Manifold Learning-based Data Sampling for Model Training

  • Shuqing Chen
  • Sabrina Dorn
  • Michael Lell
  • Marc Kachelrieß
  • Andreas Maier
Conference paper
Part of the Informatik aktuell book series (INFORMAT)

Zusammenfassung

Training data sampling is an important task in machine learning especially for data with small sample size and data with nonuniform sample distribution. Dividing data into different data sets randomly can cause the problem that, the training model covers only parts of the sampled cases and works inaccurately for weakly sampled cases. Recent research showed the benefit of manifold learning techniques in medical image processing. In this work, we propose a manifold learning based approach to improve the data division and the model training. We evaluated the proposed approach using an atlas registration framework and a deep learning framework. The final segmentation results using methods with and without data balancing were compared. All of the final segmentations were improved after implementing the manifold learning based approach into the frameworks. The largest improvement was 24.4%. Thus, the proposed manifold learning based approach is effective for the model training.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Literatur

  1. 1.
    Aljabar P, Wolz R, Rueckert D. Manifold learning for medical image registration, segmentation, and classification. Mach Learn Comput Aid Diagn. 2012; p. 351.Google Scholar
  2. 2.
    Maier A, Schuster M, Eysholdt U, et al. QMOS: a robust visualization method for speaker dependencies with different microphones. J Pattern Recognit Res. 2009;4(1):32–51.Google Scholar
  3. 3.
    Wachinger C, Navab N. Manifold learning for multi-modal image registration. Proc BMVC. 2010 01; p. 1–12.Google Scholar
  4. 4.
    Wolz R, Aljabar P, Hajnal JV, et al. LEAP: Learning embeddings for atlas propagation. NeuroImage. 2010;49(2):1316 – 1325.Google Scholar
  5. 5.
    Chen S, Endres J, Dorn S, et al. A feasibility study of automatic multi-organ segmentation using probabilistic atlas. Proc BVM. 2017; p. 218–223.Google Scholar
  6. 6.
    Chen S, Roth H, Dorn S, et al. Towards automatic abdominal multi-organ segmentation in dual energy CT using cascaded 3D fully convolutional network.Google Scholar
  7. 7.
    Roth HR, Oda H, Hayashi Y, et al. Hierarchical 3D fully convolutional networks for multi-organ segmentation.Google Scholar
  8. 8.
    van der Maaten LJP, Postma EO, van den Herik HJ. Dimensionality reduction: a comparative review; 2008.Google Scholar
  9. 9.
    Jiménez-del Toro OA, Dicente Cid Y, Depeursinge A, et al. Hierarchic anatomical structure segmentation guided by spatial correlations (AnatSeg–Gspac): VISCERAL Anatomy3. Proc Visc Chall ISBI. 2015 Apr; p. 22–26.Google Scholar
  10. 10.
    Roweis ST, Saul LK. Nonlinear dimensionality reduction by locally linear embedding. Science. 2000;290(5500):2323–2326.Google Scholar

Copyright information

© Springer-Verlag GmbH Deutschland 2018

Authors and Affiliations

  • Shuqing Chen
    • 1
  • Sabrina Dorn
    • 2
  • Michael Lell
    • 3
  • Marc Kachelrieß
    • 2
  • Andreas Maier
    • 1
  1. 1.Pattern Recognition LabFAU Erlangen-NürnbergErlangenDeutschland
  2. 2.German Cancer Research Center (DKFZ)HeidelbergDeutschland
  3. 3.University Hospital Nürnberg, Paracelsus Medical UniversityNürnbergDeutschland

Personalised recommendations