Generalizable biomarker prediction from cancer pathology slides with self-supervised deep learning: A retrospective multi-centric study.
Ontology highlight
ABSTRACT:
SUBMITTER: Niehues JM
PROVIDER: S-EPMC10140458 | biostudies-literature | 2023 Apr
REPOSITORIES: biostudies-literature
Publications
Generalizable biomarker prediction from cancer pathology slides with self-supervised deep learning: A retrospective multi-centric study.
Niehues Jan Moritz JM Quirke Philip P West Nicholas P NP Grabsch Heike I HI van Treeck Marko M Schirris Yoni Y Veldhuizen Gregory P GP Hutchins Gordon G A GGA Richman Susan D SD Foersch Sebastian S Brinker Titus J TJ Fukuoka Junya J Bychkov Andrey A Uegami Wataru W Truhn Daniel D Brenner Hermann H Brobeil Alexander A Hoffmeister Michael M Kather Jakob Nikolas JN
Cell reports. Medicine 20230322 4
Deep learning (DL) can predict microsatellite instability (MSI) from routine histopathology slides of colorectal cancer (CRC). However, it is unclear whether DL can also predict other biomarkers with high performance and whether DL predictions generalize to external patient populations. Here, we acquire CRC tissue samples from two large multi-centric studies. We systematically compare six different state-of-the-art DL architectures to predict biomarkers from pathology slides, including MSI and m ...[more]
Similar Datasets
Project description:AimsOne of the most important complications of heart transplantation is organ rejection, which is diagnosed on endomyocardial biopsies by pathologists. Computer-based systems could assist in the diagnostic process and potentially improve reproducibility. Here, we evaluated the feasibility of using deep learning in predicting the degree of cellular rejection from pathology slides as defined by the International Society for Heart and Lung Transplantation (ISHLT) grading system.Methods and resultsWe collected 1079 histopathology slides from 325 patients from three transplant centres in Germany. We trained an attention-based deep neural network to predict rejection in the primary cohort and evaluated its performance using cross-validation and by deploying it to three cohorts. For binary prediction (rejection yes/no), the mean area under the receiver operating curve (AUROC) was 0.849 in the cross-validated experiment and 0.734, 0.729, and 0.716 in external validation cohorts. For a prediction of the ISHLT grade (0R, 1R, 2/3R), AUROCs were 0.835, 0.633, and 0.905 in the cross-validated experiment and 0.764, 0.597, and 0.913; 0.631, 0.633, and 0.682; and 0.722, 0.601, and 0.805 in the validation cohorts, respectively. The predictions of the artificial intelligence model were interpretable by human experts and highlighted plausible morphological patterns.ConclusionWe conclude that artificial intelligence can detect patterns of cellular transplant rejection in routine pathology, even when trained on small cohorts.
| S-EPMC10232288 | biostudies-literature
Project description:BackgroundHuman epidermal growth factor receptor 2 (HER2)-low breast cancer has emerged as a new subtype of tumor, for which novel antibody-drug conjugates have shown beneficial effects. Assessment of HER2 requires several immunohistochemistry tests with an additional in situ hybridization test if a case is classified as HER2 2+. Therefore, novel cost-effective methods to speed up the HER2 assessment are highly desirable.MethodsWe used a self-supervised attention-based weakly supervised method to predict HER2-low directly from 1437 histopathological images from 1351 breast cancer patients. We built six distinct models to explore the ability of classifiers to distinguish between the HER2-negative, HER2-low, and HER2-high classes in different scenarios. The attention-based model was used to comprehend the decision-making process aimed at relevant tissue regions.ResultsOur results indicate that the effectiveness of classification models hinges on the consistency and dependability of assay-based tests for HER2, as the outcomes from these tests are utilized as the baseline truth for training our models. Through the use of explainable AI, we reveal histologic patterns associated with the HER2 subtypes.ConclusionOur findings offer a demonstration of how deep learning technologies can be applied to identify HER2 subgroup statuses, potentially enriching the toolkit available for clinical decision-making in oncology.
| S-EPMC11331614 | biostudies-literature
Project description:BackgroundTumor mutation burden (TMB) and VHL mutation play a crucial role in the management of patients with clear cell renal cell carcinoma (ccRCC), such as guiding adjuvant chemotherapy and improving clinical outcomes. However, the time-consuming and expensive high-throughput sequencing methods severely limit their clinical applicability. Predicting intratumoral heterogeneity poses significant challenges in biology and clinical settings. Our aimed to develop a self-supervised attention-based multiple instance learning (SSL-ABMIL) model to predict TMB and VHL mutation status from hematoxylin and eosin-stained histopathological images.MethodsWe obtained whole slide images (WSIs) and somatic mutation data of 350 ccRCC patients from The Cancer Genome Atlas for developing SSL-ABMIL model. In parallel, 163 ccRCC patients from Clinical Proteomic Tumor Analysis Consortium cohort was used as independent external validation set. We systematically compared three different models (Wang-ABMIL, Ciga-ABMIL, and ImageNet-MIL) for their ability to predict TMB and VHL alterations.ResultsWe first identified two groups of populations with high- and low-TMB (cut-off point = 0.9). In two independent cohorts, the Wang-ABMIL model achieved the highest performance with decent generalization performance (AUROC = 0.83 ± 0.02 and 0.8 ± 0.04 in predicting TMB and VHL, respectively). Attention heatmaps revealed that the Wang-ABMIL model paid the highest attention to tumor regions in high-TMB patients, while in VHL mutation prediction, non-tumor regions were also assigned high attention, particularly the stromal regions infiltrated by lymphocytes.ConclusionsOur results indicated that SSL-ABMIL can effectively extract histological features for predicting TMB and VHL mutation, demonstrating promising results in linking tumor morphology and molecular biology.
| S-EPMC11336896 | biostudies-literature
Project description:BackgroundEstimating tumor purity is especially important in the age of precision medicine. Purity estimates have been shown to be critical for correction of tumor sequencing results, and higher purity samples allow for more accurate interpretations from next-generation sequencing results. Molecular-based purity estimates using computational approaches require sequencing of tumors, which is both time-consuming and expensive.MethodsHere we propose an approach, weakly-supervised purity (wsPurity), which can accurately quantify tumor purity within a digitally captured hematoxylin and eosin (H&E) stained histological slide, using several types of cancer from The Cancer Genome Atlas (TCGA) as a proof-of-concept.FindingsOur model predicts cancer type with high accuracy on unseen cancer slides from TCGA and shows promising generalizability to unseen data from an external cohort (F1-score of 0.83 for prostate adenocarcinoma). In addition we compare performance of our model on tumor purity prediction with a comparable fully-supervised approach on our TCGA held-out cohort and show our model has improved performance, as well as generalizability to unseen frozen slides (0.1543 MAE on an independent test cohort). In addition to tumor purity prediction, our approach identified high resolution tumor regions within a slide, and can also be used to stratify tumors into high and low tumor purity, using different cancer-dependent thresholds.InterpretationOverall, we demonstrate our deep learning model's different capabilities to analyze tumor H&E sections. We show our model is generalizable to unseen H&E stained slides from data from TCGA as well as data processed at Weill Cornell Medicine.FundingStarr Cancer Consortium Grant (SCC I15-0027) to Iman Hajirasouliha.
| S-EPMC9157012 | biostudies-literature
Project description:Deep Learning (DL) can predict biomarkers from cancer histopathology. Several clinically approved applications use this technology. Most approaches, however, predict categorical labels, whereas biomarkers are often continuous measurements. We hypothesize that regression-based DL outperforms classification-based DL. Therefore, we develop and evaluate a self-supervised attention-based weakly supervised regression method that predicts continuous biomarkers directly from 11,671 images of patients across nine cancer types. We test our method for multiple clinically and biologically relevant biomarkers: homologous recombination deficiency score, a clinically used pan-cancer biomarker, as well as markers of key biological processes in the tumor microenvironment. Using regression significantly enhances the accuracy of biomarker prediction, while also improving the predictions' correspondence to regions of known clinical relevance over classification. In a large cohort of colorectal cancer patients, regression-based prediction scores provide a higher prognostic value than classification-based scores. Our open-source regression approach offers a promising alternative for continuous biomarker analysis in computational pathology.
| S-EPMC10858881 | biostudies-literature
Project description:Beat-by-beat arrhythmia detection in ambulatory electrocardiogram (ECG) monitoring is critical for the evaluation and prognosis of cardiac arrhythmias, however, it is a highly professional demanding and time-consuming task. Current methods for automatic beat-by-beat arrhythmia detection suffer from poor generalization ability due to the lack of large-sample and finely-annotated (labels are given to each beat) ECG data for model training. In this work, we propose a weakly supervised deep learning framework for arrhythmia detection (WSDL-AD), which permits training a fine-grained (beat-by-beat) arrhythmia detector with the use of large amounts of coarsely annotated ECG data (labels are given to each recording) to improve the generalization ability. In this framework, heartbeat classification and recording classification are integrated into a deep neural network for end-to-end training with only recording labels. Several techniques, including knowledge-based features, masked aggregation, and supervised pre-training, are proposed to improve the accuracy and stability of the heartbeat classification under weak supervision. The developed WSDL-AD model is trained for the detection of ventricular ectopic beats (VEB) and supraventricular ectopic beats (SVEB) on five large-sample and coarsely-annotated datasets and the model performance is evaluated on three independent benchmarks according to the recommendations from the Association for the Advancement of Medical Instrumentation (AAMI). The experimental results show that our method improves the F 1 score of supraventricular ectopic beats detection by 8%-290% and the F1 of ventricular ectopic beats detection by 4%-11% on the benchmarks compared with the state-of-the-art methods of supervised learning. It demonstrates that the WSDL-AD framework can leverage the abundant coarsely-labeled data to achieve a better generalization ability than previous methods while retaining fine detection granularity. Therefore, this framework has a great potential to be used in clinical and telehealth applications. The source code is available at https://github.com/sdnjly/WSDL-AD.
| S-EPMC9037749 | biostudies-literature
Project description:Cancer diagnosis and management depend upon the extraction of complex information from microscopy images by pathologists, which requires time-consuming expert interpretation prone to human bias. Supervised deep learning approaches have proven powerful, but are inherently limited by the cost and quality of annotations used for training. Therefore, we present Histomorphological Phenotype Learning, a self-supervised methodology requiring no labels and operating via the automatic discovery of discriminatory features in image tiles. Tiles are grouped into morphologically similar clusters which constitute an atlas of histomorphological phenotypes (HP-Atlas), revealing trajectories from benign to malignant tissue via inflammatory and reactive phenotypes. These clusters have distinct features which can be identified using orthogonal methods, linking histologic, molecular and clinical phenotypes. Applied to lung cancer, we show that they align closely with patient survival, with histopathologically recognised tumor types and growth patterns, and with transcriptomic measures of immunophenotype. These properties are maintained in a multi-cancer study.
| S-EPMC11525555 | biostudies-literature
Project description:ObjectiveThis work investigates the problem of clinical coreference resolution in a model that explicitly tracks entities, and aims to measure the performance of that model in both traditional in-domain train/test splits and cross-domain experiments that measure the generalizability of learned models.MethodsThe two methods we compare are a baseline mention-pair coreference system that operates over pairs of mentions with best-first conflict resolution and a mention-synchronous system that incrementally builds coreference chains. We develop new features that incorporate distributional semantics, discourse features, and entity attributes. We use two new coreference datasets with similar annotation guidelines - the THYME colon cancer dataset and the DeepPhe breast cancer dataset.ResultsThe mention-synchronous system performs similarly on in-domain data but performs much better on new data. Part of speech tag features prove superior in feature generalizability experiments over other word representations. Our methods show generalization improvement but there is still a performance gap when testing in new domains.DiscussionGeneralizability of clinical NLP systems is important and under-studied, so future work should attempt to perform cross-domain and cross-institution evaluations and explicitly develop features and training regimens that favor generalizability. A performance-optimized version of the mention-synchronous system will be included in the open source Apache cTAKES software.
| S-EPMC5508069 | biostudies-literature
Project description:BackgroundPrediction of clinical outcomes for individual cancer patients is an important step in the disease diagnosis and subsequently guides the treatment and patient counseling. In this work, we develop and evaluate a joint outcome and biomarker supervised (estrogen receptor expression and ERBB2 expression and gene amplification) multitask deep learning model for prediction of outcome in breast cancer patients in two nation-wide multicenter studies in Finland (the FinProg and FinHer studies). Our approach combines deep learning with expert knowledge to provide more accurate, robust, and integrated prediction of breast cancer outcomes.Materials and methodsUsing deep learning, we trained convolutional neural networks (CNNs) with digitized tissue microarray (TMA) samples of primary hematoxylin-eosin-stained breast cancer specimens from 693 patients in the FinProg series as input and breast cancer-specific survival as the endpoint. The trained algorithms were tested on 354 TMA patient samples in the same series. An independent set of whole-slide (WS) tumor samples from 674 patients in another multicenter study (FinHer) was used to validate and verify the generalization of the outcome prediction based on CNN models by Cox survival regression and concordance index (c-index). Visual cancer tissue characterization, i.e., number of mitoses, tubules, nuclear pleomorphism, tumor-infiltrating lymphocytes, and necrosis was performed on TMA samples in the FinProg test set by a pathologist and combined with deep learning-based outcome prediction in a multitask algorithm.ResultsThe multitask algorithm achieved a hazard ratio (HR) of 2.0 (95% confidence interval [CI] 1.30-3.00), P < 0.001, c-index of 0.59 on the 354 test set of FinProg patients, and an HR of 1.7 (95% CI 1.2-2.6), P = 0.003, c-index 0.57 on the WS tumor samples from 674 patients in the independent FinHer series. The multitask CNN remained a statistically independent predictor of survival in both test sets when adjusted for histological grade, tumor size, and axillary lymph node status in a multivariate Cox analyses. An improved accuracy (c-index 0.66) was achieved when deep learning was combined with the tissue characteristics assessed visually by a pathologist.ConclusionsA multitask deep learning algorithm supervised by both patient outcome and biomarker status learned features in basic tissue morphology predictive of survival in a nationwide, multicenter series of patients with breast cancer. The algorithms generalized to another independent multicenter patient series and whole-slide breast cancer samples and provide prognostic information complementary to that of a comprehensive series of established prognostic factors.
| S-EPMC8794033 | biostudies-literature
Project description:BackgroundHistopathology images of tumor biopsies present unique challenges for applying machine learning to the diagnosis and treatment of cancer. The pathology slides are high resolution, often exceeding 1GB, have non-uniform dimensions, and often contain multiple tissue slices of varying sizes surrounded by large empty regions. The locations of abnormal or cancerous cells, which may constitute a small portion of any given tissue sample, are not annotated. Cancer image datasets are also extremely imbalanced, with most slides being associated with relatively common cancers. Since deep representations trained on natural photographs are unlikely to be optimal for classifying pathology slide images, which have different spectral ranges and spatial structure, we here describe an approach for learning features and inferring representations of cancer pathology slides based on sparse coding.ResultsWe show that conventional transfer learning using a state-of-the-art deep learning architecture pre-trained on ImageNet (RESNET) and fine tuned for a binary tumor/no-tumor classification task achieved between 85% and 86% accuracy. However, when all layers up to the last convolutional layer in RESNET are replaced with a single feature map inferred via a sparse coding using a dictionary optimized for sparse reconstruction of unlabeled pathology slides, classification performance improves to over 93%, corresponding to a 54% error reduction.ConclusionsWe conclude that a feature dictionary optimized for biomedical imagery may in general support better classification performance than does conventional transfer learning using a dictionary pre-trained on natural images.
| S-EPMC6302377 | biostudies-literature