Hidden Variables in Deep Learning Digital Pathology and Their Potential to Cause Batch Effects: Prediction Model Study.

Journal: Journal of medical Internet research
Published Date: Feb 2, 2021

Abstract

BACKGROUND: An increasing number of studies within digital pathology show the potential of artificial intelligence (AI) to diagnose cancer using histological whole slide images, which requires large and diverse data sets. While diversification may result in more generalizable AI-based systems, it can also introduce hidden variables. If neural networks are able to distinguish/learn hidden variables, these variables can introduce batch effects that compromise the accuracy of classification systems.

Authors

  • Max Schmitt
    National Center for Tumor Diseases, German Cancer Research Center, Heidelberg, Germany.
  • Roman Christoph Maron
    Digital Biomarkers for Oncology Group, National Center for Tumor Diseases, German Cancer Research Center (DKFZ), Heidelberg, Germany.
  • Achim Hekler
    National Center for Tumor Diseases, Department of Translational Oncology, German Cancer Research Center, Heidelberg, Germany.
  • Albrecht Stenzinger
    From the Division of Radiology, German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany (P.S., J.P.R., P.K., H.P.S., D.B.); Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany (S.K., K.H.M.H.); Department of Urology, University of Heidelberg Medical Center, Heidelberg, Germany (J.P.R., M.H.); Division of Biostatistics, German Cancer Research Center (DKFZ), Heidelberg, Germany (M.W.); Department of Neuroradiology, University of Heidelberg Medical Center, Heidelberg, Germany (P.K.); Junior Group Medical Imaging and Radiology-Cancer Prevention, German Cancer Research Center (DKFZ), Heidelberg, Germany (S.B.); Division of Medical Physics, German Cancer Research Center (DKFZ), Heidelberg, Germany (T.A.K.); Institute of Pathology, University of Heidelberg Medical Center, Heidelberg, Germany (A.S.); and German Cancer Consortium (DKTK), Heidelberg, Germany (H.P.S., K.H.M.H., D.B.).
  • Axel Hauschild
    Department of Dermatology, University Hospital Kiel, Kiel, Germany.
  • Michael Weichenthal
    Department of Dermatology, University Hospital Kiel, Kiel, Germany.
  • Markus Tiemann
    Institute for Hematopathology Hamburg, Hamburg, Germany.
  • Dieter Krahl
    Private Laboratory of Dermatohistopathology, Mönchhofstraße 52, 69120 Heidelberg.
  • Heinz Kutzner
    Dermatopathology Laboratory, Friedrichshafen, Germany.
  • Jochen Sven Utikal
    Skin Cancer Unit, German Cancer Research Center, Heidelberg, Germany.
  • Sebastian Haferkamp
    Department of Dermatology, University Hospital Regensburg, Regensburg, Germany.
  • Jakob Nikolas Kather
    Department of Medicine III, University Hospital RWTH Aachen, Aachen, Germany.
  • Frederick Klauschen
    Pathologisches Institut, Ludwig-Maximilians-Universität München, Thalkirchner Str. 36, 80337, München, Deutschland. f.klauschen@lmu.de.
  • Eva Krieghoff-Henning
    Digital Biomarkers for Oncology Group (DBO), National Center for Tumor Diseases (NCT), German Cancer Research Center (DKFZ), Heidelberg, Germany.
  • Stefan Fröhling
    National Center for Tumor Diseases (NCT), German Cancer Research Center (DKFZ), Im Neuenheimer Feld 460, 69120 Heidelberg, Germany.
  • Christof von Kalle
    National Center for Tumor Diseases, Department of Translational Oncology, German Cancer Research Center, Heidelberg, Germany.
  • Titus Josef Brinker
    National Center for Tumor Diseases, Department of Translational Oncology, German Cancer Research Center, Heidelberg, Germany.

Keywords

External Resources

Popular Topics
Recent Journals