Advertisement

Japanese Journal of Radiology

, Volume 37, Issue 1, pp 34–72 | Cite as

Machine learning studies on major brain diseases: 5-year trends of 2014–2018

  • Koji SakaiEmail author
  • Kei Yamada
Invited Review

Abstract

In the recent 5 years (2014–2018), there has been growing interest in the use of machine learning (ML) techniques to explore image diagnosis and prognosis of therapeutic lesion changes within the area of neuroradiology. However, to date, the majority of research trend and current status have not been clearly illuminated in the neuroradiology field. More than 1000 papers have been published during the past 5 years on subject classification and prediction focused on multiple brain disorders. We provide a survey of 209 papers in this field with a focus on top ten active areas of research; i.e., Alzheimer’s disease/mild cognitive impairment, brain tumor; schizophrenia, depressive disorders, Parkinson’s disease, attention-deficit hyperactivity disorder, autism spectrum disease, epilepsy, multiple sclerosis, stroke, and traumatic brain injury. Detailed information of these studies, such as ML methods, sample size, type of inputted features and reported accuracy, are summarized. This paper reviews the evidences, current limitations and status of studies using ML to assess brain disorders in neuroimaging data. The main bottleneck of this research field is still the limited sample size, which could be potentially addressed by modern data sharing models, such as ADNI.

Keywords

Artificial intelligence Machine learning Neurological disorder Neuroimaging Diagnosis 

Introduction

Wang and Summers [1] predicted the future of machine learning (ML) in radiology and stated that “statistical approaches will be a major direction on ML study in radiology.” They expected machine learning to be “a critical component of advanced software systems in the field of radiology.” Conceptually their prediction is correct; however, up to the present day its practical application in radiology has not yet been proven to be entirely feasible. Limited sample size has been the largest issue hindering ML advancements in the field of radiology during the last 6 years. To achieve significant statistical power, large amounts of data are necessary. Wang and Summers’ prediction on the statistical underlying basis of ML was correct; however, curious enough, the basis of their own hypothesis has been the major hindering factor for its development.

Does the “small N” problem [2] still have an adverse effect on ML study in the field of neuroradiology? Accordingly, has there been any recent progress on ML studies? In this review, we tried answering these questions by acquiring the ML research trend thorough literature search. We used keyword paper search on Pubmed [3].

In this article, we focused on the efficiency of ML techniques, with special focus on brain diseases. To capture the research trend from 2014 to 2018 (5 years), we picked up top ten active studies according to the number of papers (conference papers were excluded). These included Alzheimer’s disease (AD)/mild cognitive impairment (MCI), brain tumor; schizophrenia (SCZ), depressive disorders, Parkinson’s disease (PD), attention-deficit hyperactivity disorder (ADHD), autism spectrum disease (ASD), epilepsy, multiple sclerosis (MS), stroke, and traumatic brain injury (TBI). Detailed information about those studies such as ML methods, sample size, type of input features and reported accuracy were summarized. Based on this survey, we will try to clarify the current status and bottlenecks of ML studies.

Survey

Based on PubMed search from 2014 to 2018, 1337 papers on neuroimage-based ML researches were found. The key search terms: brain, machine learning, and diagnosis. Figure 1 summarizes the literature selection procedure for this paper.
Fig. 1

Literature selection procedure

An initial 1337 studies were identified using the search strategy described above, and 511 of these were screened by the title after removing the duplicates. A total of 467 abstracts were screened and 44 studies were excluded at this point. This left 340 studies screened by full-text, with 127 studies excluded at this juncture. We limited our search to journal papers in English, published up to 19 September 2018. In a few instances, the full paper was not found and therefore those studies were excluded from this survey. This review was conducted partially according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for systematic reviews [4].

Total of 209 papers were eventually selected based on the number of papers on top ten researches selected. The key aspects of each study, such as research purpose, machine learning method, imaging modality (ultrasonography was excluded), sample size, inputted features for ML, and remarks were investigated. Those that did not show results as numerical values were considered not eligible.

A list of all abbreviations used in the tables and the manuscript is provided in Table 1.
Table 1

Abbreviations used in the tables and manuscript

Abbreviation

Full term

(a) Used in informatics

 

 3D HOG

Histogram of oriented 3D gradients

 BIRN

The Biomedical Informatics Research Network

 CRF

Conditional random fields

 CV

Cross validation

 DANS

Discriminant autoencoder network with sparsity constraint

 DBM

Deep Boltzmann machine

 DBN

Deep belief network

 DDN

Deep delief network

 DESRN

Deep ensemble sparse regression network

 DF

Deformation fields

 DLDA

Diagonal linear discriminate analysis

 DNN

Deep neural network

 DQDA

Diagonal quadratic discriminate analysis

 DRBMs

Discriminative restricted Boltzmann machines

 ELM

Extreme learning machines

 EPNN

Enhanced probabilistic neural network

 FASMA

Fast spectroscopic multiple analysis

 FCM

Fuzzy C-means

 FWT

Fast wavelet decomposition

 GMLVQ

Generalized matrix learning vector quantization

 GNB

Gaussian Naïve Bayes

 IFWT

Fast wavelet reconstruction

 IPSO

Improved particle swarm optimization

 IVM

Import vector machine

 JFSS

Joint feature-sample selection

 k-NN

k-Nearest neighbor

 KSOM

Kohonen self-organizing map

 KPLS

Kernel partial least squares regression

 LASSO

Least absolute shrinkage and selection operator

 LDA

Linear discriminant analysis

 LDS

Low density separation

 LG

Label generation

 LINDA

Lesion identification with neighborhood data analysis

 LOOCV

Leave-one-out cross validation

 LOSPGOCV

Leave-one-subject-per-group-out cross validation

 LOSOCV

Leave-one-subject-out cross validation

 LPOCV

Leave-pair-out cross validation

 LSOCV

Leave-site-out cross validation

 LTOCV

Leave-two-out cross validation

 LR

Logistic regression

 LR

Logistic regression

 M2TL

Multimodal manifold-regularized transfer learning

 MCCV

Monte carlo cross-validation

 MIMC

Multi-label inductive matrix completion

 ML

Machine learning

 MLDA

Maximum entropy linear discriminant analysis

 MMC

Maximum margin clustering

 MPA

Multivoxel pattern analyses

 MVPA

Multi-voxel pattern analysis

 NMF

Non-negative matrix factorization

 NN

Neural Network

 OASIS

Open access series of imaging studies

 OC-SVM

One-class support vector machine.

 OOB

Out-of-bag

 OVA

One-versus-all

 PNN

Robabilistic neural network

 RBF

Radial based function

 RBM

Restricted Boltzmann machine

 RDF

Random decision forest

 RELM

Regularized extreme learning machine

 RF

Random forest

 RFE

Recursive feature elimination

 RLDA

Robust discriminant analysis

 SBFE

Searchlight based feature extraction

 SLFN

Single layer feedforward networks

 SMSMA

Supervised multiblock sparse multivariable analysis

 SVM

Support vector machine

 SWCSD

Supervised within-class-similar discriminative

 TGSL

Temporally-constrained group sparse learning

(b) Used in medicine

 

 11C-MET

l-[Methyl-11C methionine

 FP-CIT SPECT

123Iodine-labelled N-(3-fluoropropyl)-2β-carbomethoxy-3β-(4-iodophenyl) nortropane (123I-FP-CIT) single photon emission computerized tomography

 AD

Alzheimer’s disease

 ADC

Apparent diffusion coefficient

 ADNI

Alzheimer’s disease Neuroimaging Initiative

 AH

Auditory hallucinations

 AIBL

The Australian Imaging, Biomarker and Lifestyle Flagship Study of Ageing

 ASD

Autism spectrum disease

 AUC

Area under the curve

 BD

Bipolar disorder

 BIRN

The Biomedical Informatics Research Network

 BRATS

Multimodal Brain Tumor Segmentation Challenge

 CL

Cognitive loss

 CLIMB

The Comprehensive Longitudinal Investigation of MS at the Brigham and Women’s Hospital Partners MS Center

 cMDD

Current major depressive disorder patients

 CMRGlc

Cerebral metabolic rate of glucose consumption

 COBRE

The Center for Biomedical Research Excellence

 CSF

Cerebrospinal fluid

 CT

Computed tomography

 cT1WI

Contrast enhanced T1WI

  DBS

Deep brain stimulation

 DKI

Diffusion kurtosis image

 dMRI

Diffusion MRI

 DSM

Diagnosis and Statistical Manual of Mental Disorders

 DTI

Diffusion tensor image

 DZNE

The German center for neurodegenerative diseases

 ECT

Electroconvulsive therapy

 EEG

Electroencephalogram

 EP

Ependymoma

 FA

Fractional anisotropy

 FDG-PET

2-Deoxy-2-[18F]fluoroglucose PET

 FLAIR

Fluid attenuated inversion recovery

 fMRI

Functional MRI

 GBM

Glioblastoma multiforme

 GM

Gray matter

 HARDI

High angular resolution diffusion images

 HC

Healthy control

 HNPPH

Henan provincial people’s hospital

 HR

High risk

 LBP

Local binary pattern

 LGG

low grade glioma

 MB

Medulloblastoma

 MBs

Medulloblastomas

 MCI

Mild cognitive impairment

 MCI-C

MCI converter

 MCI-NC

MCI non-converter

 MD

Mean diffusivity

 MDD

Major depressive disorder

 MEG

Magnetoencephalography

 MEN

Meningioma

 MI

Motor function impaired

 MP

Motor function preserved

 MPRAGE

Magnetization-prepared rapid acquisition with gradient echo

 MR

Magnetic resonance

 MRI

Magnetic resonance image

 MS

Multiple sclerosis

 mTBI

Mild traumatic brain injury

 MTS

Mesial temporal sclerosis

 NC

Normal control

 NIH

National Institutes of Health

 NITRC

The Neuroimaging Informatics Tools and Resources Clearinghouse

 OASIS

The Open Access Series of Imaging Studies

 OCD

Obsessive compulsive disorder

 PAs

Pilocytic astrocytomas

 PCS

Post concussion syndrome

 PD

Parkinson’s disease

 PDWI

Proton density weighted image

 PET

Photon emission tomography

 PiB

Pittsburg compound B

 pMCI

Progressive MCI

 PPI

Psychophysiological interaction

 PPMI

Parkinson’s progression markers initiative

 PRODEM

The prospective registry on dementia

 PSP

Progressive supranuclear palsy

 PTSD

Posttraumatic stress disorder

 R2*WI

R2* weighted image

 rMDD

Remitted major depressive disorder patients

 RNA

Ribonucleic acid

 rCBV

Relative cerebral blood volume

 rRCBV

Relative regional cerebral blood volume

 rs-fMRI

Resting state fMRI

 sMCI

Stable MCI

 sMRI

Structural MRI

 SCI

Subjective cognitive impairment

 SCZ

Schizophrenia

 SIB

Non-affected siblings of patients with schizophrenia

 SICH

Symptomatic intracranial haemorrhage

 sMRI

Structural MRI

 SNP

Single nucleotide polymorphism

 SUVr

Standard uptake value ratio

 SWEDD

Subjects with scans without evidence of dopaminergic deficit

 T1WI

T1 weighted image

 T2WI

T2 weighted image

 TBI

Traumatic brain injury

 TCGA

The cancer genome atlas

 TCIA

The cancer imaging archive

 WM

White matter

Results

Top ten studied brain diseases

Table 2 shows the top 10 studied brain diseases, their number of publications from 2014 to 2018, and relating large data sets. The top ten research areas for this survey were AD/MCI (62 papers), brain tumor (37 papers), schizophrenia (21 papers), depressive disorders (20 papers), PD (13 papers), ADHD (13 papers), ASD (12 papers), epilepsy (10 papers), MS (8 papers), stroke (7 papers), and TBI (7 papers).
Table 2

Top ten studied brain diseases, their number of publications from 2014 to 2018, and related large data sets

Rank

Disease

Number of papers

Large data set based studies

Large data set

1

Alzheimer’s disease/MCI

61

51 (83.6%)

ADNI, OASIS, DZNE, PRODEM, AIBL

2

Brain tumor

37

6 (16.2%)

BRATS, TCIA, TCGA

3

Schizophrenia

21

4 (19.0%)

COBRE, NITRC, BIRN

4

Depression

20

0 (0.0%)

5

Parkinson’s disease

13

9 (69.2%)

PPMI

5

ADHD

13

8 (61.5%)

ADHD-200a, ABIDEb

7

Autism

12

10 (83.3%)

ABIDEb

8

Epilepsy

10

0 (0.0%)

9

Multiple sclerosis

8

1 (12.5%)

CLIMB

10

Stroke

7

0 (0.0%)

10

Traumatic brain injury

7

0 (0.0%)

 

Sum

209

89 (43.8%)

 

aADHD-200: http://fcon_1000.projects.nitrc.org/indi/adhd200/

bABIDE: http://fcon_1000.projects.nitrc.org/indi/abide/databases.html

Alzheimer’s disease/mild cognitive impairment

MCI is considered a prodromal phase to dementia, especially the AD type. AD is the most common neurodegenerative disorder, which is increasingly prevalent among adults aged 65 years and older. Considering the prevalence and severity of MCI/AD, the largest number of neuroimaging-based, automatic classification publications has been published. Also, a number of these studies investigated the possibility of automatic classification of stable MCI (sMCI) from progressive MCI (pMCI) and converter MCI from non-converter MCI using ML plus neuroimaging. Table 3 summarizes the 55 papers. 48 papers (83%) among 55 were studied about classification of AD, MCI (including subtypes), and normal controls.
Table 3

Summary of AD/MCI studies

Purpose

Disorder

Modality

Feature

Classifier

Validation

Overall accuracy

Number of subjects

Data basea

Author

Year

References

Classification

MCI

rs-fMRI

Functional connectivity network

SVM

LOOCV

0.919 (MCI vs NC)

MCI = 12, NC = 25, total = 37

Internal

Jie et al.

2014

[5]

Classification

AD/MCI

sMRI

Local binary pattern

SVM

10-fold CV

0.828 (AD vs HC), 0.615 (MCI vs NC)

AD = 80, pMCI = 141, HC = 142, total = 363

ADNI

Li et al.

2014

[6]

Classification

AD

sMRI, DTI

DTI features

SVM

LOOCV

0.943 (AD vs HC),

AD = 21, HC = 15, total = 36

Internal

Li et al.

2014

[7]

Classification

MCI

sMRI

Cortical features

SVM

LOOCV

0.8 (aMCI vs NC)

aMCI = 24, NC = 26, total = 50

Internal

Li et al.

2014

[8]

Classification

AD/MCI

T1WI

Voxel-wise imaging features

SVM

10-fold CV

0.9 (AD vs HC), 0.87 (pMCI vs NC), 0.71 (pMCIvs sMCI)

AD = 198, pMCI = 167, sMCI = 236, HC = 229, total = 830

ADNI

Liu et al.

2014

[9]

Classification

MCI

PET

Neuropsychological test data

SVM

N/A

0.89 (cMCI vs sMCI)

cMCI = 25, sMCI = 20, total = 45

Internal

Segovia et al.

2014

[10]

Classification

AD/MCI

sMRI, FDG-PET

GM structure, PET, CSF biomarker

SVM

10-fold CV

0.95 (AD vs HC), 0.8 (MCI vs NC), 0.746 (MCI vs AD), 0.72 (ncMCI vs cMCI)

AD = 51, cMCI = 43, ncMCI = 56, HC = 52, total = 202

ADNI

Suk et al.

2014

[11]

Classification

MCI

MRI, PET

189 features (93 MRI, 93PET, 3 CSF)

M2TL

10-fold CV

0.801 (cMCI vs ncMCI)

AD = 51, cMCI = 43, ncMCI = 56, NC = 52, total = 202

ADNI

Cheng et al.

2015

[12]

Classification

AD

dMRI, rs-fMRI, sMRI

Fiber tract integrity, graph-theoretical measures, GM volume

SVM

LOOCV

0.85 (AD vs HC)

AD = 28, HC = 25, total = 53

DZNE

Dyrba et al.

2015

[13]

Classification

AD

MRI

Longitudinal percentage of brain volume changes

SVM + RBF

k-fold CV

0.917 (AD vs HC)

AD = 30, HC = 30, total = 60

ADNI

Farzan et al.

2015

[14]

Classification

AD/MCI

MRI, FDG-PET

GM volume, average intensity of PET

SVM

10-fold CV

0.9503 (AD vs HC), 0.7927 (MCI vs NC), 0.689 (cMCI vs ncMCI)

AD = 51, pMCI = 99, HC = 52, total = 202

ADNI

Jie et al.

2015

[15]

Classification

AD

rs-fMRI

Network-based features

SVM

LOOCV

1.0 (AD vs HC)

AD = 20, HC = 20, total = 40

ADNI

Khazaee et al.

2015

[16]

Classification

AD/MCI

MRI

GM volumetric features

SVM

10-fold CV

0.925 (AD vs HC), 0.79 (pMCIvs sMCI)

AD = 97, pMCI = 117, sMCI = 117, HC = 128, total = 459

ADNI

Liu et al.

2015

[17]

Classification

AD/MCI

MRI, PET

GM volume, regional ave. CMRGlc

SVM

10-fold CV

0.91 (AD vs HC), 0.82 (MCI vs NC)

AD = 85, cMCI = 67, ncMCI = 102, HC = 77, total = 331

ADNI

Liu et al.

2015

[18]

Classification

AD

MPRAGE

Smoothed GM density values

SVM, LDS

2 nested CV loops

0.74 (AD vs NC)

AD = 200, pMCI = 164, sMCI = 100, HC = 231, total = 695

ADNI

Moradi et al.

2015

[19]

Classification

AD/MCI

MRI, FDG-PET, florbetapir PET

GM volume, SUVr

wmSRC

N/A

0.948 (AD vs HC), 0.745 (MCI vs HC), 0.778 (pMCIvs sMCI)

AD = 113, pMCI = 110, HC = 117, total = 340

ADNI

Xu et al.

2015

[20]

Classification

AD/MCI

MRI, PET

HOG

SVM

10-fold CV

0.913 (AD vs HC), 0.781 (MCI vs HC), 0.755 (pMCI vs sMCI)

AD = 198, pMCI = 124, sMCI = 118, HC = 229, total = 669

ADNI

Zhu et al.

2015

[21]

Classification

AD

11C-PiB PET

3D HOG, amyloid status

SVM

LOOCV

1.00 (AD vs HC)

AD = 167, HC = 42, total = 209

ADNI

Cattell et al.

2016

[22]

Classification

AD/MCI

PET, rs-fMRI

Feature graph

SVM

LOOCV

0.9714 (AD vs HC), 0.9 (MCI vs NC)

AD = 30, EMCI = 30, HC = 60, total = 120

ADNI

Hu et al.

2016

[23]

Classification

AD/MCI

rs-fMRI

Graph measures

SVM

23-fold CV

0.873 (NC vs AD/MCI), 0.975 (AD vs NC/MCI), 0.72 (MCI vs NC/AD)

AD = 34, MCI = 89, NC = 45, total = 168

ADNI

Khazaee et al.

2016

[24]

Classification

AD/MCI

T1WI

GM density map

SVM

10-fold CV

0.9306 (AD vs HC), 0.7925 (pMCI vs sMCI)

AD = 97, pMCI = 117, sMCI = 117, HC = 128, total = 459

ADNI

Liu et al.

2016

[25]

Classification

AD

rs-fMRI

rs-fMRI feature

SVM

LOOCV

0.76 (AD vs HC)

AD = 25, HC = 34, total = 59

ADNI

Ni et al.

2016

[26]

Classification

MCI

rs-fMRI

Functional connectivity

RBM

LOOCV

0.7258 (MCI vs NC)

MCI = 31, NC = 31, total = 62

ADNI2

Suk et al.

2016

[27]

Classification

AD/MCI

MRI, PET

Feature graph

SVM

10-fold CV

0.926 (AD vs HC), 0.8 (MCI vs NC)

AD = 50, pMCI = 97, HC = 52, total = 199

ADNI

Yu et al.

2016

[28]

Classification

AD/MCI

MRI, FDG-PET

GM volume, average intensity of PET

SVM

10-fold CV

0.9595 (AD vs HC), 0.8026 (MCI vs NC)

AD = 51, pMCI = 99, HC = 52, total = 202

ADNI

Zu et al.

2016

[29]

Classification

AD/MCI

sMRI

MRI, SNP features

SVM

N/A

0.92 (AD vs HC), 0.8 (MCI vs HC), 0.81 (pMCI vs sMCI)

AD = 171, pMCI = 157, sMCI = 205, HC = 204, total = 737

ADNI

An et al.

2017

[30]

Classification

MCI

T1WI

GM region atrophy

SVM

10-fold CV

0.93 (pMCI vs sMCI)

AD = 92, pMCI = 70, sMCI = 65, HC = 94, total = 321

ADNI

Beheshti et al.

2017

[31]

Classification

MCI

rs-fMRI, DTI

Functional connectivity

SVM

LOOCV

0.787 (MCI vs NC)

MCI = 54, NC = 54, total = 108

ADNI

Chen et al.

2017

[32]

Classification

AD

T1WI

GM density map

LASSO

10-fold CV

0.81 (AD vs HC)

AD = 137, SCI = 38, MCI = 78, HC = 355, total = 608

ADNI

Doan et al.

2017

[33]

Classification

AD

MRI, PET

5 ROI, 6 shape, 2 volume features

SVM

CV

0.8813 (MCI vs NC)

AD = 137, cMCI = 76, HC = 162, total = 375

ADNI

Glozman et al.

2017

[34]

Classification

MCI

MRI

Cortical surface based measurements

SVM

Stratified shuffle split CV

0.77 (naMCI vs aMCI), 0.81 (aMCI vs CN), 0.7 (naMCI vs CN)

aMCI = 40, naMCI = 27, NC = 117, total = 184

Internal

Guan et al.

2017

[35]

Classification

AD

rs-fMRI

Connectivity hyper-networks

SVM

10-fold CV

0.916 (AD vs HC)

AD = 38, HC = 28, total = 66

Internal

Guo et al.

2017

[36]

Classification

MCI

rs-fMRI

Graph measures

SVM

LOOCV, 9-fold CV

0.914

cMCI = 18, ncMCI = 62, total = 80

ADNI

Hojjati et al.

2017

[37]

Classification

AD

T1WI, PET

GM volume, average intensity of PET

SMSMA

10-fold CV

AUC = 0.95

AD = 52, HC = 48, total = 100

ADNI

Kawaguchi et al.

2017

[38]

Classification

AD/MCI

rs-fMRI

Graph measures

Naïve Bayes

10-fold CV

0.9329 (MCI vs AD)

AD = 34, MCI = 89, NC =45, total = 168

ADNI

Khazaee et al.

2017

[39]

Classification

AD

MRI, PET

Structural measures

SVM, IVM, RELM

70/30 CV, 10-fold CV, LOOCV

0.7788 (AD vs HC),

AD = 70, pMCI = 74, HC = 70, total = 214

ADNI

Lama et al.

2017

[40]

Classification

AD/MCI

sMRI, FDG-PET, florbetapir PET

Mean GM volume, SUVr

SCDDL

10-fold CV

0.9736 (AD vs HC), 0.7766 (MCI vs HC)

AD = 113, pMCI = 110, HC = 117, total = 340

ADNI

Li et al.

2017

[41]

Classification

AD/MCI

MRI

GM, amygdara

SVM

10-fold CV

0.965 (AD vs HC), 0.9174 (pMCI vs HC), 0.8899 (pMCI vs sMCI)

AD = 65, pMCI = 95, sMCI = 132, HC = 135, total = 427

ADNI

Long et al.

2017

[42]

Classification

AD

T1WI

Regional volume

Automated volumetric assessment tool

N/A

1.00 (AD vs HC)

AD = 30, HC = 25, total = 55

Internal

Min et al.

2017

[43]

Classification

AD

MRI, blood lipid

Lipidomics

Random forest

Training/test (2/1)

0.71(AD vs HC)

AD = 142, HC = 135, total = 277

Internal

Proitsi et al.

2017

[44]

Classification

AD/MCI

sMRI

Cortical thickness, hippocampal shape, hippocampal texture, and volumetry

LDA

10-fold CV

0.969 (NC vs AD/MCI), 0.612 (AD vs NC/MCI), 0.287 (MCI vs NC/AD)

N/A

The CADDementia challenge

Sørensen et al.

2017

[45]

Classification

AD/MCI

sMRI

Spatially normalized GM densities

Deep ESR net

10-fold CV

0.91 (AD vs HC), 0.73 (MCI vs NC), 0.748 (pMCIvs sMCI)

AD = 186, pMCI = 167, sMCI = 226, HC = 226, total = 805

ADNI

Suk et al.

2017

[46]

Classification

AD/MCI

MPRAGE, PiB-PET

GM volume, intensity of PET

Simple MKL

10-fold CV

0.957 (AD vs HC), 0.958 (MCI vs NC), 0.951 (MCI vs AD)

AD = 52, pMCI = 108, HC = 120, total = 280

AIBL

Youssofzadeh et al.

2017

[47]

Classification

AD/MCI

MRI

GM volume

SVM

10-fold CV

0.903 (AD vs HC), 0.722 (MCI vs NC), 0.713 (sMCI vs cMCI)

AD = 186, pMCI = 393, HC = 226, total = 805

ADNI

Zhu et al.

2017

[48]

Classification

AD/MCI

MRI, PET

GM volume, intensity of PET

SVM

10-fold CV

0.957 (AD vs HC), 0.799 (MCI vs HC), 0.724 (ncMCI vs cMCI)

AD = 51, cMCI = 143, ncMCI = 156, HC = 52, total = 402

ADNI

Zhu et al.

2017

[49]

Classification

AD

MRI

Hippocampal

HUMAN

N/A

0.20 (AD vs NC)

AD = 132, pMCI = 250, HC = 174, total = 556

ADNI

Amoroso et al.

2018

[50]

Classification

AD

FDG-PET, florbetapir PET

PET images

Deep CNN

10-fold CV

0.96 (AD vs HC)

AD = 139, pMCI = 79, sMCI = 92, HC = 182, total = 492

ADNI-2

Choi et al.

2018

[51]

Classification

AD

FDG-PET

Intensity of PET

SVM

LOOCV

0.9788 (AD vs HC)

AD = 81, pMCI = 29, HC = 61, total = 171

Internal

Garali et al.

2018

[52]

Classification

AD/MCI

MRI, FDG-PET

Volume, mean intensity, CSF features (p-tau, t-tau, Aβ42)

ELM

10-fold CV

0.971 (AD vs HC), 0.871 (MCI vs HC)

AD = 51, pMCI = 99, HC = 52, total = 202

ADNI

Kim et al.

2018

[53]

Classification

AD/MCI

T1WI, FDG-PET

3D image patch

CNN

10-fold CV

0.933 (AD vs HC), 0.83 (sMCI vs NC), 0.64 (pMCI vs NC)

AD = 93, pMCI = 204, HC = 100, total = 397

ADNI

Liu et al.

2018

[54]

Classification

AD

MRI, PET, SPECT

FDG-PET

SVM

LOOCV

0.93 (AD vs HC)

AD = 20, HC = 18, total = 38

Internal

Rondina et al.

2018

[55]

Classification

AD/MCI

MRI

Kaggle-neuroimaging-challenge: 429 features (demographic, clinical features)

SVM

5-fold CV

0.96 (AD vs HC), 0.9 (cMCI vs AD), 0.78 (ncMCI vs AD), 0.62 (ncMCI vs cMCI), 0.79 (cMCI vs HC), 0.59 (ncMCI vs HC)

AD = 44, cMCI = 44, ncMCI = 44, HC = 44, total = 176

ADNI

Salvatore et al.

2018

[56]

Classification

AD/MCI

MRI

GM density map

SVM

Training/test (1/1)

0.928 (AD vs HC), 0.708 (MCI vs NC), 0.64 (sMCI vs cMCI), 0.657 (MCI vs AD)

AD = 186, cMCI = 167, sMCI = 226, HC = 226, total = 805

ADNI

Sun et al.

2018

[57]

Classification

AD

MPRAGE

Image intensity

CNN

Training/test (1/1)

0.977 (AD vs HC)

AD = 98, HC = 98, total = 196

OASIS, internal

Wang et al.

2018

[58]

Detection

AD

DTI, DKI

DTI, DKI measures

SVM

LOOCV

0.96 (AD vs HC)

AD = 27, HC = 26, total = 53

Internal

Chen et al.

2017

[59]

Diagnosis

AD

DTI

FA-ICA

Elastc net

10-fold CV

0.85 (AD vs HC)

AD = 77, HC = 173, total = 250

PRODEM

Schouten et al.

2017

[60]

Diagnosis

AD

MRI, FDG-PET

Hippocampal texture

SVM

20-fold CV

AUC = 0.74/0.83 (MCI to AD)

AD = 101, pMCI = 233, HC = 169, total = 503/AD = 28, pMCI = 25, HC = 88, total =  = 141

ADNI/AIBL

Sørensen et al.

2016

[61]

Diagnosis

AD

MRI, PET

HYDRA

SVM

10-fold CV

AUC = 0.91

AD = 123, HC = 177, total = 300

ADNI

Varol et al.

2017

[62]

Identification

AD

MRI

Morphological measures

Naïve Bayes, logistic regression, SVM

MCCV

0.85 (AD vs HC)

AD = 75, HC = 75, total = 150

OASIS

Cai et al.

2017

[63]

Prediction

AD/MCI

FDG-PET, florbetapir PET

PET images

Deep CNN

10-fold CV

0.84 (AD converter)

AD = 139, pMCI = 79, sMCI = 92, HC = 182, total = 492

ADNI-2

Choi et al.

2018

[64]

Prediction

AD

MRI

Total GM volume

tgLASSO

10-fold CV

0.757 (AD converter)

AD = 91, cMCI = 104, ncMCI = 98, HC = 152, total = 445

ADNI

Jie et al.

2017

[65]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from institutional and/or public through media channels

Brain tumor

A central goal of ML on brain tumor is the identification of quantitative imaging indicators that predict important clinical outcomes, including prognosis and response or resistance to a specific cancer treatment [65]. The main purposes of the surveyed ML studies on brain tumor were segmentation of the tumor, classification of the type of tumor, and prediction of survival or genotype. MRI was the main modality for ML studies. Depending on the purpose, variety of MR imaging methods were employed. Table 4 summarizes the brain tumor studies.
Table 4

Summary of brain tumor studies

Purpose

Type

Modality

Feature

Classifier

Validation

Accuracy

Number of subjects

Data basea

Author

Year

References

Detection

Glioma progression

(18)F-FDG PET, MRS

SUVmax, Cho conc., Cr conc.

SVM

LOOCV

0.83

Total = 12 (6II, 6III)

Internal

Imani et al.

2014

[66]

Segmentation

Brain tumor

T1WI, MRS

3D texture

ELM-IPSO

LOOCV

0.9915

Training = 35 (12 meningiomas, 23 gliomas)

Internal

Nachimuthu et al.

2014

[67]

Detection

Brain metastasis

T1WI, T2WI, FLAIR

Histogram, morphology, texture

Template-matching, k-means cluster, ANN

Training/test (80/60)

AUC = 0.874

Training = 80 (450 BM nodules), Test = 30 (134BM nodules), 30 (without nodules), Total = 140

Internal

Sunwoo et al.

2017

[68]

Segmentation

Tumor compartment

T1WI, cT1WI, T2WI, FLAIR

Superpixel

SVM

2-fold CV

0.975

Total = 20 GBM

BRATS2012

Wu et al.

2014

[69]

Segmentation

Tumor compartment, GBM

T2WI, DWI, PWI

T2WI, ADC, rCBV maps

FCM

N/A

0.84 - 0.92

Total = 13 GBM

Internal

Fathi et al.

2015

[70]

Segmentation

Tumor compartment

T1WI, cT1WIFLAIR, PWI

rRCBV, sveral ROIs

KFCM

N/A

0.7183

Total = 20 (9GBM, 4HG, 1DA, 5MEN, 1MET)

Internal

Szwarc et al.

2015

[71]

Segmentation

Brain tumor

T1WI, T1WIC, T2WI, FLAIR

Shape, size, contrast

DNN

Training/test (30/10)

0.88

Training = 30 (20 high grade, 10 low grade), test = 10 (high grade), Total = 40

BRATS2013

Havaei et al.

2017

[72]

Segmentation

LGG

T2WI, MRS

Image intensity, 11 metabolites feature vector

IFWT

LOOCV

0.9

Total = 7 LGG

Internal

Li et al.

2017

[73]

Segmentation

Tumor compartment

T1WI, cT1WI, FLAIR, PWI, DWI

6 contrasts

NMF

N/A

0.8

Total = 21 HGG

Internal

Sauwen et al.

2017

[74]

Segmentation

Brain tumor

MRI, FALIR

Contrast

CNN

10-fold CV

0.76

Total = 44 GBM

Internal

AlBadawy et al.

2018

[75]

Segmentation

Tumor compartment, glioma

18F-FET PET

PET image

3D U-net CNN

Training/test (26/11)

0.8231

Training = 26, Test = 11, Total = 37

Internal

Blanc-Durand et al.

2018

[76]

Segmentation

Tumor compartment

DTI

3D supervoxel

RF

N/A

0.89

Total = 30 (20 high grade, 10 low grade)

BRATS2013

Soltaninejad et al.

2018

[77]

Segmentation

LGG, Glioma

T1WI, T2WI, FLAIR, cT1WI

Image intensity

CRF

N/A

0.823 (LGG), 0.823 (HGG)

Total = 54 LGG, 200 HGG, Test = 161, Total = 215

BRATS2013, 2015,HNPPH

Zhao et al.

2018

[78]

Classification

High vs low grade

T1WI, T2WI, DTI

Metrics

SVM

LOOCV

0.80

Total = 33 (14 LGG, 19HGG)

Internal

Inano et al.

2014

[79]

Classification

Brain tumor

T1WI, T2WI

3D texture

ANN

LOOCV

0.92

Total = 44 (21 MB, 20 PA, 7 EP)

Internal

Fetit et al.

2015

[80]

Classification

Glioblastoma

T1WI, T2WI, PWI, DTI

Texture features

DLDA, SQDA, SVM

Training/test (11/7)

0.82

Training = 11 (60 biopsies), test = 7 (22biopsies), Total = 18

Internal

Hu et al.

2015

[81]

Classification

Brain tumor

MRI, MRS,DWI, DTI, PWI

Computed metrics

SVM

10-fold CV

0.89

Total = 126

Internal

Tsolaki et al.

2015

[82]

Classification

Glioma

MRI

LBP features

Binary logistic regression

LOOCV

0.93

Total = 107 (73 LGG, 34 GBM)

TCIA

Li-Chun et al.

2017

[83]

Classification

Glioma

MRI

Histogram, texture feature

SVM, IBK

LOOCV

0.945 (high vs low grade), 0.961(II/III/IV)

Total = 120 (28LGG, 92HGG)

Internal

Zhang et al.

2017

[84]

Classification/Prediction

Brain tumor, glioma

T1WI, T2WI, FLAIR, DWI

Computed metrics

RF

5-fold CV

AUC = 0.98(II/III), AUC = 1.00(II/IV), AUC = 0.97(III/IV), AUC = 0.88(IDH status)

Total = 381 (57II, 63III, 261IV)

Internal

De Looze et al.

2018

[85]

Purpose

Type

Modality

Feature

Classifier

Validation

Accuracy

Number of subjects

Data basea

Author

Year

References

Classification

Brain Tumour

T1WI, T2WI

3D texture attributes

SVM

LOOCV

AUC = 0.86

Total = 134 (45 MB, 71PA, 19EP)

Multi-center

Fetit et al.

2018

[86]

Classification

LGGs

cT1WI,T2WI, FLAIR

Radiomic features

SVM

5-fold CV

0.87

Total = 21 1p/19q intact, 26 1p/19q codeleted

Internal

Shofty et al.

2018

[87]

Classification

Brain tumor

MRS

MRS feature

SVM

10-fold CV

0.86

Total = 41(17 medulloblastomas, 20 pilocytic astrocytomas, 4 ependymomas)

Internal

Zarinabad et al.

2018

[88]

Stratification

Pseudo-/true progression

MRI, DTI

FA

SVM

10-fold CV

0.87

Total = 35 GB (13PsP, 22TTP)

Internal

Qian et al.

2016

[89]

Stratification

Pseudo-/true progression

MRI, DTI

FA

SVM

5-fold CV

AUC = 0.87

Total = 79 GBM (23PsP. 56TTP)

Internal

Zhang et al.

2016

[90]

Prediction

Glioma, 4 year survival associations

cT1WI, T2WI, FLAIR, DWI, DSC-MRI

Computed metrics

SVM

10-fold CV

AUC = 0.82

Total = 94 (31II, 14III, 49IV)

Internal

Emblem et al.

2014

[91]

Prediction

Glioma, 3 years survival associations

MRI, DWI, DSC-MRI

Computed metrics

SVM

Training/test (101/134)

AUC = 0.851

Training = 101, test = 134, total = 235

Internal

Emblem et al.

2015

[92]

Prediction

Glioblastoma, overall survival

MT1WI, T2WI, FLAIR, DTI, DSC-MRI

Computed metrics

SVM

Training/test (34/31)

AUC = 0.84

Training = 34, test = 31, Total = 65

Internal

Akbari et al.

2016

[93]

Prediction

Glioblastoma, recurrence location

T1WI, FLAI

Volumetric, shape, texture, parametric, histogram features.

RF

Training/test (84/42)

AUC = 0.72

Training = 84, test = 42, total = 126

Internal

Chang et al.

2016

[94]

Prediction

Glioblastoma, methylation status

T1WI, T2WI

Co-occurrence, run length texture features

SVM

N/A

AUC = 0.85

Total = 155 (66 methylated, 89 unmethylated)

Internal

Korfiatis et al.

2016

[95]

Prediction

Glioblastoma, molecular subtypes

T1WI, cT1WI, T2WI, FLAIR, DTI, DSC-MRI

Imaging features

SVM

Training/test (105/29)

0.76

Training = 105 GB, test = 29 GB, total = 134

Internal

Macyszyn et al.

2016

[96]

Prediction

LGG, Chromosomal Arms 1P/19q deletion

cT1WI,T2WI

Image intensity

CNN

Training/test (387/90 slices)

0.88

Total = 159LGG (97OS, 45OD, 17AC)

Internal

Akkus et al.

2017

[97]

Prediction

Glioma, MGMT/IDH1 status

rs-fMRI, DTI

Metrics, networks, regions, clinical feature

MIMC

10-fold CV

0.7174 (MGMT), 0.83(IDH1)

Total = 47 MGMT (26 positive, 20 negative, 1 unlabeled), 44 IDH1 (13 positive, 33 negative, 1 unlabeled)

Internal

Chen et al.

2017

[98]

Prediction

Glioma, IDH genotype of HGG

T1WI, T2WI, DWI

Imaging features

RF

Training/test (90/30)

0.89

Total = 120 (35III, 85IV)

N/A

Zhang et al.

2017

[99]

Prediction

GBM, survival group

cT1WI, T2WI, FLAIR

Signal intensity

SVM

LOOCV

0.88

Total = 32GBM

TCGA

Zhou et al.

2017

[100]

Prediction

LGG, p53 status

T2WI

Radiomic features

SVM

Training/test (180/92)

AUC = 0.763

Training = 180 gliomas (II/III), Test = 92 gliomas (II/III), total = 272

Internal

Li et al.

2018

[101]

Prediction

Glioma, survival

11C-MET PET

Image, ex vivo, patient features

Geometric probability covering algorithms

MCCV

0.89

Total = 70 (1I, 21II, 31III, 17IV)

Internal

Papp et al.

2018

[102]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from insitutional and/or public through media channels

Schizophrenia

Considering the absence of standard clinical test for schizophrenia, there is a growing interest in automatic diagnosis of schizophrenia based on neuroimaging features [2]. Therefore, all studies aimed to identify and classify SCZ from healthy control and/or other psychiatric diseases such as bipolar disorder [106] or OCD [104]. We surveyed 21 papers, which are presented in Table 5.
Table 5

Summary of schizophrenia studies

Purpose

Disease

Modality

Feature

Classifier

Validation

Accuracy

Number of subjects

Data basea

Author

Year

References

Classification

SCZ/bipolar illness

sMRI, rs-fMRI

Functional connectivity (thalamas seeded)

SVM

LOOCV

0.74

SCZ = 90, HC = 90, total = 180

Internal

Anticevic et al.

2014

[103]

Classification

SCZ

T1WI, T2*WI, fMRI (working memory task)

MVPA, GLM

SBFE

LTOCV

0.91

P = 17 pure SCZ, 16 SCZ-OCD, HC = 20, total = 37

Internal

Bleich-Cohen et al.

2014

[104]

Classification

SCZ

T1WI

SBM

Bagged SVM

LOOCV

0.73

SCZ = 110, HC = 124, total = 234

Multisite

Castro et al.

2014

[105]

Classification

SCZ/BPD

T1WI

GM density

SVM

Training/test (198/136)

0.9(SCZ vs HC), 0.88(SCZ vs BPD), 0.67(BPD vs HC)

P = 66 SCZ, 66 BPD, HC = 66, total = 198

Internal

Schnack et al.

2014

[106]

Classification

SCZ

rs-fMRI

Functional network

SVM

LOOCV

0.79

SCZ = 19, HC = 29, total = 48

Internal

Cheng et al.

2015

[107]

Classification

SCZ, AH

T1WI, fMRI

Local fMRI activity measures

SVM

10-fold CV

1.00

P = 26 SCZ-AH (14 SCZ, 12 SAD, 1 SPD), 14 SCZ (5 SCZ, 8 SAD, 1 SPD), HC = 28, total = 68

Internal

Chyzhyk et al.

2015

[108]

Classification

SCZ

MPRAGE, rs-fMRI

Functional activity map features

Ensemble ELM, SLFN

10-fold CV

0.91

SCZ = 73, HC = 74, total = 147

Internal

Chyzhyk et al.

2015

[109]

Classification

SCZ

fMRI, MPRAGE

Regional fMRI activation pattern

SVM

LOOCV

0.93

SCZ = 44, HC = 44, total = 88

Internal

Koch et al.

2015

[110]

Classification

SCZ

T1WI

Voxel intensity

SVM

k-fold CV

0.92

SCZ = 19, HC = 16, total = 35

Internal

Chu et al.

2016

[111]

Classification

SCZ

rs-fMRI

Functional connectivity

DNN

5-fold CV

0.86

SCZ = 50, HC = 50, total = 100

NITRC

Kim et al.

2016

[112]

Classification

SCZ

T1WI

Voxel intensity

SVM

 

0.88

SCZ = 41, HC = 42, total = 83

Internal

Lu et al.

2016

[113]

Classification

SCZ

T1WI

Cortical thickness, anatomical structure volume

DBN

3-fold CV

0.74

SCZ = 143, HC = 83, total = 226

Internal

Pinaya et al.

2016

[114]

Classification

SCZ/SIB

MRI

Thalamic grey matter volume

RF

LOOCV

0.81(SCZ vs HC), 0.75(SIB vs HC)

SCZ = 96, SIB = 55, HC = 249, total = 400

Internal

Pergola et al.

2017

[115]

Classification

SCZ/PD

rs-fMRI

Functional conncetivity (thalamas seeded)

SVM

10-fold CV

0.72

SCZ = 86, HC = 84HC, total = 170

COBRE

Pläschke et al.

2017

[116]

Classification

SCZ

MPRAGE, rs-fMRI

Cortex structural features, global connectivity measure

ELM

10-fold CV

1.00

SCZ = 72, HC = 72, total = 144

COBRE

Qureshi et al.

2017

[117]

Classification

SSD

fMRI

Functional conncetivity (thalamas seeded)

Regularized linear discriminant analysis classifiers

Training/test (373/147)

0.77

SCZ = 182 HC = 348, total = 530

Multisite

Skåtun et al.

2017

[118]

Classification

SCZ

T1WI

Segmented GM, WM, CSF, behavioural analysis

SVM

LOOCV

0.94

SCZ = 17, HR = 17, total = 34

N/A

Zarogianni et al.

2017

[119]

Classification

SCZ

rs-fMRI

3D spatial maps

SVM

LOOCV

0.98

SCZ = 25, HC = 25, total = 50

BIRN

Juneja et al.

2018

[120]

Classification

SSD

DTI

FA

SVM

LTOCV

0.62

SCZ = 77, HC = 77, total = 154

Internal

Mikolas et al.

2018

[121]

Classification

SCZ

fMRI

Functional connectomes

SVM

LSOCV

0.84

SCZ = 191HC = 191, total = 382

Multisite

Orban et al.

2018

[122]

Classification

SCZ

fMRI

Functional connectivity

Deep DANS NN

LSOCV

0.85

SCZ = 474HC = 607, total = 1081

Multisite

Zeng et al.

2018

[123]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from insitutional and/or public through media channels

Depressive disorders

The diagnosis of major depression disorders (MDD) is based on application of criteria from the Diagnosis and Statistical Manual of Mental Disorders (DSM) and clinician judgment. Based on the diagnosis, initial pharmacotherapy treatment is effective in approximately 50% of patients [124]. Therefore, the main purposes of ML studies on MDD were classification and prediction of the first onset [140], treatment response [142, 143], and clinical depression score [144]. We reviewed 19 studies that used neuroimaging for automatic diagnoses of MDD. Those studies are listed in Table 6.
Table 6

Summary of depressive disorder studies

Purpose

Type

Modality

Feature

Classifier

Validation

Overall accuracy

Number of subjects

Data basea

Author

Year

References

Classification

MDD

rs-fMRI

Functional connectivity

SVM

LOOCV

0.84

MDD = 39, HC = 37, total = 76

Internal

Cao et al.

2014

[125]

Classification

DEP

DTI

FA map

SVM

LOOCV

0.83

DEP = 29, HC = 30, total = 59

Internal

Qin et al.

2014

[126]

Classification

MDD

rs-fMRI

Connectivity patterns

LG-MMC

LOOCV

0.93

MDD = 24, HC = 29, total = 53

Internal

Zeng et al.

2014

[127]

Classification

depressive comorbidity, panic disorder

fMRI

Beta values, t-statistic images

SVM

LOOCV

0.79

PD (−DEP) = 33, PD (+DEP) = 26

Internal

Lueken et al.

2015

[128]

Classification

cMDD, rMDD

DTI, fMRI

5 physiological parameters, 3 network measures

SVM

LOOCV

1.00(DEP vs HC), 0.9767(cMDD vs rMDD)

cMDD = 28, rMDD = 15, HC = 30, total = 73

Internal

Qin et al.

2015

[129]

Classification

MDD

fMRI

PPI t-maps

MLDA

LOSOCV

0.78

MDD = 25, HC = 21, total = 46

Internal

Sato et al.

2015

[130]

Classification

DEP

fMRI (verbal fluency)

Brain activity Z scores

SVM

10-fold CV

0.95

DEP = 31, HC = 31, total = 62

Internal

Shimizu et al.

2015

[131]

Classification

PUD

T1WI

Morphometric measurements

SVM

LOOCV

0.78

PUD = 25, 2HC = 6, total = 31

Internal

Wu et al.

2015

[132]

Classification

MDD, severity

rs-fMRI, fMRI (emotional-face)

fMRI volume space (masked)

SVM

5-fold CV

0.66(severe), 0.58(mild to moderate)

MDD = 45, HC = 19, total = 64

Internal

Ramasubbu et al.

2016

[133]

Classification

MDD

rs-fMRI

Connectivity patterns

SVM + elastic net

LOOCV

0.76

MDD = 38, HC = 29, total = 67

Multi site

Bhaumik et al.

2017

[134]

Classification

MDD

[carbonyl-11C]WAY-100635 PET, 3D T1WI

PET data

SVM

10-fold CV

0.75

MDD = 19, HC = 62, total = 81

Internal

Kautzky et al.

2017

[135]

Classification

MDD

DTI

FA map

SVM

LOSPGOCV

0.76

MDD = 25, HC = 25, total = 50

Internal

Schnyer et al.

2017

[136]

Classification

MDD, ECT

3D T1WI, rs-fMRI

GM volume, functional connectivity

SVM

LOOCV

0.83

MDD = 23, HC = 25, total = 48

Internal

Wang et al.

2017

[137]

Classification

MDD, BD

DTI

Tract profiles

SVM

N/A

0.68

BD = 31, MDD = 36, HC = 45, total = 112

Internal

Deng et al.

2018

[138]

Classification

MDD

[Carbonyl-11C]WAY-100635 PET, T1WI

PET data

MVPA

LOOCV

AUC = 0.58(MDD vs HC), AUC = 0.80(HR vs HC), AUC = 0.49(MDD vs HR)

N/A

N/A

Milak et al.

2018

[139]

Prediction

MDD, first onset

T1WI

Cortical thickness

SVM

10-fold CV

0.70

MDD = 18, HC = 15, total = 33

Internal

Foland-Ross et al.

2015

[140]

Prediction

TRD

MPRAGE

GM volume

SVM

LOOCV

0.85

TRD = 20, HC = 21, total = 41

Internal

Johnston et al.

2015

[141]

Prediction

Late-life depression, treatment response

T1WI, T2WI, DTI, rs-fMRI

functional, Structural image features

SVM

LOOCV

0.8727 (late-life), 0.8947 (treat. response)

DEP = 22 (late-life), 19 (treat. response), HC = 28, total = 69

Internal

Patel et al.

2015

[142]

Prediction

ECT response

sMRI

Voxel-based morphometry

SVM

LOSOCV

0.78

ECT = 23, medication = 23, HC = 21, total = 67

Internal

Redlich et al.

2016

[143]

Prediction

Clinical depression scores

rs-fMRI

Functional connectivity

KPLS-poly + LDA

LOOCV

0.81

DEP = 58, HC = 65, total = 123

Internal

Yoshida et al.

2017

[144]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from insitutional and/or public through media channels

Parkinson’s disease

The main purposes of ML studies on PD were classification and differentiation of PD from normal control (NC), progressive supranuclear palsy (PSP), and/or subjects with scans without evidence of dopaminergic deficit (SWEDD). We surveyed 13 papers, which are presented in Table 7.
Table 7

Summary of Parkinson’s disease studies

Purpose

Type

Modality

Features

Classifier

Validation

Overall accuracy

Number of Subjects

Data basea

Author

Year

References

Classification

PD

T1WI

WM, GM, CSF volume in 98 ROIs

LS-SVM

10-fold CV

0.82

PD = 374, NC = 169, total = 543

PPMI

Adeli et al.

2016

[145]

Classification

PD, non-PD

FP-CIT SPECT

Uptake measures

Elastic-Net

Training/test (379/75)

0.99

PD = 72, SWEDD = 77, HC = 10, total = 159

PPMI + internal

Choi et al.

2017

[146]

Classification

PD

T1WI

Multilevel ROI features

JFSS + LDA

10-fold CV

0.86

PD = 69, HC = 103, total = 172

PPMI

Peng et al.

2017

[147]

Classification

PD

T1WI

Network measures, clinical scores

RF + SVM

10-fold CV

0.93

PD = 374, NC = 169, total = 543

PPMI

Amoroso et al.

2018

[148]

Classification

PD

FP-CIT SPECT

CSF, RNA, Serum test, image features

SVM

LOOCV

0.97

PD = 168, SWEDD = 26, HC = 194, total = 388

PPMI

Castillo-Barnes et al.

2018

[149]

Classification

PD

FP-CIT SPECT

Uptake measures

deep CNN

LOOCV

0.98

PD = 443, HC = 209, Total = 652

PPMI

Oliveira et al.

2018

[150]

Differentiation

PD, PSP

3D T1WI

Voxel-based pattern distribution

EPNN

LOOCV

0.858 (PD vs HC), 0.89(PSP vs HC), 0.89 (PD vs PSP)

PD = 28, PSP = 28, HC = 28, total = 84

Internal

Salvatore et al.

2014

[151]

Differentiation

PD, SWEDD

FP-CIT SPECT

Motor, non-motor, neuroimaging features

EPNN

10-fold CV

0.925(PD vs HC), AUC = 0.97 (SWEDD vs HC), 0.97(PD vs SWEDD)

N/A

PPMI

Hirschauer et al.

2015

[152]

Differentiation

PD, vascular PD

FP-CIT SPECT

Uptake measures

RLDA

10-fold CV

0.90

VPD = 80, PD = 164, total = 244

Internal

Huertas-Fernández et al.

2015

[153]

Differentiation

PD, SWEDD

T1WI

WM, GM, KSOM feature

LS-SVM

8-fold CV

0.99

PD = 518, SWEDD = 68, HC = 245, total = 831

PPMI

Singh et al.

2015

[154]

Differentiation

PD

T1WI

WM, GM features

SVM

8-fold CV

0.71

PD = 56, HC = 56, total = 112

PPMI

Liu et al.

2016

[155]

Differentiation

PD, MSA, PSP

DTI, R2*WI

FA, MD

SVM

10-fold CV

AUC = 0.88 (HC vs PD/MSA/PSP), 0.91 (HC vs PD), 0.94(PD vs MSA/PSP), 0.99(PD vs MSA), 0.99(PD vs PSP), 0.98(MSA vs PSP)

PD = 35, MSA = 16, PSP = 19, HC = 36, total = 106

Internal

Du et al.

2017

[156]

Prediction

PD, treatment response

DWI, fMRI

Subject-level general linear model

SVM

LOOCV

0.85

PD = 34, HC = 42, total = 76

Internal

Ye et al.

2016

[157]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from institutional and/or public through media channels

Attention-deficit hyperactivity disorder

ADHD is one of the most commonly found functional disorders affecting children. In 2011, a global competition (called ADHD-200) was held to use neuroimaging as well as phenotypic measures to automatically detect ADHD [158]. Approximately, half of the studies reviewed in this survey were based on that challenge. The main characteristics of those studies are tabulated in Table 8.
Table 8

Summary of ADHD studies

Purpose

Type

Modality

Feature

Classifier

Validation

Overall accuracy

Number of subjects

Data basea

Author

Year

References

Classification

ADHD, TDC, ADHD-I, ADHD-C

T1WI

Cortical features

SVM, ELM

10-fold CV

0.61

935

ADHD-200 dataset

Qureshi et al.

2016

[159]

Classification

ADHD, TDC, ADHD-I, ADHD-C

MRI, rs-fMRI

Structural, functional

SVM, ELM

Training/test (159/42)

0.76

935

ADHD-200 dataset

Qureshi et al.

2017

[160]

Classification

ADHD

T1WI, fMRI

Functional

SVM

10-fold CV

0.69

215

ADHD-200 dataset

Tan et al.

2017

[161]

Identification

ADHD

fMRI

Individual statistical maps

SVM (Gaussian process classifier)

LOOCV

0.77

60

Internal

Hart et al.

2014

[162]

Identification

ADHD

T1WI

T1WI

SVM

LOOCV

0.93

68

Internal

Johnston et al.

2014

[163]

Identification

ADHD

MRI, fMRI

HOG

SVM

5-fold CV

0.70

940

ADHD-200 dataset

Ghiassian et al.

2016

[164]

Identification

ADHD

MRI, fMRI

HOG

SVM

5-fold CV

0.65

1111

ABIDE

Ghiassian et al.

2016

[164]

Identification

ADHD

MRI

Structural

Least absolute shrinkage and selection operator (Lasso)

LOOCV

0.81

47

Internal

Xiao et al.

2016

[165]

Identification

ADHD

T1WI, DTI

GM, WM volume, FA, TR

SVM (non-linear GRBF kernel)

10-fold CV

0.66

133

Internal

Chaim-Avancini et al.

2017

[166]

Identification

ADHD

fMRI

Imaging + non-imaging data

SVM

LOOCV

0.87

N/A

ADHD-200 dataset

Riaz et al.

2018

[167]

Identification

ADHD

MRI, fMRI

Structural, functional

SVM

5-fold CV

0.67

729

ADHD-200 dataset

Sen et al.

2018

[168]

Identification

ADHD

fMRI

Structural, functional

SVM

5-fold CV

0.64

1099

ABIDE

Sen et al.

2018

[168]

Prediction

ADHD, treatment responder

fMRI

N/A

SVM (second-order polynomial kernel)

10-fold CV

0.85

N/A

Internal

Kim et al.

2015

[169]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from insitutional and/or public through media channels

Autism

ASD is a common disorder affecting youth, and share a high degree of comorbidity with other psychiatric disorders [170]. Both autism spectrum disorder (ASD) and ADHD are currently diagnosed on the basis of parent interview and clinical observation [171]. The identification of objective and reliable biomarkers is thus a critical yet elusive goal for neuroimaging researchers [172]. We surveyed 11 papers in classification of ASD mostly using functional MRI-based features and a prediction of ASD severity based on structure MRI. Those studies are listed in Table 9.
Table 9

Summary of autism studies

Purpose

Disease

Modality

Features

Classifier

Validation

Overall accuracy

Number of subjects

Data basea

Author

Year

References

Classification

Autism

fMRI, HRADI

Group level factors

GNB

N/A

0.97

Autism = 17, NC = 17, total = 34

Internal

Just et al.

2014

[173]

Classification

ASD

rs-fMR

Connectivity

RF

out-of-bag

0.91

ASD = 126, TD = 126TD, Total = 252

ABIDE

Chen et al.

2015

[174]

Classification

ASD

T1WI

Morphological

SVM

LPOCV

AUC = 0.74

ASD = 21, NC = 20, total = 41

Internal

Gori et al.

2015

[175]

Classification

ASD

rs-fMRI

Connectivity

PNN

LOOCV

0.90

ASD = 312, TD = 328TD, total = 640

ABIDE

Iidaka et al.

2015

[176]

Classification

ASD

rs-fMRI

Connectivity

L2LR

LOOCV

0.77

ASD = 59, TD = 59TD, total = 118

ABIDE

Plitt et al.

2015

[177]

classification

ASD

rs-fMRI

Connectivity

RF

out-of-bag

0.71

ASD = 126, NC = 126, Total = 252

ABIDE

Jahedi et al.

2017

[178]

Classification

ASD

rs-fMRI

Connectivity

DRBMs

10-fold CV

0.81

ASD = 61, NC = 72, total = 133

ABIDE(UM)

Kam et al.

2017

[179]

Classification

ASD

rs-fMRI, sMRI

Connectivity, structure

DBN

10-fold CV

0.66

ASD = 116, NC = 69, total = 185

ABIDE I, II

Aghdam et al.

2018

[180]

Classification

ASD

rs-fMRI

Connectivity

random SVM cluster

Training/test (58/26)

0.96

ASD = 45, TD = 39TD, Total = 84

ABIDE

Bi et al.

2018

[181]

Classification

ASD

rs-fMRI

Functional activity

DNN

10-fold CV

0.70

ASD = 505, NC = 530, total = 1035

ABIDE

Heinsfeld et al.

2018

[182]

Classification

ASD

rs-fMRI

Connectivity

DTL-NN

5-fold CV

0.67

ASD = 149, NC = 161, total = 310

ABIDE

Li et al.

2018

[183]

Prediction

ASD, severity

T1WI

Cortical thickness

SVM

10-fold CV

r = 0.51 ± 0.04

ASD = 156

ABIDE

Moradi et al.

2017

[184]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from insitutional and/or public through media channels

Epilepsy

The main purposes of ML studies on epilepsy were classification of epilepsy from other type of diseases, detection, and prediction of treatment outcome. MR imaging was a main modality used for ML studies. Depending on the purpose, variety of MR imaging methods, such as T1WI and DTI were employed. Table 10 summarizes the epilepsy studies.
Table 10

Summary of epilepsy studies

Purpose

Tyep

Modality

Feature

Classifier

Validation

Overall accuracy

Number of subjects

Data basea

Author

Year

References

Classification

Epilepsy, MTS

MPRAGE

Morphological measures

SVM-RFE

10-fold CV

0.81

wMTS = 85, w/oMTS = 84

Internal

Rudie et al.

2015

[185]

Classification

Epilepsy, FCD

T2WI

Cortical thickness

SVM

LOOCV

0.98

FCD type I = 13, type II = 28, NC = 41, total = 82

Internal

Hong et al.

2016

[186]

Classification

TLE, lateralization

DTI

Structural connectomes

SVM

LOOCV

0.897 (right), 0.86 (left), 0.864 (control)

TLE = 44, NC = 14, total = 58

Internal

Kamiya et al.

2016

[187]

Classification

TLE

DTI, DKI

Diffusion metrics

SVM

5-fold CV

0.82

Epilepsy = 32, NC = 36, total = 68

Internal

Del Gaizo et al.

2017

[188]

Detection

Epilepsy, cortial dysplasia type II

MRI

Surface-based features

LDA

LOOCV

0.74 (sensitivity)

Epilepsy = 19, NC = 24, total = 43

Internal

Hong et al.

2014

[189]

Detection

Epilepsy

T1WI

Texture parameters

OC-SVM

N/A

0.769 (sensitivity)

Epilepsy = 11 NC = 77, total = 88

Internal

Azami et a.l

2016

[190]

Detection

Epilepsy

T1WI

Surface-based features

NN

LOOCV

0.87 (AUC)

Epilepsy = 22, NC = 28, Total = 50

Internal

Adler et al.

2017

[191]

Prediction

TLE, treatment outcome

T1WI

Surface-based features

k-Means clustering

LOOCV

0.81

TLE-I = 24, TLE-II = 32, TLE-III = 34, TLE-IV = 24, NC = 42, total = 156

Internal

Bernhardt et al.

2015

[192]

Prediction

Epilepsy, treatment outcome

DTI

Structural connectome

Elastic net

10-fold CV

0.80

Epilepsy = 35, NC = 18, total = 53

Internal

Munsell et al.

2015

[193]

Prediction

Epilepsy, laterality

PET, MRI, and DTI

Glucose metabolism, cortical thickness, WM atrophy

Logistic regression

N/A

1.00

Left TLE = 28, right TLE = 30

Internal

Pustina et al.

2015

[194]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from insitutional and/or public through media channels

Multiple sclerosis

The main purposes of ML studies on MS were classification, detection, and segmentation of MS lesion. Both structural and functional MRI were the main imaging methods used for ML studies. Table 11 summarizes the MS studies.
Table 11

Summary of multiple sclerosis studies

Purpose

Type

Modality

Feature

Classifier

Validation

Overall accuracy

Number of subjects

Data basea

Author

Year

References

Classification

Motor function

rs-fMRI, DTI

GMM + FC

SVM

LOOCV

0.8834 (HC vs MI), 0.84 (HC vs MP), 0.86 (MP vs MI)

MP = 26MP, MI = 25, NC = 21, total = 72

Internal

Zhong et al.

2017

[195]

Classification

MS, NAWM, HC

MRI

Myelin, T1WI features

Deep CNN, RF

11-fold CV

0.88

MS = 55, NC = 44, total = 99

Internal

Yoo et al.

2018

[196]

Classification

Relapsing-remitting MS

T1WI, rs-fMRI, DTI

FA, connectivity, function correlation

SVM

10-fold CV, LOOCV

0.89

MS = 104, NC = 46, total = 150

Internal

Zurita et al.

2018

[197]

Detection

MS new lesion

rs-fMRI

Functional connectivity

SVM, RF

5-fold CV

0.86

MS = 18, NC = 19, total = 37

Internal

Saccà et al.

2018

[198]

Detection

MS

T2WI

DF information

LRM

LOOCV

0.77

MS = 36

Internal

Salem et al.

2018

[199]

Prediction

Worsening cases

T2WI

Normalized T2 hyperintense lesion volume (T2LV)

SVM + bagging

10-fold CV

0.62

MS = 1693

CLIMB

Zhao et al.

2017

[200]

Segmentation

MS lesion

T1WI, MPRAGE, T2WI, PDWI, FLAIR

structural

Dictionary learning

LOSOCV

0.50

MS = 14

Internal

Deshpande et al.

2015

[201]

Segmentation

MS lesion

MPRAGE, MP2RAGE, FLAIR, DIR

Image intensity

k-NN

LOOCV

0.75

MS = 39

Internal

Fartaria et al.

2016

[202]

aInternal: subjects were recruited from insitutional and/or public through media channels

Stroke

ML techniques have been applied for stroke imaging in two different aspects: i.e., automatic or accurate diagnosis and prediction of prognosis [203]. Automatic lesion identification or segmentation is one of the most important elements in precision medicine dealing with huge datasets of brain imaging. This is because manual lesion segmentation is cumbersome and inconsistent across raters [204]. Prediction of treatment complications may be useful for screening a high-risk group receiving acute treatment, such as thrombolysis [205], whereas prediction of neurological long-term outcomes may guide the stroke management [206]. Both CT and structural MRI were a main modalities. Table 12 summarizes the stroke studies.
Table 12

Summary of stroke studies

Purpose

Type

Modality

Feature

Classifier

Validation

Overall accuracy

Number of subjects

Data basea

Author

Year

References

Detection

Early infarction signs

CT

ASPECTS score

e-ASPECTS, version 6.0b

N/A

0.67

Stroke = 119

Internal

Guberina et al.

2018

[207]

Identification

Motor disability

rs-fMRI

Resting-state connectivity

SVM

LOOCV

0.88

Stroke = 20 (Training), 20 (test), NC = 20, Total = 60

Internal

Rehme et al.

2015

[208]

Prediction

Post intra-arterial therapy outcome

CT

The presence of acute stroke and intracranial haemorrhage

ANN, SVM

LOOCV

0.87

Stroke = 107

Internal

Asadi et al.

2014

[209]

Prediction

Thrombolysis outcome, SICH

CT

CT-value

SVM

10-fold CV

0.744 (AUC)

SICH = 63, noSICH = 49

Internal

Bentley et al.

2014

[210]

Segmentation

Stroke lesion

MRI

Intensity feature, the weighted local mean, the 2D center distance and the local histogram

RDF

N/A

0.67

Stroke = 37

Internal

Maier et al.

2015

[211]

Segmentation

Chronic stroke lesion

T1WI

Feature map

Gaussian naïve Bayes

LOOCV

0.66

Stroke = 30

Internal

Griffis et al.

2016

[212]

Segmentation

Chronic stroke lesion

T1WI

12 geometric features

LINDA

6-fold CV

0.696

Stroke = 60 (training), 45 (test), NC = 80, total = 185

Internal

Pustina et al.

2016

[213]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from insitutional and/or public through media channels

be-ASPECTS. Brainomix®, Oxford, UK. https://www.brainomix.com/ (accessed 10/2018)

Traumatic brain injury

Neuroimaging plays a critical role in the acute setting to guide appropriate management by detecting injuries that require intervention or further monitoring [214]. Among various MR imaging techniques, the DTI metrics are thought to reflect the integrity of microstructural properties of white matter and have been applied extensively as neuroimaging biomarkers to study a range of clinical conditions [215]. The main purposes of ML studies on TBI were classification of mild traumatic brain injury (mTBI) from NC and/or post concussion syndrome (PCS) and posttraumatic stress disorder (PTSD). CT, DTI, and fMRI were the main modality used for ML studies on TBI. Table 13 summarizes the TBI studies.
Table 13

Summary of traumatic brain injury studies

Purpose

Type

Modality

Feature

Classifier

Validation

Overall accuracy

Number of patients

Data basea

Author

Year

References

Classification

TBI

DTI

FA measures, network connectivity

Decision ensemble

10-fold CV

0.68

TBI = 179, NC = 146, total = 325

Internal

Mitra et al.

2016

[216]

Classification

mTBI

CT

N/A

RF

10-fold CV

0.78

mTBI = 334 (179 training, 155 test), HC = 328, total = 662

Internal

Peacock IV et al.

2017

[217]

Classification

PTSD, mTBI

MRI, DTI

Static and dynamic functional connectivity

RF

6-fold CV

0.84

PTSD = 17, PCS + PTSD = 42, NC = 28, total = 87

Internal

Rangaprakash et al.

2017

[218]

Classification

mTBI

rs-fMRI, DTI

Network connectivity, FA

SVM

LOOCV

0.84

mTBI = 50, NC = 50, total = 100

Internal

Vergara et al.

2017

[219]

Classification

Disease foci

rs-fMRI

Effective connectivity networks

RF

6-fold CV

0.81

PTSD = 87, PCS + PTSD = 87, NC = 87, total = 261

Internal

Rangaprakash et al.

2018

[220]

Identification

TBI

DTI

WM connectivity

SVM

10-fold CV

0.93

TBI = 52, NC = 25, total = 77

Internal

Fagerholm et al.

2015

[221]

Prediction

Severity

CT

N/A

RCE-SVM

N/A

0.98 (AUC)

TBI = 39, NC = 156, total = 195

Internal

Chong et al.

2015

[222]

N/A indicates information was not available or could not be found

aInternal: subjects were recruited from insitutional and/or public through media channels

Study characteristics

Number of publications

Figure 2 shows the number of papers published in each year for each disease type. The number of studies showed a growing trend since 2014 (2018 was not an entire year, the term was 1 January–19 September). As shown in Table 2, top three researches are strongly supported by big data sets. Typically, 92.7% of AD/MCI studies were carried out on large data sets, such as ADNI, OASIS [223], DZNE [224], PRODEM [225], and AIBL [226]. This clearly indicates that the well-maintained database supports the progress of technology and their publications.
Fig. 2

Total number of papers within each year for every disease or disorder

Overall accuracy vs. total sample size

Figure 3 shows the overall accuracy against the total sample size used in the studies. Almost all studies that reported very high accuracies (69.8% of studies were higher than 0.8) had sample sizes smaller than 100 (45.6% of studies). The reported overall accuracy decreases with sample size in most of the disorders, such as schizophrenia and ADHD. This pattern raises a serious concern regarding generalizability of many of those studies with small sample sizes. On the other hands, only PD studies (* in Fig. 3) showed positive relationships between the accuracy and the total sample size. Parkinson’s Progression Markers Initiative (PPMI) may play an important role. In contrast, the result of MRI measures varies from facility-to-facility and the standardization still appears to be an open problem [227].
Fig. 3

Scatter plot of overall reported accuracy versus the total sample size

Total sample size distribution

Figure 4 shows the sample size distribution. The dashed lines represent mean (red) and median (blue) sizes, which are 231 and 120, respectively.
Fig. 4

Histogram of the sample sizes of the surveyed studies. Vertical dashed lines indicate mean (red) and median (blue) sample size among all studies, which are 231 and 120, respectively

When using diagnosis models that are very complex or have many parameters on datasets with small number of samples, overfitting tends to take place. Usually, ML creates classifier from training dataset. In the training session, ML sometimes extracts unique characteristics, which are only based on the training datasets. Therefore, an overfitted model provides good results on the training data and poor results on the test data. Neuroimaging datasets tend to have limited sample size and millions of voxels per sample. The majority of surveyed studies built predictive models based on a very small number of subjects. Therefore, it is plausible that many surveyed studies suffer from overfitting problem. Cross validation, simple classifier, and proper feature selections can help avoid overfitting.

Accuracy for each disorder

Figure 5 illustrates the summary statistics of reported overall accuracy for each disorder. AD/MCI and PD studies exceeded 0.9 in median. In contrast, ADHD and stroke studies were less than 0.7 in their medians. Others had around 0.8 to < 0.9 in their median of accuracy.
Fig. 5

Disorder-specific box plots of reported overall accuracies of the surveyed papers. Dot including circles indicate the median accuracy

Classification method

Figure 6 shows the classifiers used in the studies. In terms of classification methods, support vector machine (SVM) was still the most popular method (more than 55%, 117/209 papers). Different strategy of SVM such as linear, non-linear with different kernel, SVM with recursive feature elimination, SVM with L1 regularization and SVM with L1 and L2 regularization (elastic net) have been used for classification of various disorders.
Fig. 6

Classifiers used for ML of the surveyed papers

Although deep neural network (DNN) is attracting attention in recent years [203, 228, 229], there were still only 18 papers published in the last 5 years using DNN. Additionally, random forest (RF) was used in 11 papers.

Validation method

Figure 7 shows the validation methods used in the studies. In terms of classifier validation methods, k-hold cross validation (CV) is the most popular method (more than 80%, 169/209 papers). The constant k was varying; k = 1–10. Among them, the popular methods were leave-one-out cross validation (LOOCV) and tenfold CV.
Fig. 7

Validation methods used for ML of the surveyed papers

Ideally, training/test scheme should be chosen for the validation of classifier, but more than 80% of papers made k-hold CV because of lack of subject. Only 9.1% (19 papers) made training/test scheme for the validation of classifier.

Limitations

There are several limitations in this work. Firstly, we limited our search to English journal papers. Secondly, there are other studies based on other modalities such as EEG MEG, and US, which were not included in this study. Moreover, other important conditions such bipolar and anxiety disorders were not reviewed in this work. Many of the papers contained multiple experiments under different scenarios, such as multiple ML methods, data reduction methods, and multiple data sets, but we just reported one of the most successful remarks.

Summary

In this paper, we comprehensively reviewed 11 diseases neuroimaging-based 209 ML studies in the recent 5 years period of time. To clarify the recent trends of ML studies in the field of neuroradiology, we summarized ML methods used in these research, the number of data, image features, and the overall accuracy. We have also shown that there are several bottlenecks, such as feature selection bias and overfitting by small sample size.

It is clear that the result of ML, which is a statistical model fitting method, depends on its sample size. Therefore, as many previous review papers have pointed out, the existence of large data sets is indispensable for promoting ML study. More importantly, multidisciplinary cooperation remains a crucial aspect, and if achieved, Wang and Summers’s expectations might become true.

Notes

Funding

One of the authors (K. Y.) was funded by following companies (within the past 12 months): Nihon Medi-Physics Co., Ltd., Daiichi Sankyo Co., Ltd., Fuji Pharma Co.,Ltd., Doctor-Net Inc, and Fujifilm RI Pharma Co., Ltd.

Compliance with ethical standards

Ethical statements

This article does not contain any studies with human participants or animals performed by any of the authors.

Conflict of interest

The authors declare that they have no conflict of interest.

References

  1. 1.
    Wang S, Summers RM. Machine learning and radiology. Med Image Anal. 2012;16:933–51.CrossRefPubMedPubMedCentralGoogle Scholar
  2. 2.
    Arbabshirani MR, Plis S, Sui J, Calhoun VD. Single subject prediction of brain disorders in neuroimaging: promises and pitfalls. Neuroimage. 2017;145:137–65.CrossRefPubMedGoogle Scholar
  3. 3.
    US national library of medicine national institutes of health. https://www.ncbi.nlm.nih.gov/pubmed/. Accessed Oct 2018.
  4. 4.
    Moher D. Preferred reporting items for systematic reviews and meta-analyses. J Clin Epidemiol. 2009;62:1006–12.CrossRefGoogle Scholar
  5. 5.
    Jie B, Zhang D, Gao W, Wang Q, Wee CY, Shen D. Integration of network topological and connectivity properties for neuroimaging classification. IEEE Trans Biomed Eng. 2014;61(2):576–89.CrossRefPubMedPubMedCentralGoogle Scholar
  6. 6.
    Li M, Oishi K, He X, Qin Y, Gao F, Mori S, et al. An efficient approach for differentiating Alzheimer’s disease from normal elderly based on multicenter MRI using gray-level invariant features. PLoS One. 2014;9(8):e105563.CrossRefPubMedPubMedCentralGoogle Scholar
  7. 7.
    Li M, Qin Y, Gao F, Zhu W, He X. Discriminative analysis of multivariate features from structural MRI and diffusion tensor images. Magn Reson Imaging. 2014;32(8):1043–51.CrossRefPubMedGoogle Scholar
  8. 8.
    Li S, Yuan X, Pu F, Li D, Fan Y, Wu L, et al. Abnormal changes of multidimensional surface features using multivariate pattern classification in amnestic mild cognitive impairment patients. J Neurosci. 2014;34(32):10541–53.CrossRefPubMedPubMedCentralGoogle Scholar
  9. 9.
    Liu M, Zhang D, Shen D, Alzheimer’s Disease Neuroimaging I. Identifying informative imaging biomarkers via tree structured sparse learning for AD diagnosis. Neuroinformatics. 2014;12(3):381–94.CrossRefPubMedPubMedCentralGoogle Scholar
  10. 10.
    Segovia F, Bastin C, Salmon E, Gorriz JM, Ramirez J, Phillips C. Combining PET images and neuropsychological test data for automatic diagnosis of Alzheimer’s disease. PLoS One. 2014;9(2):e88687.CrossRefPubMedPubMedCentralGoogle Scholar
  11. 11.
    Suk HI, Lee SW, Shen D, Alzheimer’s Disease Neuroimaging I. Hierarchical feature representation and multimodal fusion with deep learning for AD/MCI diagnosis. Neuroimage. 2014;101:569–82.CrossRefPubMedPubMedCentralGoogle Scholar
  12. 12.
    Cheng B, Liu M, Suk HI, Shen D, Zhang D, Alzheimer’s Disease Neuroimaging I. Multimodal manifold-regularized transfer learning for MCI conversion prediction. Brain Imaging Behav. 2015;9(4):913–26.CrossRefPubMedPubMedCentralGoogle Scholar
  13. 13.
    Dyrba M, Grothe M, Kirste T, Teipel SJ. Multimodal analysis of functional and structural disconnection in Alzheimer’s disease using multiple kernel SVM. Hum Brain Mapp. 2015;36(6):2118–31.CrossRefGoogle Scholar
  14. 14.
    Farzan A, Mashohor S, Ramli AR, Mahmud R. Boosting diagnosis accuracy of Alzheimer’s disease using high dimensional recognition of longitudinal brain atrophy patterns. Behav Brain Res. 2015;290:124–30.CrossRefPubMedGoogle Scholar
  15. 15.
    Jie B, Zhang D, Cheng B, Shen D, Alzheimer’s Disease Neuroimaging I. Manifold regularized multitask feature learning for multimodality disease classification. Hum Brain Mapp. 2015;36(2):489–507.CrossRefGoogle Scholar
  16. 16.
    Khazaee A, Ebrahimzadeh A, Babajani-Feremi A. Identifying patients with Alzheimer’s disease using resting-state fMRI and graph theory. Clin Neurophysiol. 2015;126(11):2132–41.CrossRefPubMedGoogle Scholar
  17. 17.
    Liu M, Zhang D, Shen D, Alzheimer’s Disease Neuroimaging I. View-centralized multi-atlas classification for Alzheimer’s disease diagnosis. Hum Brain Mapp. 2015;36(5):1847–65.CrossRefPubMedGoogle Scholar
  18. 18.
    Liu S, Liu S, Cai W, Che H, Pujol S, Kikinis R, et al. Multimodal neuroimaging feature learning for multiclass diagnosis of Alzheimer’s disease. IEEE Trans Biomed Eng. 2015;62(4):1132–40.CrossRefPubMedGoogle Scholar
  19. 19.
    Moradi E, Pepe A, Gaser C, Huttunen H, Tohka J, Alzheimer’s Disease Neuroimaging I. Machine learning framework for early MRI-based Alzheimer’s conversion prediction in MCI subjects. Neuroimage. 2015;104:398–412.CrossRefPubMedGoogle Scholar
  20. 20.
    Xu L, Wu X, Chen K, Yao L. Multi-modality sparse representation-based classification for Alzheimer’s disease and mild cognitive impairment. Comput Methods Progr Biomed. 2015;122(2):182–90.CrossRefGoogle Scholar
  21. 21.
    Zhu X, Suk HI, Zhu Y, Thung KH, Wu G, Shen D. Multi-view classification for identification of Alzheimer’s disease. Mach Learn Med Imaging. 2015;9352:255–62.CrossRefPubMedPubMedCentralGoogle Scholar
  22. 22.
    Cattell L, Platsch G, Pfeiffer R, Declerck J, Schnabel JA, Hutton C, et al. Classification of amyloid status using machine learning with histograms of oriented 3D gradients. Neuroimage Clin. 2016;12:990–1003.CrossRefPubMedPubMedCentralGoogle Scholar
  23. 23.
    Hu C, Sepulcre J, Johnson KA, Fakhri GE, Lu YM, Li Q. Matched signal detection on graphs: theory and application to brain imaging data classification. Neuroimage. 2016;125:587–600.CrossRefPubMedGoogle Scholar
  24. 24.
    Khazaee A, Ebrahimzadeh A, Babajani-Feremi A. Application of advanced machine learning methods on resting-state fMRI network for identification of mild cognitive impairment and Alzheimer’s disease. Brain Imaging Behav. 2016;10(3):799–817.CrossRefPubMedGoogle Scholar
  25. 25.
    Liu M, Zhang D, Shen D. Relationship induced multi-template learning for diagnosis of Alzheimer’s disease and mild cognitive impairment. IEEE Trans Med Imaging. 2016;35(6):1463–74.CrossRefPubMedPubMedCentralGoogle Scholar
  26. 26.
    Ni H, Zhou L, Ning X, Wang L, Alzheimer’s Disease Neuroimaging I. Exploring multifractal-based features for mild Alzheimer’s disease classification. Magn Reson Med. 2016;76(1):259–69.CrossRefPubMedGoogle Scholar
  27. 27.
    Suk HI, Wee CY, Lee SW, Shen D. State-space model with deep learning for functional dynamics estimation in resting-state fMRI. Neuroimage. 2016;129:292–307.CrossRefPubMedPubMedCentralGoogle Scholar
  28. 28.
    Yu G, Liu Y, Shen D. Graph-guided joint prediction of class label and clinical scores for the Alzheimer’s disease. Brain Struct Funct. 2016;221(7):3787–801.CrossRefPubMedGoogle Scholar
  29. 29.
    Zu C, Jie B, Liu M, Chen S, Shen D, Zhang D, et al. Label-aligned multi-task feature learning for multimodal classification of Alzheimer’s disease and mild cognitive impairment. Brain Imaging Behav. 2016;10(4):1148–59.CrossRefPubMedPubMedCentralGoogle Scholar
  30. 30.
    An L, Adeli E, Liu M, Zhang J, Lee SW, Shen D. A hierarchical feature and sample selection framework and its application for Alzheimer’s disease diagnosis. Sci Rep. 2017;7:45269.CrossRefPubMedPubMedCentralGoogle Scholar
  31. 31.
    Beheshti I, Demirel H, Matsuda H, Alzheimer’s Disease Neuroimaging I. Classification of Alzheimer’s disease and prediction of mild cognitive impairment-to-Alzheimer’s conversion from structural magnetic resource imaging using feature ranking and a genetic algorithm. Comput Biol Med. 2017;83:109–19.CrossRefPubMedGoogle Scholar
  32. 32.
    Chen X, Zhang H, Zhang L, Shen C, Lee SW, Shen D. Extraction of dynamic functional connectivity from brain grey matter and white matter for MCI classification. Hum Brain Mapp. 2017;38(10):5019–34.CrossRefPubMedPubMedCentralGoogle Scholar
  33. 33.
    Doan NT, Engvig A, Zaske K, Persson K, Lund MJ, Kaufmann T, et al. Distinguishing early and late brain aging from the Alzheimer’s disease spectrum: consistent morphological patterns across independent samples. Neuroimage. 2017;158:282–95.CrossRefPubMedGoogle Scholar
  34. 34.
    Glozman T, Solomon J, Pestilli F, Guibas L, Alzheimer’s Disease Neuroimaging I. Shape-attributes of brain structures as biomarkers for Alzheimer’s disease. J Alzheimers Dis. 2017;56(1):287–95.CrossRefPubMedPubMedCentralGoogle Scholar
  35. 35.
    Guan H, Liu T, Jiang J, Tao D, Zhang J, Niu H, et al. Classifying MCI subtypes in community-dwelling elderly using cross-sectional and longitudinal MRI-Based biomarkers. Front Aging Neurosci. 2017;9:309.CrossRefPubMedPubMedCentralGoogle Scholar
  36. 36.
    Guo H, Zhang F, Chen J, Xu Y, Xiang J. Machine learning classification combining multiple features of a hyper-network of fMRI data in Alzheimer’s disease. Front Neurosci. 2017;11:615.CrossRefPubMedPubMedCentralGoogle Scholar
  37. 37.
    Hojjati SH, Ebrahimzadeh A, Khazaee A, Babajani-Feremi A, Alzheimer’s Disease Neuroimaging I. Predicting conversion from MCI to AD using resting-state fMRI, graph theoretical approach and SVM. J Neurosci Methods. 2017;282:69–80.CrossRefPubMedGoogle Scholar
  38. 38.
    Kawaguchi A, Yamashita F, Alzheimer’s Disease Neuroimaging I. Supervised multiblock sparse multivariable analysis with application to multimodal brain imaging genetics. Biostatistics. 2017;18(4):651–65.CrossRefPubMedGoogle Scholar
  39. 39.
    Khazaee A, Ebrahimzadeh A, Babajani-Feremi A, Alzheimer’s Disease Neuroimaging I. Classification of patients with MCI and AD from healthy controls using directed graph measures of resting-state fMRI. Behav Brain Res. 2017;322(Pt B):339–50.CrossRefPubMedGoogle Scholar
  40. 40.
    Lama RK, Gwak J, Park JS, Lee SW. Diagnosis of Alzheimer’s disease based on structural MRI images using a regularized extreme learning machine and PCA features. J Healthc Eng. 2017;2017:5485080.CrossRefPubMedPubMedCentralGoogle Scholar
  41. 41.
    Li Q, Wu X, Xu L, Chen K, Yao L, Li R. Multi-modal discriminative dictionary learning for Alzheimer’s disease and mild cognitive impairment. Comput Methods Progr Biomed. 2017;150:1–8.CrossRefGoogle Scholar
  42. 42.
    Long X, Chen L, Jiang C, Zhang L, Alzheimer’s Disease Neuroimaging I. Prediction and classification of Alzheimer disease based on quantification of MRI deformation. PLoS One. 2017;12(3):e0173372.CrossRefPubMedPubMedCentralGoogle Scholar
  43. 43.
    Min J, Moon WJ, Jeon JY, Choi JW, Moon YS, Han SH. Diagnostic efficacy of structural MRI in patients with mild-to-moderate Alzheimer disease: automated volumetric assessment versus visual assessment. AJR Am J Roentgenol. 2017;208(3):617–23.CrossRefPubMedGoogle Scholar
  44. 44.
    Proitsi P, Kim M, Whiley L, Simmons A, Sattlecker M, Velayudhan L, et al. Association of blood lipids with Alzheimer’s disease: a comprehensive lipidomics analysis. Alzheimers Dement. 2017;13(2):140–51.CrossRefPubMedGoogle Scholar
  45. 45.
    Sorensen L, Igel C, Pai A, Balas I, Anker C, Lillholm M, et al. Differential diagnosis of mild cognitive impairment and Alzheimer’s disease using structural MRI cortical thickness, hippocampal shape, hippocampal texture, and volumetry. Neuroimage Clin. 2017;13:470–82.CrossRefPubMedGoogle Scholar
  46. 46.
    Suk HI, Lee SW, Shen D, Alzheimer’s Disease Neuroimaging I. Deep ensemble learning of sparse regression models for brain disease diagnosis. Med Image Anal. 2017;37:101–13.CrossRefPubMedPubMedCentralGoogle Scholar
  47. 47.
    Youssofzadeh V, McGuinness B, Maguire LP, Wong-Lin K. Multi-Kernel learning with Dartel improves combined MRI-PET classification of Alzheimer’s disease in AIBL data: group and individual analyses. Front Hum Neurosci. 2017;11:380.CrossRefPubMedPubMedCentralGoogle Scholar
  48. 48.
    Zhu X, Suk HI, Lee SW, Shen D. Discriminative self-representation sparse regression for neuroimaging-based Alzheimer’s disease diagnosis. Brain Imaging Behav. 2017.  https://doi.org/10.1007/s11682-017-9731-x.CrossRefPubMedPubMedCentralGoogle Scholar
  49. 49.
    Zhu X, Suk HI, Wang L, Lee SW, Shen D, Alzheimer’s Disease Neuroimaging I. A novel relational regularization feature selection method for joint regression and classification in AD diagnosis. Med Image Anal. 2017;38:205–14.CrossRefPubMedGoogle Scholar
  50. 50.
    Amoroso N, Rocca M, Bellotti R, Fanizzi A, Monaco A, Tangaro S, et al. Alzheimer’s disease diagnosis based on the hippocampal unified multi-atlas network (HUMAN) algorithm. Biomed Eng Online. 2018;17(1):6.CrossRefPubMedPubMedCentralGoogle Scholar
  51. 51.
    Choi H, Jin KH, Alzheimer’s Disease Neuroimaging I. Predicting cognitive decline with deep learning of brain metabolism and amyloid imaging. Behav Brain Res. 2018;344:103–9.CrossRefPubMedGoogle Scholar
  52. 52.
    Garali I, Adel M, Bourennane S, Guedj E. Histogram-Based features selection and volume of interest ranking for brain PET image classification. IEEE J Transl Eng Health Med. 2018;6:2100212.CrossRefPubMedGoogle Scholar
  53. 53.
    Kim J, Lee B. Identification of Alzheimer’s disease and mild cognitive impairment using multimodal sparse hierarchical extreme learning machine. Hum Brain Mapp. 2018.  https://doi.org/10.1002/hbm.24207.CrossRefPubMedPubMedCentralGoogle Scholar
  54. 54.
    Liu M, Cheng D, Wang K, Wang Y, Alzheimer’s Disease Neuroimaging I. Multi-modality cascaded convolutional neural networks for Alzheimer’s disease diagnosis. Neuroinformatics. 2018;16(3–4):295–308.CrossRefPubMedGoogle Scholar
  55. 55.
    Rondina JM, Ferreira LK, de Souza Duran FL, Kubo R, Ono CR, Leite CC, et al. Selecting the most relevant brain regions to discriminate Alzheimer’s disease patients from healthy controls using multiple kernel learning: a comparison across functional and structural imaging modalities and atlases. Neuroimage Clin. 2018;17:628–41.CrossRefPubMedGoogle Scholar
  56. 56.
    Salvatore C, Castiglioni I. A wrapped multi-label classifier for the automatic diagnosis and prognosis of Alzheimer’s disease. J Neurosci Methods. 2018;302:58–65.CrossRefPubMedGoogle Scholar
  57. 57.
    Sun Z, Qiao Y, Lelieveldt BPF, Staring M, Alzheimer’s Disease NeuroImaging I. Integrating spatial-anatomical regularization and structure sparsity into SVM: improving interpretation of Alzheimer’s disease classification. Neuroimage. 2018;178:445–60.CrossRefPubMedGoogle Scholar
  58. 58.
    Wang SH, Phillips P, Sui Y, Liu B, Yang M, Cheng H. Classification of Alzheimer’s disease based on eight-layer convolutional neural network with leaky rectified linear unit and max pooling. J Med Syst. 2018;42(5):85.CrossRefPubMedGoogle Scholar
  59. 59.
    Chen Y, Sha M, Zhao X, Ma J, Ni H, Gao W, et al. Automated detection of pathologic white matter alterations in Alzheimer’s disease using combined diffusivity and kurtosis method. Psychiatry Res Neuroimaging. 2017;264:35–45.CrossRefPubMedGoogle Scholar
  60. 60.
    Schouten TM, Koini M, Vos F, Seiler S, Rooij M, Lechner A, et al. Individual classification of Alzheimer’s disease with diffusion magnetic resonance imaging. Neuroimage. 2017;152:476–81.CrossRefPubMedGoogle Scholar
  61. 61.
    Sorensen L, Igel C, Liv Hansen N, Osler M, Lauritzen M, Rostrup E, et al. Early detection of Alzheimer’s disease using MRI hippocampal texture. Hum Brain Mapp. 2016;37(3):1148–61.CrossRefPubMedGoogle Scholar
  62. 62.
    Varol E, Sotiras A, Davatzikos C, Alzheimer’s Disease Neuroimaging I. HYDRA: revealing heterogeneity of imaging and genetic patterns through a multiple max-margin discriminative analysis framework. Neuroimage. 2017;145(Pt B):346–64.CrossRefPubMedGoogle Scholar
  63. 63.
    Cai K, Xu H, Guan H, Zhu W, Jiang J, Cui Y, et al. Identification of early-stage Alzheimer’s disease using sulcal morphology and other common neuroimaging indices. PLoS One. 2017;12(1):e0170875.CrossRefPubMedPubMedCentralGoogle Scholar
  64. 64.
    Jie B, Liu M, Liu J, Zhang D, Shen D. Temporally constrained group sparse learning for longitudinal data analysis in Alzheimer’s disease. IEEE Trans Biomed Eng. 2017;64(1):238–49.CrossRefPubMedGoogle Scholar
  65. 65.
    Zhou M, Scott J, Chaudhury B, Hall L, Goldgof D, Yeom KW, Iv M, Ou Y, Kalpathy-Cramer J, Napel S, Gillies R, Gevaert O, Gatenby R. Radiomics in brain tumor: image assessment, quantitative feature descriptors, and machine-learning approaches. AJNR Am J Neuroradiol. 2018;39:208–16.CrossRefPubMedGoogle Scholar
  66. 66.
    Imani F, Boada FE, Lieberman FS, Davis DK, Mountz JM. Molecular and metabolic pattern classification for detection of brain glioma progression. Eur J Radiol. 2014;83(2):e100–5.CrossRefPubMedGoogle Scholar
  67. 67.
    Nachimuthu DS, Baladhandapani A. Multidimensional texture characterization: on analysis for brain tumor tissues using MRS and MRI. J Digit Imaging. 2014;27(4):496–506.CrossRefPubMedPubMedCentralGoogle Scholar
  68. 68.
    Sunwoo L, Kim YJ, Choi SH, Kim KG, Kang JH, Kang Y, et al. Computer-aided detection of brain metastasis on 3D MR imaging: observer performance study. PLoS One. 2017;12(6):e0178265.CrossRefPubMedPubMedCentralGoogle Scholar
  69. 69.
    Wu W, Chen AY, Zhao L, Corso JJ. Brain tumor detection and segmentation in a CRF (conditional random fields) framework with pixel-pairwise affinity and superpixel-level features. Int J Comput Assist Radiol Surg. 2014;9(2):241–53.CrossRefPubMedGoogle Scholar
  70. 70.
    Fathi Kazerooni A, Mohseni M, Rezaei S, Bakhshandehpour G, Saligheh Rad H. Multi-parametric (ADC/PWI/T2-w) image fusion approach for accurate semi-automatic segmentation of tumorous regions in glioblastoma multiforme. MAGMA. 2015;28(1):13–22.CrossRefPubMedGoogle Scholar
  71. 71.
    Szwarc P, Kawa J, Rudzki M, Pietka E. Automatic brain tumour detection and neovasculature assessment with multiseries MRI analysis. Comput Med Imaging Graph. 2015;46(Pt 2):178–90.CrossRefPubMedGoogle Scholar
  72. 72.
    Havaei M, Davy A, Warde-Farley D, Biard A, Courville A, Bengio Y, et al. Brain tumor segmentation with deep neural networks. Med Image Anal. 2017;35:18–31.CrossRefPubMedGoogle Scholar
  73. 73.
    Li Y, Liu X, Wei F, Sima DM, Van Cauter S, Himmelreich U, et al. An advanced MRI and MRSI data fusion scheme for enhancing unsupervised brain tumor differentiation. Comput Biol Med. 2017;81:121–9.CrossRefPubMedGoogle Scholar
  74. 74.
    Sauwen N, Acou M, Sima DM, Veraart J, Maes F, Himmelreich U, et al. Semi-automated brain tumor segmentation on multi-parametric MRI using regularized non-negative matrix factorization. BMC Med Imaging. 2017;17(1):29.CrossRefPubMedPubMedCentralGoogle Scholar
  75. 75.
    AlBadawy EA, Saha A, Mazurowski MA. Deep learning for segmentation of brain tumors: Impact of cross-institutional training and testing. Med Phys. 2018;45(3):1150–8.CrossRefPubMedGoogle Scholar
  76. 76.
    Blanc-Durand P, Van Der Gucht A, Schaefer N, Itti E, Prior JO. Automatic lesion detection and segmentation of 18F-FET PET in gliomas: a full 3D U-Net convolutional neural network study. PLoS One. 2018;13(4):e0195798.CrossRefPubMedPubMedCentralGoogle Scholar
  77. 77.
    Soltaninejad M, Yang G, Lambrou T, Allinson N, Jones TL, Barrick TR, et al. Supervised learning based multimodal MRI brain tumour segmentation using texture features from supervoxels. Comput Methods Progr Biomed. 2018;157:69–84.CrossRefGoogle Scholar
  78. 78.
    Zhao Z, Yang G, Lin Y, Pang H, Wang M. Automated glioma detection and segmentation using graphical models. PLoS One. 2018;13(8):e0200745.CrossRefPubMedPubMedCentralGoogle Scholar
  79. 79.
    Inano R, Oishi N, Kunieda T, Arakawa Y, Yamao Y, Shibata S, et al. Voxel-based clustered imaging by multiparameter diffusion tensor images for glioma grading. Neuroimage Clin. 2014;5:396–407.CrossRefPubMedPubMedCentralGoogle Scholar
  80. 80.
    Fetit AE, Novak J, Peet AC, Arvanitits TN. Three-dimensional textural features of conventional MRI improve diagnostic classification of childhood brain tumours. NMR Biomed. 2015;28(9):1174–84.CrossRefPubMedGoogle Scholar
  81. 81.
    Hu LS, Ning S, Eschbacher JM, Gaw N, Dueck AC, Smith KA, et al. Multi-Parametric MRI and texture analysis to visualize spatial histologic heterogeneity and tumor extent in glioblastoma. PLoS One. 2015;10(11):e0141506.CrossRefPubMedPubMedCentralGoogle Scholar
  82. 82.
    Tsolaki E, Svolos P, Kousi E, Kapsalaki E, Fezoulidis I, Fountas K, et al. Fast spectroscopic multiple analysis (FASMA) for brain tumor classification: a clinical decision support system utilizing multi-parametric 3T MR data. Int J Comput Assist Radiol Surg. 2015;10(7):1149–66.CrossRefPubMedGoogle Scholar
  83. 83.
    Li-Chun Hsieh K, Chen CY, Lo CM. Quantitative glioma grading using transformed gray-scale invariant textures of MRI. Comput Biol Med. 2017;83:102–8.CrossRefPubMedGoogle Scholar
  84. 84.
    Zhang X, Yan LF, Hu YC, Li G, Yang Y, Han Y, et al. Optimizing a machine learning based glioma grading system using multi-parametric MRI histogram and texture features. Oncotarget. 2017;8(29):47816–30.PubMedPubMedCentralGoogle Scholar
  85. 85.
    De Looze C, Beausang A, Cryan J, Loftus T, Buckley PG, Farrell M, et al. Machine learning: a useful radiological adjunct in determination of a newly diagnosed glioma’s grade and IDH status. J Neurooncol. 2018;139(2):491–9.CrossRefPubMedGoogle Scholar
  86. 86.
    Fetit AE, Novak J, Rodriguez D, Auer DP, Clark CA, Grundy RG, et al. Radiomics in paediatric neuro-oncology: a multicentre study on MRI texture analysis. NMR Biomed. 2018.  https://doi.org/10.1002/nbm.3781.CrossRefPubMedGoogle Scholar
  87. 87.
    Shofty B, Artzi M, Ben Bashat D, Liberman G, Haim O, Kashanian A, et al. MRI radiomics analysis of molecular alterations in low-grade gliomas. Int J Comput Assist Radiol Surg. 2018;13(4):563–71.CrossRefPubMedGoogle Scholar
  88. 88.
    Zarinabad N, Abernethy LJ, Avula S, Davies NP, Rodriguez Gutierrez D, Jaspan T, et al. Application of pattern recognition techniques for classification of pediatric brain tumors by in vivo 3T (1) H-MR spectroscopy-A multi-center study. Magn Reson Med. 2018;79(4):2359–66.CrossRefPubMedGoogle Scholar
  89. 89.
    Qian X, Tan H, Zhang J, Zhao W, Chan MD, Zhou X. Stratification of pseudoprogression and true progression of glioblastoma multiform based on longitudinal diffusion tensor imaging without segmentation. Med Phys. 2016;43(11):5889.CrossRefPubMedPubMedCentralGoogle Scholar
  90. 90.
    Zhang J, Yu H, Qian X, Liu K, Tan H, Yang T, et al. Pseudo progression identification of glioblastoma with dictionary learning. Comput Biol Med. 2016;73:94–101.CrossRefPubMedPubMedCentralGoogle Scholar
  91. 91.
    Emblem KE, Due-Tonnessen P, Hald JK, Bjornerud A, Pinho MC, Scheie D, et al. Machine learning in preoperative glioma MRI: survival associations by perfusion-based support vector machine outperforms traditional MRI. J Magn Reson Imaging. 2014;40(1):47–54.CrossRefPubMedGoogle Scholar
  92. 92.
    Emblem KE, Pinho MC, Zollner FG, Due-Tonnessen P, Hald JK, Schad LR, et al. A generic support vector machine model for preoperative glioma survival associations. Radiology. 2015;275(1):228–34.CrossRefPubMedGoogle Scholar
  93. 93.
    Akbari H, Macyszyn L, Da X, Bilello M, Wolf RL, Martinez-Lage M, et al. Imaging surrogates of infiltration obtained via multiparametric imaging pattern analysis predict subsequent location of recurrence of glioblastoma. Neurosurgery. 2016;78(4):572–80.CrossRefPubMedPubMedCentralGoogle Scholar
  94. 94.
    Chang K, Zhang B, Guo X, Zong M, Rahman R, Sanchez D, et al. Multimodal imaging patterns predict survival in recurrent glioblastoma patients treated with bevacizumab. Neuro Oncol. 2016;18(12):1680–7.CrossRefPubMedPubMedCentralGoogle Scholar
  95. 95.
    Korfiatis P, Kline TL, Coufalova L, Lachance DH, Parney IF, Carter RE, et al. MRI texture features as biomarkers to predict MGMT methylation status in glioblastomas. Med Phys. 2016;43(6):2835–44.CrossRefPubMedPubMedCentralGoogle Scholar
  96. 96.
    Macyszyn L, Akbari H, Pisapia JM, Da X, Attiah M, Pigrish V, et al. Imaging patterns predict patient survival and molecular subtype in glioblastoma via machine learning techniques. Neuro Oncol. 2016;18(3):417–25.CrossRefPubMedGoogle Scholar
  97. 97.
    Akkus Z, Ali I, Sedlar J, Agrawal JP, Parney IF, Giannini C, et al. Predicting deletion of chromosomal arms 1p/19q in low-grade gliomas from MR images using machine intelligence. J Digit Imaging. 2017;30(4):469–76.CrossRefPubMedPubMedCentralGoogle Scholar
  98. 98.
    Chen L, Zhang H, Thung KH, Liu L, Lu J, Wu J, et al. Multi-label inductive matrix completion for joint MGMT and IDH1 status prediction for glioma patients. Med Image Comput Comput Assist Interv. 2017;10434:450–8.PubMedPubMedCentralGoogle Scholar
  99. 99.
    Zhang B, Chang K, Ramkissoon S, Tanguturi S, Bi WL, Reardon DA, et al. Multimodal MRI features predict isocitrate dehydrogenase genotype in high-grade gliomas. Neuro Oncol. 2017;19(1):109–17.CrossRefPubMedGoogle Scholar
  100. 100.
    Zhou M, Chaudhury B, Hall LO, Goldgof DB, Gillies RJ, Gatenby RA. Identifying spatial imaging biomarkers of glioblastoma multiforme for survival group prediction. J Magn Reson Imaging. 2017;46(1):115–23.CrossRefPubMedGoogle Scholar
  101. 101.
    Li Y, Qian Z, Xu K, Wang K, Fan X, Li S, et al. MRI features predict p53 status in lower-grade gliomas via a machine-learning approach. Neuroimage Clin. 2018;17:306–11.CrossRefPubMedGoogle Scholar
  102. 102.
    Papp L, Potsch N, Grahovac M, Schmidbauer V, Woehrer A, Preusser M, et al. Glioma survival prediction with combined analysis of in vivo (11)C-MET PET features, ex vivo features, and patient features by supervised machine learning. J Nucl Med. 2018;59(6):892–9.CrossRefPubMedGoogle Scholar
  103. 103.
    Anticevic A, Cole MW, Repovs G, Murray JD, Brumbaugh MS, Winkler AM, et al. Characterizing thalamo-cortical disturbances in schizophrenia and bipolar illness. Cereb Cortex. 2014;24(12):3116–30.CrossRefPubMedGoogle Scholar
  104. 104.
    Bleich-Cohen M, Jamshy S, Sharon H, Weizman R, Intrator N, Poyurovsky M, et al. Machine learning fMRI classifier delineates subgroups of schizophrenia patients. Schizophr Res. 2014;160(1–3):196–200.CrossRefPubMedGoogle Scholar
  105. 105.
    Castro E, Gupta CN, Martinez-Ramon M, Calhoun VD, Arbabshirani MR, Turner J. Identification of patterns of gray matter abnormalities in schizophrenia using source-based morphometry and bagging. Conf Proc IEEE Eng Med Biol Soc. 2014;2014:1513–6.PubMedPubMedCentralGoogle Scholar
  106. 106.
    Schnack HG, Nieuwenhuis M, van Haren NE, Abramovic L, Scheewe TW, Brouwer RM, et al. Can structural MRI aid in clinical classification? A machine learning study in two independent samples of patients with schizophrenia, bipolar disorder and healthy subjects. Neuroimage. 2014;84:299–306.CrossRefPubMedGoogle Scholar
  107. 107.
    Cheng H, Newman S, Goni J, Kent JS, Howell J, Bolbecker A, et al. Nodal centrality of functional network in the differentiation of schizophrenia. Schizophr Res. 2015;168(1–2):345–52.CrossRefPubMedPubMedCentralGoogle Scholar
  108. 108.
    Chyzhyk D, Grana M, Ongur D, Shinn AK. Discrimination of schizophrenia auditory hallucinators by machine learning of resting-state functional MRI. Int J Neural Syst. 2015;25(3):1550007.CrossRefPubMedPubMedCentralGoogle Scholar
  109. 109.
    Chyzhyk D, Savio A, Grana M. Computer aided diagnosis of schizophrenia on resting state fMRI data by ensembles of ELM. Neural Netw. 2015;68:23–33.CrossRefPubMedGoogle Scholar
  110. 110.
    Koch SP, Hagele C, Haynes JD, Heinz A, Schlagenhauf F, Sterzer P. Diagnostic classification of schizophrenia patients on the basis of regional reward-related FMRI signal patterns. PLoS One. 2015;10(3):e0119089.CrossRefPubMedPubMedCentralGoogle Scholar
  111. 111.
    Chu WL, Huang MW, Jian BL, Hsu CY, Cheng KS. A correlative classification study of schizophrenic patients with results of clinical evaluation and structural magnetic resonance images. Behav Neurol. 2016;2016:7849526.CrossRefPubMedPubMedCentralGoogle Scholar
  112. 112.
    Kim J, Calhoun VD, Shim E, Lee JH. Deep neural network with weight sparsity control and pre-training extracts hierarchical features and enhances classification performance: evidence from whole-brain resting-state functional connectivity patterns of schizophrenia. Neuroimage. 2016;124(Pt A):127–46.CrossRefPubMedGoogle Scholar
  113. 113.
    Lu X, Yang Y, Wu F, Gao M, Xu Y, Zhang Y, et al. Discriminative analysis of schizophrenia using support vector machine and recursive feature elimination on structural MRI images. Medicine (Baltimore). 2016;95(30):e3973.CrossRefPubMedPubMedCentralGoogle Scholar
  114. 114.
    Pinaya WH, Gadelha A, Doyle OM, Noto C, Zugman A, Cordeiro Q, et al. Using deep belief network modelling to characterize differences in brain morphometry in schizophrenia. Sci Rep. 2016;6:38897.CrossRefPubMedPubMedCentralGoogle Scholar
  115. 115.
    Pergola G, Trizio S, Di Carlo P, Taurisano P, Mancini M, Amoroso N, et al. Grey matter volume patterns in thalamic nuclei are associated with familial risk for schizophrenia. Schizophr Res. 2017;180:13–20.CrossRefPubMedGoogle Scholar
  116. 116.
    Plaschke RN, Cieslik EC, Muller VI, Hoffstaedter F, Plachti A, Varikuti DP, et al. On the integrity of functional brain networks in schizophrenia, Parkinson’s disease, and advanced age: evidence from connectivity-based single-subject classification. Hum Brain Mapp. 2017;38(12):5845–58.CrossRefPubMedPubMedCentralGoogle Scholar
  117. 117.
    Qureshi MNI, Oh J, Cho D, Jo HJ, Lee B. Multimodal discrimination of schizophrenia using hybrid weighted feature concatenation of brain functional connectivity and anatomical features with an extreme learning machine. Front Neuroinform. 2017;11:59.CrossRefPubMedPubMedCentralGoogle Scholar
  118. 118.
    Skatun KC, Kaufmann T, Doan NT, Alnaes D, Cordova-Palomera A, Jonsson EG, et al. Consistent functional connectivity alterations in schizophrenia spectrum disorder: a multisite study. Schizophr Bull. 2017;43(4):914–24.CrossRefPubMedGoogle Scholar
  119. 119.
    Zarogianni E, Storkey AJ, Johnstone EC, Owens DG, Lawrie SM. Improved individualized prediction of schizophrenia in subjects at familial high risk, based on neuroanatomical data, schizotypal and neurocognitive features. Schizophr Res. 2017;181:6–12.CrossRefPubMedGoogle Scholar
  120. 120.
    Juneja A, Rana B, Agrawal RK. A novel fuzzy rough selection of non-linearly extracted features for schizophrenia diagnosis using fMRI. Comput Methods Progr Biomed. 2018;155:139–52.CrossRefGoogle Scholar
  121. 121.
    Mikolas P, Hlinka J, Skoch A, Pitra Z, Frodl T, Spaniel F, et al. Machine learning classification of first-episode schizophrenia spectrum disorders and controls using whole brain white matter fractional anisotropy. BMC Psychiatry. 2018;18(1):97.CrossRefPubMedPubMedCentralGoogle Scholar
  122. 122.
    Orban P, Dansereau C, Desbois L, Mongeau-Perusse V, Giguere CE, Nguyen H, et al. Multisite generalizability of schizophrenia diagnosis classification based on functional brain connectivity. Schizophr Res. 2018;192:167–71.CrossRefPubMedGoogle Scholar
  123. 123.
    Zeng LL, Wang H, Hu P, Yang B, Pu W, Shen H, et al. Multi-site diagnostic classification of schizophrenia using discriminant deep learning with functional connectivity MRI. EBioMedicine. 2018;30:74–85.CrossRefPubMedPubMedCentralGoogle Scholar
  124. 124.
    Papakostas GI. Managing partial response or nonresponse: switching, augmentation, and combination strategies for major depressive disorder. J Clin Psychol. 2009;70(Suppl. 6):16–25.Google Scholar
  125. 125.
    Cao L, Guo S, Xue Z, Hu Y, Liu H, Mwansisya TE, et al. Aberrant functional connectivity for diagnosis of major depressive disorder: a discriminant analysis. Psychiatry Clin Neurosci. 2014;68(2):110–9.CrossRefPubMedGoogle Scholar
  126. 126.
    Qin J, Wei M, Liu H, Chen J, Yan R, Hua L, et al. Abnormal hubs of white matter networks in the frontal-parieto circuit contribute to depression discrimination via pattern classification. Magn Reson Imaging. 2014;32(10):1314–20.CrossRefPubMedGoogle Scholar
  127. 127.
    Zeng LL, Shen H, Liu L, Hu D. Unsupervised classification of major depression using functional connectivity MRI. Hum Brain Mapp. 2014;35(4):1630–41.CrossRefPubMedGoogle Scholar
  128. 128.
    Lueken U, Straube B, Yang Y, Hahn T, Beesdo-Baum K, Wittchen HU, et al. Separating depressive comorbidity from panic disorder: a combined functional magnetic resonance imaging and machine learning approach. J Affect Disord. 2015;184:182–92.CrossRefPubMedGoogle Scholar
  129. 129.
    Qin J, Wei M, Liu H, Chen J, Yan R, Yao Z, et al. Altered anatomical patterns of depression in relation to antidepressant treatment: evidence from a pattern recognition analysis on the topological organization of brain networks. J Affect Disord. 2015;180:129–37.CrossRefPubMedGoogle Scholar
  130. 130.
    Sato JR, Moll J, Green S, Deakin JF, Thomaz CE, Zahn R. Machine learning algorithm accurately detects fMRI signature of vulnerability to major depression. Psychiatry Res. 2015;233(2):289–91.CrossRefPubMedPubMedCentralGoogle Scholar
  131. 131.
    Shimizu Y, Yoshimoto J, Toki S, Takamura M, Yoshimura S, Okamoto Y, et al. Toward probabilistic diagnosis and understanding of depression based on functional MRI data analysis with logistic group LASSO. PLoS One. 2015;10(5):e0123524.CrossRefPubMedPubMedCentralGoogle Scholar
  132. 132.
    Wu MJ, Wu HE, Mwangi B, Sanches M, Selvaraj S, Zunta-Soares GB, et al. Prediction of pediatric unipolar depression using multiple neuromorphometric measurements: a pattern classification approach. J Psychiatr Res. 2015;62:84–91.CrossRefPubMedPubMedCentralGoogle Scholar
  133. 133.
    Ramasubbu R, Brown MR, Cortese F, Gaxiola I, Goodyear B, Greenshaw AJ, et al. Accuracy of automated classification of major depressive disorder as a function of symptom severity. Neuroimage Clin. 2016;12:320–31.CrossRefPubMedPubMedCentralGoogle Scholar
  134. 134.
    Bhaumik R, Jenkins LM, Gowins JR, Jacobs RH, Barba A, Bhaumik DK, et al. Multivariate pattern analysis strategies in detection of remitted major depressive disorder using resting state functional connectivity. Neuroimage Clin. 2017;16:390–8.CrossRefPubMedGoogle Scholar
  135. 135.
    Kautzky A, James GM, Philippe C, Baldinger-Melich P, Kraus C, Kranz GS, et al. The influence of the rs6295 gene polymorphism on serotonin-1A receptor distribution investigated with PET in patients with major depression applying machine learning. Transl Psychiatry. 2017;7(6):e1150.CrossRefPubMedPubMedCentralGoogle Scholar
  136. 136.
    Schnyer DM, Clasen PC, Gonzalez C, Beevers CG. Evaluating the diagnostic utility of applying a machine learning algorithm to diffusion tensor MRI measures in individuals with major depressive disorder. Psychiatry Res Neuroimaging. 2017;264:1–9.CrossRefPubMedGoogle Scholar
  137. 137.
    Wang J, Wei Q, Bai T, Zhou X, Sun H, Becker B, et al. Electroconvulsive therapy selectively enhanced feedforward connectivity from fusiform face area to amygdala in major depressive disorder. Soc Cogn Affect Neurosci. 2017;12(12):1983–92.CrossRefPubMedPubMedCentralGoogle Scholar
  138. 138.
    Deng F, Wang Y, Huang H, Niu M, Zhong S, Zhao L, et al. Abnormal segments of right uncinate fasciculus and left anterior thalamic radiation in major and bipolar depression. Prog Neuropsychopharmacol Biol Psychiatry. 2018;81:340–9.CrossRefPubMedGoogle Scholar
  139. 139.
    Milak MS, Pantazatos S, Rashid R, Zanderigo F, DeLorenzo C, Hesselgrave N, et al. Higher 5-HT1A autoreceptor binding as an endophenotype for major depressive disorder identified in high risk offspring—a pilot study. Psychiatry Res Neuroimaging. 2018;276:15–23.CrossRefPubMedPubMedCentralGoogle Scholar
  140. 140.
    Foland-Ross LC, Sacchet MD, Prasad G, Gilbert B, Thompson PM, Gotlib IH. Cortical thickness predicts the first onset of major depression in adolescence. Int J Dev Neurosci. 2015;46:125–31.CrossRefPubMedPubMedCentralGoogle Scholar
  141. 141.
    Johnston BA, Steele JD, Tolomeo S, Christmas D, Matthews K. Structural MRI-based predictions in patients with treatment-refractory depression (TRD). PLoS One. 2015;10(7):e0132958.CrossRefPubMedPubMedCentralGoogle Scholar
  142. 142.
    Patel MJ, Andreescu C, Price JC, Edelman KL, Reynolds CF 3rd, Aizenstein HJ. Machine learning approaches for integrating clinical and imaging features in late-life depression classification and response prediction. Int J Geriatr Psychiatry. 2015;30(10):1056–67.CrossRefPubMedPubMedCentralGoogle Scholar
  143. 143.
    Redlich R, Opel N, Grotegerd D, Dohm K, Zaremba D, Burger C, et al. Prediction of individual response to electroconvulsive therapy via machine learning on structural magnetic resonance imaging data. JAMA Psychiatry. 2016;73(6):557–64.CrossRefPubMedGoogle Scholar
  144. 144.
    Yoshida K, Shimizu Y, Yoshimoto J, Takamura M, Okada G, Okamoto Y, et al. Prediction of clinical depression scores and detection of changes in whole-brain using resting-state functional MRI data with partial least squares regression. PLoS One. 2017;12(7):e0179638.CrossRefPubMedPubMedCentralGoogle Scholar
  145. 145.
    Adeli E, Shi F, An L, Wee CY, Wu G, Wang T, et al. Joint feature-sample selection and robust diagnosis of Parkinson’s disease from MRI data. Neuroimage. 2016;141:206–19.CrossRefPubMedPubMedCentralGoogle Scholar
  146. 146.
    Choi H, Ha S, Im HJ, Paek SH, Lee DS. Refining diagnosis of Parkinson’s disease with deep learning-based interpretation of dopamine transporter imaging. Neuroimage Clin. 2017;16:586–94.CrossRefPubMedPubMedCentralGoogle Scholar
  147. 147.
    Peng B, Wang S, Zhou Z, Liu Y, Tong B, Zhang T, et al. A multilevel-ROI-features-based machine learning method for detection of morphometric biomarkers in Parkinson’s disease. Neurosci Lett. 2017;651:88–94.CrossRefPubMedGoogle Scholar
  148. 148.
    Amoroso N, La Rocca M, Monaco A, Bellotti R, Tangaro S. Complex networks reveal early MRI markers of Parkinson’s disease. Med Image Anal. 2018;48:12–24.CrossRefPubMedGoogle Scholar
  149. 149.
    Castillo-Barnes D, Ramirez J, Segovia F, Martinez-Murcia FJ, Salas-Gonzalez D, Gorriz JM. Robust ensemble classification methodology for I123-ioflupane SPECT images and multiple heterogeneous biomarkers in the diagnosis of Parkinson’s disease. Front Neuroinform. 2018;12:53.CrossRefPubMedPubMedCentralGoogle Scholar
  150. 150.
    Oliveira FPM, Faria DB, Costa DC, Castelo-Branco M, Tavares J. Extraction, selection and comparison of features for an effective automated computer-aided diagnosis of Parkinson’s disease based on [(123)I]FP-CIT SPECT images. Eur J Nucl Med Mol Imaging. 2018;45(6):1052–62.CrossRefPubMedGoogle Scholar
  151. 151.
    Salvatore C, Cerasa A, Castiglioni I, Gallivanone F, Augimeri A, Lopez M, et al. Machine learning on brain MRI data for differential diagnosis of Parkinson’s disease and progressive supranuclear palsy. J Neurosci Methods. 2014;222:230–7.CrossRefPubMedGoogle Scholar
  152. 152.
    Hirschauer TJ, Adeli H, Buford JA. Computer-Aided diagnosis of Parkinson’s disease using enhanced probabilistic neural network. J Med Syst. 2015;39(11):179.CrossRefPubMedGoogle Scholar
  153. 153.
    Huertas-Fernandez I, Garcia-Gomez FJ, Garcia-Solis D, Benitez-Rivero S, Marin-Oyaga VA, Jesus S, et al. Machine learning models for the differential diagnosis of vascular parkinsonism and Parkinson’s disease using [(123)I]FP-CIT SPECT. Eur J Nucl Med Mol Imaging. 2015;42(1):112–9.CrossRefPubMedGoogle Scholar
  154. 154.
    Singh G, Samavedham L. Unsupervised learning based feature extraction for differential diagnosis of neurodegenerative diseases: a case study on early-stage diagnosis of Parkinson disease. J Neurosci Methods. 2015;256:30–40.CrossRefPubMedGoogle Scholar
  155. 155.
    Liu L, Wang Q, Adeli E, Zhang L, Zhang H, Shen D. Feature selection based on iterative canonical correlation analysis for automatic diagnosis of Parkinson’s disease. Med Image Comput Comput Assist Interv. 2016;9901:1–8.PubMedPubMedCentralGoogle Scholar
  156. 156.
    Du G, Lewis MM, Kanekar S, Sterling NW, He L, Kong L, et al. Combined diffusion tensor imaging and apparent transverse relaxation rate differentiate Parkinson disease and atypical parkinsonism. AJNR Am J Neuroradiol. 2017;38(5):966–72.CrossRefPubMedPubMedCentralGoogle Scholar
  157. 157.
    Ye Z, Rae CL, Nombela C, Ham T, Rittman T, Jones PS, et al. Predicting beneficial effects of atomoxetine and citalopram on response inhibition in Parkinson’s disease with clinical and neuroimaging measures. Hum Brain Mapp. 2016;37(3):1026–37.CrossRefPubMedPubMedCentralGoogle Scholar
  158. 158.
    ADHD200 - international neuroimaging data-sharing initiative - NITRC. http://.fcon_1000.projects.nitrc.org/indi/adhd200. Accessed Oct 2018.Google Scholar
  159. 159.
    Qureshi MN, Min B, Jo HJ, Lee B. Multiclass classification for the differential diagnosis on the ADHD subtypes using recursive feature elimination and hierarchical extreme learning machine: structural MRI study. PLoS One. 2016;11(8):e0160697.CrossRefPubMedPubMedCentralGoogle Scholar
  160. 160.
    Qureshi MNI, Oh J, Min B, Jo HJ, Lee B. Multi-modal, multi-measure, and multi-class discrimination of ADHD with hierarchical feature extraction and extreme learning machine using structural and functional brain MRI. Front Hum Neurosci. 2017;11:157.PubMedPubMedCentralGoogle Scholar
  161. 161.
    Tan L, Guo X, Ren S, Epstein JN, Lu LJ. A computational model for the automatic diagnosis of attention deficit hyperactivity disorder based on functional brain volume. Front Comput Neurosci. 2017;11:75.CrossRefPubMedPubMedCentralGoogle Scholar
  162. 162.
    Hart H, Chantiluke K, Cubillo AI, Smith AB, Simmons A, Brammer MJ, et al. Pattern classification of response inhibition in ADHD: toward the development of neurobiological markers for ADHD. Hum Brain Mapp. 2014;35(7):3083–94.CrossRefPubMedGoogle Scholar
  163. 163.
    Johnston BA, Mwangi B, Matthews K, Coghill D, Konrad K, Steele JD. Brainstem abnormalities in attention deficit hyperactivity disorder support high accuracy individual diagnostic classification. Hum Brain Mapp. 2014;35(10):5179–89.CrossRefPubMedGoogle Scholar
  164. 164.
    Ghiassian S, Greiner R, Jin P, Brown MR. Using functional or structural magnetic resonance images and personal characteristic data to identify ADHD and autism. PLoS One. 2016;11(12):e0166934.CrossRefPubMedPubMedCentralGoogle Scholar
  165. 165.
    Xiao C, Bledsoe J, Wang S, Chaovalitwongse WA, Mehta S, Semrud-Clikeman M, et al. An integrated feature ranking and selection framework for ADHD characterization. Brain Inform. 2016;3(3):145–55.CrossRefPubMedPubMedCentralGoogle Scholar
  166. 166.
    Chaim-Avancini TM, Doshi J, Zanetti MV, Erus G, Silva MA, Duran FLS, et al. Neurobiological support to the diagnosis of ADHD in stimulant-naive adults: pattern recognition analyses of MRI data. Acta Psychiatr Scand. 2017;136(6):623–36.CrossRefPubMedGoogle Scholar
  167. 167.
    Riaz A, Asad M, Alonso E, Slabaugh G. Fusion of fMRI and non-imaging data for ADHD classification. Comput Med Imaging Graph. 2018;65:115–28.CrossRefPubMedGoogle Scholar
  168. 168.
    Sen B, Borle NC, Greiner R, Brown MRG. A general prediction model for the detection of ADHD and Autism using structural and functional MRI. PLoS One. 2018;13(4):e0194856.CrossRefPubMedPubMedCentralGoogle Scholar
  169. 169.
    Kim JW, Sharma V, Ryan ND. Predicting methylphenidate response in ADHD using machine learning approaches. Int J Neuropsychopharmacol. 2015;18(11):pyv052.CrossRefPubMedPubMedCentralGoogle Scholar
  170. 170.
    Matson JL, Rieske RD, Williams LW. The relationship between autism spectrum disorders and attention-deficit/hyperactivity disorder: an overview. Res Dev Disabil. 2013;34:2475–84.CrossRefPubMedGoogle Scholar
  171. 171.
    Lord C, Risi S, Lambrecht L, Cook EH Jr, Leventhal BL, DiLavore PC, et al. The autism diagnostic observation schedule-generic: a standard measure of social and communication deficits associated with the spectrum of autism. J Autism Dev Disord. 2000;30:205–23.CrossRefPubMedGoogle Scholar
  172. 172.
    Uddin LQ, Dajani DR, Voorhies W, Bednarz H, Kana RK. Progress and roadblocks in the search for brain-based biomarkers of autism and attention-deficit/hyperactivity disorder. Transl Psychiatry. 2017;7:e1218.CrossRefPubMedPubMedCentralGoogle Scholar
  173. 173.
    Just MA, Cherkassky VL, Buchweitz A, Keller TA, Mitchell TM. Identifying autism from neural representations of social interactions: neurocognitive markers of autism. PLoS One. 2014;9(12):e113879.CrossRefPubMedPubMedCentralGoogle Scholar
  174. 174.
    Chen CP, Keown CL, Jahedi A, Nair A, Pflieger ME, Bailey BA, et al. Diagnostic classification of intrinsic functional connectivity highlights somatosensory, default mode, and visual regions in autism. Neuroimage Clin. 2015;8:238–45.CrossRefPubMedPubMedCentralGoogle Scholar
  175. 175.
    Gori I, Giuliano A, Muratori F, Saviozzi I, Oliva P, Tancredi R, et al. Gray matter alterations in young children with autism spectrum disorders: comparing morphometry at the voxel and regional level. J Neuroimaging. 2015;25(6):866–74.CrossRefPubMedGoogle Scholar
  176. 176.
    Iidaka T. Resting state functional magnetic resonance imaging and neural network classified autism and control. Cortex. 2015;63:55–67.CrossRefPubMedGoogle Scholar
  177. 177.
    Plitt M, Barnes KA, Martin A. Functional connectivity classification of autism identifies highly predictive brain features but falls short of biomarker standards. Neuroimage Clin. 2015;7:359–66.CrossRefPubMedGoogle Scholar
  178. 178.
    Jahedi A, Nasamran CA, Faires B, Fan J, Muller RA. Distributed intrinsic functional connectivity patterns predict diagnostic status in large autism cohort. Brain Connect. 2017;7(8):515–25.CrossRefPubMedPubMedCentralGoogle Scholar
  179. 179.
    Kam TE, Suk HI, Lee SW. Multiple functional networks modeling for autism spectrum disorder diagnosis. Hum Brain Mapp. 2017;38(11):5804–21.CrossRefPubMedGoogle Scholar
  180. 180.
    Akhavan Aghdam M, Sharifi A, Pedram MM. Combination of rs-fMRI and sMRI data to discriminate autism spectrum disorders in young children using deep belief network. J Digit Imaging. 2018;31(6):895–903.CrossRefPubMedGoogle Scholar
  181. 181.
    Bi XA, Wang Y, Shu Q, Sun Q, Xu Q. Classification of autism spectrum disorder using random support vector machine cluster. Front Genet. 2018;9:18.CrossRefPubMedPubMedCentralGoogle Scholar
  182. 182.
    Heinsfeld AS, Franco AR, Craddock RC, Buchweitz A, Meneguzzi F. Identification of autism spectrum disorder using deep learning and the ABIDE dataset. Neuroimage Clin. 2018;17:16–23.CrossRefPubMedGoogle Scholar
  183. 183.
    Li H, Parikh NA, He L. A novel transfer learning approach to enhance deep neural network classification of brain functional connectomes. Front Neurosci. 2018;12:491.CrossRefPubMedPubMedCentralGoogle Scholar
  184. 184.
    Moradi E, Khundrakpam B, Lewis JD, Evans AC, Tohka J. Predicting symptom severity in autism spectrum disorder based on cortical thickness measures in agglomerative data. Neuroimage. 2017;144(Pt A):128–41.CrossRefPubMedGoogle Scholar
  185. 185.
    Rudie JD, Colby JB, Salamon N. Machine learning classification of mesial temporal sclerosis in epilepsy patients. Epilepsy Res. 2015;117:63–9.CrossRefPubMedGoogle Scholar
  186. 186.
    Hong SJ, Kim H, Schrader D, Bernasconi N, Bernhardt BC, Bernasconi A. Automated detection of cortical dysplasia type II in MRI-negative epilepsy. Neurology. 2014;83(1):48–55.CrossRefPubMedPubMedCentralGoogle Scholar
  187. 187.
    Kamiya K, Amemiya S, Suzuki Y, Kunii N, Kawai K, Mori H, et al. Machine learning of DTI structural brain connectomes for lateralization of temporal lobe epilepsy. Magn Reson Med Sci. 2016;15(1):121–9.CrossRefPubMedGoogle Scholar
  188. 188.
    Del Gaizo J, Mofrad N, Jensen JH, Clark D, Glenn R, Helpern J, et al. Using machine learning to classify temporal lobe epilepsy based on diffusion MRI. Brain Behav. 2017;7(10):e00801.CrossRefPubMedPubMedCentralGoogle Scholar
  189. 189.
    Hong SJ, Bernhardt BC, Schrader DS, Bernasconi N, Bernasconi A. Whole-brain MRI phenotyping in dysplasia-related frontal lobe epilepsy. Neurology. 2016;86(7):643–50.CrossRefPubMedPubMedCentralGoogle Scholar
  190. 190.
    El Azami M, Hammers A, Jung J, Costes N, Bouet R, Lartizien C. Detection of lesions underlying intractable epilepsy on T1-weighted MRI as an outlier detection problem. PLoS One. 2016;11(9):e0161498.CrossRefPubMedPubMedCentralGoogle Scholar
  191. 191.
    Adler S, Wagstyl K, Gunny R, Ronan L, Carmichael D, Cross JH, et al. Novel surface features for automated detection of focal cortical dysplasias in paediatric epilepsy. Neuroimage Clin. 2017;14:18–27.CrossRefPubMedGoogle Scholar
  192. 192.
    Bernhardt BC, Hong SJ, Bernasconi A, Bernasconi N. Magnetic resonance imaging pattern learning in temporal lobe epilepsy: classification and prognostics. Ann Neurol. 2015;77(3):436–46.CrossRefPubMedGoogle Scholar
  193. 193.
    Munsell BC, Wee CY, Keller SS, Weber B, Elger C, da Silva LA, et al. Evaluation of machine learning algorithms for treatment outcome prediction in patients with epilepsy based on structural connectome data. Neuroimage. 2015;118:219–30.CrossRefPubMedPubMedCentralGoogle Scholar
  194. 194.
    Pustina D, Avants B, Sperling M, Gorniak R, He X, Doucet G, et al. Predicting the laterality of temporal lobe epilepsy from PET, MRI, and DTI: a multimodal study. Neuroimage Clin. 2015;9:20–31.CrossRefPubMedPubMedCentralGoogle Scholar
  195. 195.
    Zhong J, Chen DQ, Nantes JC, Holmes SA, Hodaie M, Koski L. Combined structural and functional patterns discriminating upper limb motor disability in multiple sclerosis using multivariate approaches. Brain Imaging Behav. 2017;11(3):754–68.CrossRefPubMedGoogle Scholar
  196. 196.
    Yoo Y, Tang LYW, Brosch T, Li DKB, Kolind S, Vavasour I, et al. Deep learning of joint myelin and T1w MRI features in normal-appearing brain tissue to distinguish between multiple sclerosis patients and healthy controls. Neuroimage Clin. 2018;17:169–78.CrossRefPubMedGoogle Scholar
  197. 197.
    Zurita M, Montalba C, Labbe T, Cruz JP, Dalboni da Rocha J, Tejos C, et al. Characterization of relapsing-remitting multiple sclerosis patients using support vector machine classifications of functional and diffusion MRI data. Neuroimage Clin. 2018;20:724–30.CrossRefPubMedPubMedCentralGoogle Scholar
  198. 198.
    Sacca V, Sarica A, Novellino F, Barone S, Tallarico T, Filippelli E, et al. Evaluation of machine learning algorithms performance for the prediction of early multiple sclerosis from resting-state FMRI connectivity data. Brain Imaging Behav. 2018.  https://doi.org/10.1007/s11682-018-9926-9.CrossRefGoogle Scholar
  199. 199.
    Salem M, Cabezas M, Valverde S, Pareto D, Oliver A, Salvi J, et al. A supervised framework with intensity subtraction and deformation field features for the detection of new T2-w lesions in multiple sclerosis. Neuroimage Clin. 2018;17:607–15.CrossRefPubMedGoogle Scholar
  200. 200.
    Zhao Y, Healy BC, Rotstein D, Guttmann CR, Bakshi R, Weiner HL, et al. Exploration of machine learning techniques in predicting multiple sclerosis disease course. PLoS One. 2017;12(4):e0174866.CrossRefPubMedPubMedCentralGoogle Scholar
  201. 201.
    Deshpande H, Maurel P, Barillot C. Classification of multiple sclerosis lesions using adaptive dictionary learning. Comput Med Imaging Graph. 2015;46(Pt 1):2–10.CrossRefPubMedGoogle Scholar
  202. 202.
    Fartaria MJ, Bonnier G, Roche A, Kober T, Meuli R, Rotzinger D, et al. Automated detection of white matter and cortical lesions in early stages of multiple sclerosis. J Magn Reson Imaging. 2016;43(6):1445–54.CrossRefPubMedGoogle Scholar
  203. 203.
    Lee Eun-Jae, Kim Yong-Hwan, Kim Namkug, Kanga Dong-Wha. Deep into the brain: artificial intelligence in stroke imaging. J Stroke. 2017;19(3):277–85.CrossRefPubMedPubMedCentralGoogle Scholar
  204. 204.
    Ashton EA, Takahashi C, Berg MJ, Goodman A, Totterman S, Ekholm S. Accuracy and reproducibility of manual and semiautomated quantification of MS lesions by MRI. J Magn Reson Imaging. 2003;17:300–8.CrossRefPubMedGoogle Scholar
  205. 205.
    Bentley P, Ganesalingam J, Carlton Jones AL, Mahady K, Epton S, Rinne P, et al. Prediction of stroke thrombolysis outcome using CT brain machine learning. Neuroimage Clin. 2014;4:635640.CrossRefGoogle Scholar
  206. 206.
    Kim BJ, Kim YH, Kim N, Kwon SU, Kim SJ, Kim JS, et al. Lesion location-based prediction of visual field improvement after cerebral infarction. PLoS One. 2015;10:e0143882.CrossRefPubMedPubMedCentralGoogle Scholar
  207. 207.
    Guberina N, Dietrich U, Radbruch A, Goebel J, Deuschl C, Ringelstein A, et al. Detection of early infarction signs with machine learning-based diagnosis by means of the Alberta Stroke Program Early CT score (ASPECTS) in the clinical routine. Neuroradiology. 2018;60(9):889–901.CrossRefPubMedGoogle Scholar
  208. 208.
    Rehme AK, Volz LJ, Feis DL, Bomilcar-Focke I, Liebig T, Eickhoff SB, et al. Identifying neuroimaging markers of motor disability in acute stroke by machine learning techniques. Cereb Cortex. 2015;25(9):3046–56.CrossRefPubMedGoogle Scholar
  209. 209.
    Asadi H, Dowling R, Yan B, Mitchell P. Machine learning for outcome prediction of acute ischemic stroke post intra-arterial therapy. PLoS One. 2014;9(2):e88225.CrossRefPubMedPubMedCentralGoogle Scholar
  210. 210.
    Bentley P, Ganesalingam J, Carlton Jones AL, Mahady K, Epton S, Rinne P, et al. Prediction of stroke thrombolysis outcome using CT brain machine learning. Neuroimage Clin. 2014;4:635–40.CrossRefPubMedPubMedCentralGoogle Scholar
  211. 211.
    Maier O, Schroder C, Forkert ND, Martinetz T, Handels H. Classifiers for ischemic stroke lesion segmentation: a comparison study. PLoS One. 2015;10(12):e0145118.CrossRefPubMedPubMedCentralGoogle Scholar
  212. 212.
    Griffis JC, Allendorfer JB, Szaflarski JP. Voxel-based Gaussian naive Bayes classification of ischemic stroke lesions in individual T1-weighted MRI scans. J Neurosci Methods. 2016;257:97–108.CrossRefPubMedGoogle Scholar
  213. 213.
    Pustina D, Coslett HB, Turkeltaub PE, Tustison N, Schwartz MF, Avants B. Automated segmentation of chronic stroke lesions using LINDA: Lesion identification with neighborhood data analysis. Hum Brain Mapp. 2016;37(4):1405–21.CrossRefPubMedPubMedCentralGoogle Scholar
  214. 214.
    Douglas David B, Iv Michael, Douglas Pamela K, Ariana Anderson, Vos Sjoerd B, Bammer Roland, Zeineh Michael, Wintermark Max. Diffusion tensor imaging of TBI: potentials and challenges. Top Magn Reson Imaging. 2015;24(5):241–51.CrossRefPubMedPubMedCentralGoogle Scholar
  215. 215.
    Johansen-Berg H. Behavioural relevance of variation in white matter microstructure. Curr Opin Neurol. 2010;23(351–358):20581685.Google Scholar
  216. 216.
    Mitra J, Shen KK, Ghose S, Bourgeat P, Fripp J, Salvado O, et al. Statistical machine learning to identify traumatic brain injury (TBI) from structural disconnections of white matter networks. Neuroimage. 2016;129:247–59.CrossRefPubMedGoogle Scholar
  217. 217.
    Peacock WFT, Van Meter TE, Mirshahi N, Ferber K, Gerwien R, Rao V, et al. Derivation of a three biomarker panel to improve diagnosis in patients with mild traumatic brain injury. Front Neurol. 2017;8:641.CrossRefPubMedPubMedCentralGoogle Scholar
  218. 218.
    Rangaprakash D, Deshpande G, Daniel TA, Goodman AM, Robinson JL, Salibi N, et al. Compromised hippocampus-striatum pathway as a potential imaging biomarker of mild-traumatic brain injury and posttraumatic stress disorder. Hum Brain Mapp. 2017;38(6):2843–64.CrossRefPubMedGoogle Scholar
  219. 219.
    Vergara VM, Mayer AR, Damaraju E, Kiehl KA, Calhoun V. Detection of mild traumatic brain injury by machine learning classification using resting state functional network connectivity and fractional anisotropy. J Neurotrauma. 2017;34(5):1045–53.CrossRefPubMedPubMedCentralGoogle Scholar
  220. 220.
    Rangaprakash D, Dretsch MN, Venkataraman A, Katz JS, Denney TS Jr, Deshpande G. Identifying disease foci from static and dynamic effective connectivity networks: illustration in soldiers with trauma. Hum Brain Mapp. 2018;39(1):264–87.CrossRefPubMedGoogle Scholar
  221. 221.
    Fagerholm ED, Hellyer PJ, Scott G, Leech R, Sharp DJ. Disconnection of network hubs and cognitive impairment after traumatic brain injury. Brain. 2015;138(Pt 6):1696–709.CrossRefPubMedPubMedCentralGoogle Scholar
  222. 222.
    Chong SL, Liu N, Barbier S, Ong ME. Predictive modeling in pediatric traumatic brain injury using machine learning. BMC Med Res Methodol. 2015;15:22.CrossRefPubMedPubMedCentralGoogle Scholar
  223. 223.
    OASIS brains - open access series of imaging studies. https://www.oasis-brains.org/. Accessed Oct 2018.
  224. 224.
    DZNE – longitudinal cognitive impairment and dementia study. https://www.dzne.de/en/research/studies/studien/delcode/. Accessed Oct 2018.
  225. 225.
    Seiler S, Schmidt H, Lechner A, Benke T. Driving cessation and dementia: results of the prospective registry on dementia in Austria PRODEM. PLoS One. 2012;7(12):1–6.CrossRefGoogle Scholar
  226. 226.
    Ellis KA, Bush AI, Darby D, De Fazio D, Foster J, Hudson P, Lautenschlager NT, Lenzo N, Martins RN, Maruff P, et al. The Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging: methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer’s disease. Int Psychogeriatr. 2009;21:672–87.CrossRefPubMedGoogle Scholar
  227. 227.
    QIBA - quantitative imaging biomarkers alliance. https://www.linkedin.com/company/rsna-qiba. Accessed Oct 2018.
  228. 228.
    Shen Dinggang, Guorong Wu, Suk Heung-Il. Deep learning in medical image analysis. Annu Rev Biomed Eng. 2017;21(19):221–48.CrossRefGoogle Scholar
  229. 229.
    Vieira S, Pinaya WHL, Mechelli A. Using deep learning to investigate the neuroimaging correlates of psychiatric and neurological disorders: methods and applications. Neurosci Biobehav Rev. 2017;74:58–75.CrossRefPubMedGoogle Scholar

Copyright information

© Japan Radiological Society 2018

Authors and Affiliations

  1. 1.Department of Radiology, Graduate School of Medical ScienceKyoto Prefectural University of MedicineKyotoJapan

Personalised recommendations