Computational pathology

Abstract

Computational pathology offers a comprehensive framework for advanced study design in a wide range of research questions, as well as for standardized pipeline development for fast and reproducible computer-assisted routine diagnostics. This new field emerges at the border of pathology and computer science and shows high potential to revolutionize established workflows in research and clinic, since not only computational models get faster and more efficient than before but also since an incredible amount of training data is being generated in modern hospitals which is mandatory for the training of informed and validated models. We review the field of computational pathology and illustrate on two research examples how it will contribute to an accurate, objective, and reproducible study design comprising informed data acquisition, advanced pattern recognition, and transparent model validation.