Journal Article

DKFZ-2023-01778

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Transformer-based biomarker prediction from colorectal cancer histology: A large-scale multicentric study.

Wagner, S. J. ; Reisenbüchler, D. ; West, N. P. ; Niehues, J. M. ; Zhu, J. ; Foersch, S. ; Veldhuizen, G. P. ; Quirke, P. ; Grabsch, H. I. ; van den Brandt, P. A. ; Hutchins, G. G. A. ; Richman, S. D. ; Yuan, T.DKFZ* ; Langer, R. ; Jenniskens, J. C. A. ; Offermans, K. ; Mueller, W. ; Gray, R. ; Gruber, S. B. ; Greenson, J. K. ; Rennert, G. ; Bonner, J. D. ; Schmolze, D. ; Jonnagaddala, J. ; Hawkins, N. J. ; Ward, R. L. ; Morton, D. ; Seymour, M. ; Magill, L. ; Nowak, M. ; Hay, J. ; Koelzer, V. H. ; Church, D. N. ; consortium, T. (Collaboration Author) ; Matek, C. ; Geppert, C. ; Peng, C. ; Zhi, C. ; Ouyang, X. ; James, J. A. ; Loughrey, M. B. ; Salto-Tellez, M. ; Brenner, H.DKFZ* ; Hoffmeister, M.DKFZ* ; Truhn, D. ; Schnabel, J. A. ; Boxberg, M. ; Peng, T. ; Kather, J. N. ; Church, D. (Contributor) ; Domingo, E. (Contributor) ; Edwards, J. (Contributor) ; Glimelius, B. (Contributor) ; Gogenur, I. (Contributor) ; Harkin, A. (Contributor) ; Hay, J. (Contributor) ; Iveson, T. (Contributor) ; Jaeger, E. (Contributor) ; Kelly, C. (Contributor) ; Kerr, R. (Contributor) ; Maka, N. (Contributor) ; Morgan, H. (Contributor) ; Oien, K. (Contributor) ; Orange, C. (Contributor) ; Palles, C. (Contributor) ; Roxburgh, C. (Contributor) ; Sansom, O. (Contributor) ; Saunders, M. (Contributor) ; Tomlinson, I. (Contributor)

2023
Elsevier New York, NY

This record in other databases:  

Please use a persistent id in citations: doi:10.1016/j.ccell.2023.08.002

Abstract: Deep learning (DL) can accelerate the prediction of prognostic biomarkers from routine pathology slides in colorectal cancer (CRC). However, current approaches rely on convolutional neural networks (CNNs) and have mostly been validated on small patient cohorts. Here, we develop a new transformer-based pipeline for end-to-end biomarker prediction from pathology slides by combining a pre-trained transformer encoder with a transformer network for patch aggregation. Our transformer-based approach substantially improves the performance, generalizability, data efficiency, and interpretability as compared with current state-of-the-art algorithms. After training and evaluating on a large multicenter cohort of over 13,000 patients from 16 colorectal cancer cohorts, we achieve a sensitivity of 0.99 with a negative predictive value of over 0.99 for prediction of microsatellite instability (MSI) on surgical resection specimens. We demonstrate that resection specimen-only training reaches clinical-grade performance on endoscopic biopsy tissue, solving a long-standing diagnostic problem.

Keyword(s): artificial intelligence ; biomarker ; colorectal cancer ; deep learning ; microsatellite instability ; multiple instance learning ; transformer

Note: 2023 Sep 11;41(9):1650-1661.e4

---

Contributing Institute(s):

1. C070 Klinische Epidemiologie und Alternf. (C070)
2. Präventive Onkologie (C120)
3. DKTK HD zentral (HD01)

Research Program(s):

1. 313 - Krebsrisikofaktoren und Prävention (POF4-313) (POF4-313)

Appears in the scientific report 2023

---

Database coverage:
; BIOSIS Previews ; Biological Abstracts ; Clarivate Analytics Master Journal List ; Current Contents - Clinical Medicine ; Current Contents - Life Sciences ; Ebsco Academic Search ; Essential Science Indicators ; IF >= 50 ; JCR ; SCOPUS ; Science Citation Index Expanded ; Web of Science Core Collection

---