Weighted Combination of Multi-Parametric MR Imaging Markers for Evaluating Radiation Therapy Related Changes in the Prostate

Abstract

Recently, multi-parametric (MP) Magnetic Resonance (MR) Imaging (T2-weighted, MR Spectroscopy (MRS), Diffusion-weighted (DWI)) has shown great potential for evaluating the early effects of radiotherapy (RT) in the prostate. In this work we present a framework for quantitatively combining MP-MRI markers in order to assess RT changes on a voxel-by-voxel basis. The suite of segmentation, registration, feature extraction, and classifier tools presented in this work will allow for identification of (a) residual disease, and (b) new foci of cancer (local recurrence) within the prostate. Our scheme involves, (a) simultaneously evaluating differences in pre-, post-RT MR imaging markers, and (b) intelligently integrating and weighting the imaging marker changes obtained in (a) to generate a combined MP-MRI difference map that can better quantify treatment specific changes in the prostate. We demonstrate the applicability of our scheme in studying intensity-modulated radiation therapy (IMRT)-related changes for a cohort of 14 MP (T2w, MRS, DWI) prostate MRI patient datasets. In the first step, the different MRI protocols from pre- and post-IMRT MRI scans are affinely registered (accounting for gland shrinkage), followed by automated segmentation of the prostate capsule using an active shape model. Individual imaging marker difference maps are generated by calculating the differences of textural, metabolic, and functional MRI marker attributes, pre- and post-RT, on a per-voxel basis. These difference maps are then combined via an intelligent optimization scheme to generate a combined weighted difference map, where higher difference values on the map signify larger change (new foci of cancer), and low difference values signify no/small change post-RT. In the absence of histological ground truth (surgical or biopsy), radiologist delineated CaP extent on pre-, and post-RT MRI was employed as the ground truth for evaluating the accuracy of our scheme in successfully identifying MP-MRI related disease changes post-RT. A mean area under the receiver operating curve (AUC) of 73.2% was obtained via the weighted MP-MRI map, when evaluated against expert delineated CaP extent on pre-, post-RT MRI. The difference maps corresponding to the individual structural (T2w intensities), functional (ADC intensities) and metabolic (choline, creatine) markers yielded a corresponding mean AUC of 54.4%, 68.6% and 70.8%.

This work was supported by the Wallace H. Coulter Foundation, the National Cancer Institute under Grants R01CA136535, R21CA127186, R03CA128081, and R03CA143991, the Cancer Institute of New Jersey, and the Department of Defense (W81XWH-09).