arXiv reaDer
HistoEncoder: a digital pathology foundation model for prostate cancer
Foundation models are trained on massive amounts of data to distinguish complex patterns and can be adapted to a wide range of downstream tasks with minimal computational resources. Here, we develop a foundation model for prostate cancer digital pathology called HistoEncoder by pre-training on 48 million prostate tissue tile images. We demonstrate that HistoEncoder features extracted from tile images with similar histological patterns map closely together in the feature space. HistoEncoder outperforms models pre-trained with natural images, even without fine-tuning or with 1000 times less training data. We describe two use cases that leverage the capabilities of HistoEncoder by fine-tuning the model with a limited amount of data and computational resources. First, we show how HistoEncoder can be used to automatically annotate large-scale datasets with high accuracy. Second, we combine histomics with commonly used clinical nomograms, significantly improving prostate cancer-specific death survival models. Foundation models such as HistoEncoder can allow organizations with limited resources to build effective clinical software tools without needing extensive datasets or significant amounts of computing.
updated: Fri Nov 22 2024 13:32:01 GMT+0000 (UTC)
published: Mon Nov 18 2024 10:46:05 GMT+0000 (UTC)
参考文献 (このサイトで利用可能なもの) / References (only if available on this site)
被参照文献 (このサイトで利用可能なものを新しい順に) / Citations (only if available on this site, in order of most recent)
Amazon.co.jpアソシエイト