Pan-cancer integrative histology-genomic analysis via multimodal deep learning

Chen, Richard J.; Lu, Ming; Williamson, Drew F. K.; Chen, Tiffany; Lipková, Jana; Noor, Zahra; Shaban, Muhammad; Shady, Maha; Williams, Mane; Joo, Bumjin; Mahmood, Faisal

doi:10.1016/j.ccell.2022.07.004

Cited by 140 publications

(131 citation statements)

References 78 publications

Supporting

Mentioning

127

Contrasting

Order By: Relevance

“…3 Since then, several "pan-cancer" studies have shown that DL-based prediction of biomarkers is feasible across the whole spectrum of human cancer. [4][5][6][7] However, these studies are overwhelmingly performed in a single large cohort without externally validating the results on a large scale. This raises a number of potential concerns, as prediction performance can be heavily biased by batch effects occurring in such single multicentric datasets.…”

Section: Main Textmentioning

confidence: 99%

See 1 more Smart Citation

Self-supervised deep learning for pan-cancer mutation prediction from histopathology

Saldanha

Loeffler

Niehues

et al. 2022

Preprint

View full text Add to dashboard Cite

The histopathological phenotype of tumors reflects the underlying genetic makeup. Deep learning can predict genetic alterations from tissue morphology, but it is unclear how well these predictions generalize to external datasets. Here, we present a deep learning pipeline based on self-supervised feature extraction which achieves a robust predictability of genetic alterations in two large multicentric datasets of seven tumor types.

show abstract

Section: Main Textmentioning

confidence: 99%

“…3 Since then, several “pan-cancer” studies have shown that DL-based prediction of biomarkers is feasible across the whole spectrum of human cancer. 4–7 . However, these studies are overwhelmingly performed in a single large cohort without externally validating the results on a large scale.…”

Section: Main Textmentioning

confidence: 99%

Self-supervised deep learning for pan-cancer mutation prediction from histopathology

Saldanha

Loeffler

Niehues

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In general, a better understanding of the inner workings of deep learning models would be desirable for this and other biomarker studies in computational pathology. In the future, attentionbased DL methods could further improve performance and interpretability [26,30,31].…”

Section: Discussionmentioning

confidence: 99%

Direct prediction of genetic aberrations from pathology images in gastric cancer with swarm learning

et al. 2022

View full text Add to dashboard Cite

Background Computational pathology uses deep learning (DL) to extract biomarkers from routine pathology slides. Large multicentric datasets improve performance, but such datasets are scarce for gastric cancer. This limitation could be overcome by Swarm Learning (SL). Methods Here, we report the results of a multicentric retrospective study of SL for prediction of molecular biomarkers in gastric cancer. We collected tissue samples with known microsatellite instability (MSI) and Epstein–Barr Virus (EBV) status from four patient cohorts from Switzerland, Germany, the UK and the USA, storing each dataset on a physically separate computer. Results On an external validation cohort, the SL-based classifier reached an area under the receiver operating curve (AUROC) of 0.8092 (± 0.0132) for MSI prediction and 0.8372 (± 0.0179) for EBV prediction. The centralized model, which was trained on all datasets on a single computer, reached a similar performance. Conclusions Our findings demonstrate the feasibility of SL-based molecular biomarkers in gastric cancer. In the future, SL could be used for collaborative training and, thus, improve the performance of these biomarkers. This may ultimately result in clinical-grade performance and generalizability.

show abstract

“…Previous reports on ccRCC genetics have illustrated remodeling of cellular metabolism 15 , abundant indel mutation 16 and adaptive immunity 17 . TCGA image data have been used to indicate that PBRM1-mutated ccRCC samples can be identified from H&E images 7 , and that combination of genomic and image data can improve prognostication compared to standard staging 18,19 . Histologically, ccRCC is characterized by nests of large tumor cells with a conspicuous cytoplasm and surrounded by stromal, vascular, necrotic, adipose, and normal renal tissue and immune cells.…”

Section: Main Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Integrative Analysis of Histological Textures and Lymphocyte Infiltration in Renal Cell Carcinoma using Deep Learning

Brummer

Pölönen

Mustjoki

et al. 2022

Preprint

View full text Add to dashboard Cite

Evaluating tissue architecture from routine hematoxylin and eosin-stained (H&E) slides is prone to subjectivity and sampling bias. Here, we extensively annotated ~40,000 images of five tissue texture types and ~25,000 images of lymphocyte quantity to train deep learning models. We defined histopathological patterns in over 400 clear-cell renal cell carcinoma H&E-stained slides of The Cancer Genome Atlas (TCGA) and resolved sampling and staining differences by harmonizing textural composition. By integrating multi-omic and imaging data, we profiled their clinical, immunological, genomic, and transcriptomic phenotypes. Histological grade, stage, adaptive immunity, the epithelial-to-mesenchymal transition signature and lower mutation burden were more common in stroma-rich samples. Histological proximity between the malignant and normal renal tissues was associated with poor survival, cellular proliferation, tumor heterogeneity, and wild-type PBRM1. This study highlights textural characterization to standardize sampling differences, quantify lymphocyte infiltration and discover novel histopathological associations both in the intratumoral and peritumoral regions.

show abstract

Pan-cancer integrative histology-genomic analysis via multimodal deep learning

Cited by 140 publications

References 78 publications

Self-supervised deep learning for pan-cancer mutation prediction from histopathology

Self-supervised deep learning for pan-cancer mutation prediction from histopathology

Direct prediction of genetic aberrations from pathology images in gastric cancer with swarm learning

Integrative Analysis of Histological Textures and Lymphocyte Infiltration in Renal Cell Carcinoma using Deep Learning

Contact Info

Product

Resources

About