Benchmarking weakly-supervised deep learning pipelines for whole slide classification in computational pathology

Narmin Ghaffari Laleh, Hannah Sophie Muti, Chiara Maria Lavinia Loeffler, Amelie Echle, Oliver Lester Saldanha, Faisal Mahmood, Ming Y. Lu, Christian Trautwein, Rupert Langer, Bastian Dislich, Roman David Bülow, Heike Irmgard Grabsch, Hermann Brenner, Jenny Chang-Claude, Elizabeth Alwers, Titus J. Brinker, Firas Khader, Daniel Truhn, Nadine T. Gaisa, Peter Boor et al. (3 additional authors not shown)

Abstract: Highlights•We provide a large-scale benchmarking comparison of multiple Deep Learning approaches for analysis of pathology slides in multiple large patient cohorts.•HIA, our computational implementation is not limited to one particular method and is highly reusable•For the first time we use Vision Transformers (ViT) in computational pathology•ViTs outperform convolutional neural networks (CNNs) in clinically relevant prediction tasks and could become the new standard in this field