Pan-cancer image-based detection of clinically actionable genetic alterations

By Jakob Nikolas Kather, Lara R. Heij, H. Grabsch, L. Kooreman, C. Loeffler, A. Echle, Jeremias Krause, H. Muti, J. Niehues, Kai A. J. Sommer and 14 more. Nature Cancer, 2019

600 citations recorded by the source corpus.

Molecular alterations in cancer can cause phenotypic changes in tumor cells and their microenvironment. Routine histopathology tissue slides, which are ubiquitously available, can reflect such morphological changes. Here, we show that deep learning can consistently infer a wide range of genetic mutations, molecular tumor subtypes, gene expression signatures and standard pathology biomarkers directly from routine histology. We developed, optimized, validated and publicly released a one-stop-shop workflow and applied it to tissue slides of more than 5,000 patients across multiple solid tumors. Our findings show that a single deep learning algorithm can be trained to predict a wide range of molecular alterations from routine, paraffin-embedded histology slides stained with hematoxylin and eosin. These predictions generalize to other populations and are spatially resolved. Our method can be implemented on mobile hardware, potentially enabling point-of-care diagnostics for personalized cancer treatment. More generally, this approach could elucidate and quantify genotype–phenotype links in cancer. Two papers by Kather and colleagues and Gerstung and colleagues develop workflows to predict a wide range of molecular alterations from pan-cancer digital pathology slides.

Subfields: Artificial Intelligence, Biophysics, Cancer Research

View Paper: https://www.semanticscholar.org/paper/882207f21ec0a72212cda52876c3a93eb1b72204

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Biophysics

In Cancer Research

Citations in the corpus

Pan-cancer computational histopathology reveals mutations, tumor composition and prognosis (2019, background): “Also, less complex CNN architectures may suffice for classifying histopathology patterns 9,18 , because tumor sections have a defined scale, unlike everyday images.”
Deep learning in cancer pathology: a new generation of clinical biomarkers (2020, background): “By contrast, DL-based evaluation of scanned routine histology slides does not incur any significant cost or time and could be deployed even on mobile hardware.(39) Notably, however, in all DL-based studies carried out so far, the performance (as measured by AUROC) has varied according to the sample ”
Clinical-grade Detection of Microsatellite Instability in Colorectal Tumors by Deep Learning. (2020, methodology): “following 4 previous studies with the intent of retraining a previously described deep learning system.(16,17) First, we used the publicly available Cancer Genome Atlas (TCGA) (n 1⁄4 616”
Artificial intelligence in histopathology: enhancing cancer research and clinical oncology (2022, background): “Highly scoring image tiles can be visualized 19,20 , allowing human experts to check the plausibility of AI systems and also discover new features with the help of AI 21,22 (Fig.”
Harnessing multimodal data integration to advance precision oncology (2021, background): “While weak labels may be incomplete, inexact or inaccurate 98 , WSL applications in computational pathology have resulted in robust models to infer genomic alterations 31 and diagnose cancer 99 .”
Artificial intelligence for digital and computational pathology (2023, background): “5) and translational considerations for clinical deployment (Box 3).”
Deep learning in cancer diagnosis, prognosis and treatment selection (2021, methodology): “[70] has also showed that CNN models trained and evaluated on TCGA H&E slides can accurately predict a range of actionable genetic alterations across multiple cancer types, including mutational status of key genes, molecular subtypes and gene expression of standard biomarkers such as hormone recepto”