Artificial Intelligence (AI) has great potential to improve health outcomes by training systems on vast digitized clinical datasets. Computational Pathology, with its massive amounts of microscopy image data and impact on diagnostics and biomarkers, is at the forefront of this development. Gigapixel pathology slides pose a unique challenge due to their enormous size and are usually divided into tens of thousands of smaller tiles for analysis. This results in a discontinuity in the machine learning process by separating the training of tile-level encoders from slide-level aggregators and the need to adopt weakly supervised learning strategies. Training models from entire pathology slides end-to-end has been largely unexplored due to its computational challenges. To overcome this problem, we propose a novel approach to jointly train both a tile encoder and a slide-aggregator fully in memory and end-to-end at high-resolution, bridging the gap between input and slide-level supervision. While more computationally expensive, detailed quantitative validation shows promise for large-scale pre-training of pathology foundation models.
Recent breakthroughs in self-supervised learning have enabled the use of large unlabeled datasets to train visual foundation models that can generalize to a variety of downstream tasks. While this training paradigm is well suited for the medical domain where annotations are scarce, large-scale pre-training in the medical domain, and in particular pathology, has not been extensively studied. Previous work in self-supervised learning in pathology has leveraged smaller datasets for both pre-training and evaluating downstream performance. The aim of this project is to train the largest academic foundation model and benchmark the most prominent self-supervised learning algorithms by pre-training and evaluating downstream performance on large clinical pathology datasets. We collected the largest pathology dataset to date, consisting of over 3 billion images from over 423 thousand microscopy slides. We compared pre-training of visual transformer models using the masked autoencoder (MAE) and DINO algorithms. We evaluated performance on six clinically relevant tasks from three anatomic sites and two institutions: breast cancer detection, inflammatory bowel disease detection, breast cancer estrogen receptor prediction, lung adenocarcinoma EGFR mutation prediction, and lung cancer immunotherapy response prediction. Our results demonstrate that pre-training on pathology data is beneficial for downstream performance compared to pre-training on natural images. Additionally, the DINO algorithm achieved better generalization performance across all tasks tested. The presented results signify a phase change in computational pathology research, paving the way into a new era of more performant models based on large-scale, parallel pre-training at the billion-image scale.
Lung cancer is the leading cause of cancer death worldwide, with lung adenocarcinoma being the most prevalent form of lung cancer. EGFR positive lung adenocarcinomas have been shown to have high response rates to TKI therapy, underlying the essential nature of molecular testing for lung cancers. Despite current guidelines consider testing necessary, a large portion of patients are not routinely profiled, resulting in millions of people not receiving the optimal treatment for their lung cancer. Sequencing is the gold standard for molecular testing of EGFR mutations, but it can take several weeks for results to come back, which is not ideal in a time constrained scenario. The development of alternative screening tools capable of detecting EGFR mutations quickly and cheaply while preserving tissue for sequencing could help reduce the amount of sub-optimally treated patients. We propose a multi-modal approach which integrates pathology images and clinical variables to predict EGFR mutational status achieving an AUC of 84% on the largest clinical cohort to date. Such a computational model could be deployed at large at little additional cost. Its clinical application could reduce the number of patients who receive sub-optimal treatments by 53.1% in China, and up to 96.6% in the US.