A deep learning model to predict RNA-Seq expression of tumours from whole slide images

Schmauch, Benoît; Romagnoni, Alberto; Pronier, Elodie; Saillard, Charlie; Maillé, Pascale; Caldéraro, Julien; Kamoun, Aurélie; Sefta, Meriem; Toldo, Sylvain; Zaslavskiy, Mikhail; Clozel, Thomas; Moarii, Matahi; Courtiol, Pierre; Wainrib, Gilles

doi:10.1038/s41467-020-17678-4

Cited by 357 publications

(357 citation statements)

References 69 publications

Supporting

Mentioning

350

Contrasting

Order By: Relevance

“…Studies published over the past 1-2 years have pursued a "pan-cancer pan-mutation" approach to try to predict any genetic alteration in any type of solid tumour directly from H&E histology. [38][39][40] However, these studies have been largely based on one particular dataset, "The Cancer Genome Atlas (TCGA)", provided by the National Cancer Institute (NCI), and so large-scale validation in genomically characterised cohorts beyond TCGA is needed to gauge the robustness of these methods in pancancer applications.…”

Section: Prediction Of Genotype and Gene Expressionmentioning

confidence: 99%

Deep learning in cancer pathology: a new generation of clinical biomarkers

et al. 2020

View full text Add to dashboard Cite

Clinical workflows in oncology rely on predictive and prognostic molecular biomarkers. However, the growing number of these complex biomarkers tends to increase the cost and time for decision-making in routine daily oncology practice; furthermore, biomarkers often require tumour tissue on top of routine diagnostic material. Nevertheless, routinely available tumour tissue contains an abundance of clinically relevant information that is currently not fully exploited. Advances in deep learning (DL), an artificial intelligence (AI) technology, have enabled the extraction of previously hidden information directly from routine histology images of cancer, providing potentially clinically useful information. Here, we outline emerging concepts of how DL can extract biomarkers directly from histology images and summarise studies of basic and advanced image analysis for cancer histology. Basic image analysis tasks include detection, grading and subtyping of tumour tissue in histology images; they are aimed at automating pathology workflows and consequently do not immediately translate into clinical decisions. Exceeding such basic approaches, DL has also been used for advanced image analysis tasks, which have the potential of directly affecting clinical decision-making processes. These advanced approaches include inference of molecular features, prediction of survival and end-to-end prediction of therapy response. Predictions made by such DL systems could simplify and enrich clinical decision-making, but require rigorous external validation in clinical settings.

show abstract

Section: Prediction Of Genotype and Gene Expressionmentioning

confidence: 99%

Deep learning in cancer pathology: a new generation of clinical biomarkers

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Multiple studies have trained and validated models using cases from TCGA without external validation or isolating sites to either the training or validation datasets. Such studies include the classification of cancer histology in lung cancer 17 , genetic mutation prediction in multiple cancer types 16,17,40 , prediction of grade in clear cell renal cancer 41 , prediction of breast cancer molecular subtype 42 , the prediction of gene expression 13 , or correlation of histology and outcome 40,43 . Survival outcomes are particularly challenging to develop rigorous models for using histology from TCGA, and model performance may be falsely elevated not only by the disparate outcomes across sites, but also the site level differences in critical factors relevant to survival such as stage and age.…”

Section: Discussionmentioning

confidence: 99%

“…Recently, deep learning approaches have been applied to identify less apparent histologic features, including clinical biomarkers such as breast cancer receptor status 4,9 , microsatellite instability 10,11 , or the presence of pathogenic virus in cancer 12 . These approaches have been further extended to infer more subtle features of disease, including gene expression [13][14][15] and pathogenic mutations 16,17 . The predictive accuracy of many of these models have been validated in external datasets, but many studies rely on single data sources for both training and validation.…”

Section: Mainmentioning

confidence: 99%

The Impact of Digital Histopathology Batch Effect on Deep Learning Model Accuracy and Bias

Howard¹,

Dolezal²,

Kochanny³

et al. 2020

Preprint

View full text Add to dashboard Cite

The Cancer Genome Atlas (TCGA) is one of the largest biorepositories of digital histology. Deep learning (DL) models have been trained on TCGA to predict numerous features directly from histology, including survival, gene expression patterns, and driver mutations. However, we demonstrate that these features vary substantially across tissue submitting sites in TCGA for over 3,000 patients with six cancer subtypes. Additionally, we show that histologic image differences between submitting sites can easily be identified with DL. This site detection remains possible despite commonly used color normalization and augmentation methods, and we quantify the digital image characteristics constituting this histologic batch effect. As an example, we show that patient ethnicity within the TCGA breast cancer cohort can be inferred from histology due to site-level batch effect, which must be accounted for to ensure equitable application of DL. Batch effect also leads to overoptimistic estimates of model performance, and we propose a quadratic programming method to guide validation that abrogates this bias.

show abstract

“…CNN are essentially made by several filters, used to extract the high number of information available within images and aggregate them into a lower amount, still relevant to complete the assigned duty. In the field of histopathology, CNN have already been used for several tasks: from the detection of a simple object as a mitotic figure [6], trough the classification of prostate cancer grading [7], to the intriguing potential to pick up, from a simple H/E staining, information regarding the prognosis [8], the response to treatment [9] or even the presence of molecular alterations [10].…”

Section: Introductionmentioning

confidence: 99%

Development of a Deep-learning Pipeline to Recognize and Characterize Macrophages in Colo-rectal Liver Metastasis

Cancian

Cortese

Donadon

et al. 2020

Preprint

View full text Add to dashboard Cite

Quantitative analysis of tumor microenvironment (TME) provides prognostic and predictive information in several human cancers but, with few exceptions, it is not performed in the daily clinical practice being time-consuming. We recently showed that the morphology of tumor associated macrophages (TAM) correlates with outcome in patients with colo-rectal liver metastases (CLM). However, as for other TME components, recognizing and characterizing hundreds of TAM in a single histopathological slide is unfeasible. To fasten this process, we explored a deep-learning based solution. We tested three convolutional neural networks (CNN), Unet, SegNet and DeepLab-v3, and compared their results according to IoU (intersection over union), a metric describing the similarity between what CNN predicts as TAM and the ground truth and SBD (symmetric best dice), which indicates the ability of CNN to separate different TAMs. Unet and SegNet showed intrinsic limitations in discriminating single TAMs (highest SBD 61.34 ± 2.21), whereas DeepLab-v3 accurately recognized TAM from the background [IoU (89.13 ± 3.85)] and separated different TAM [SBD (79.00 ± 3.72)]. This deep-learning pipeline to recognize TAMs in digital slides, will allow the characterization of TAM-related metrics in the daily clinical practice, allowing the implementation of prognostic tools

show abstract

A deep learning model to predict RNA-Seq expression of tumours from whole slide images

Cited by 357 publications

References 69 publications

Deep learning in cancer pathology: a new generation of clinical biomarkers

Deep learning in cancer pathology: a new generation of clinical biomarkers

The Impact of Digital Histopathology Batch Effect on Deep Learning Model Accuracy and Bias

Development of a Deep-learning Pipeline to Recognize and Characterize Macrophages in Colo-rectal Liver Metastasis

Contact Info

Product

Resources

About