Beyond Labels: A Self-Supervised Framework with Masked Autoencoders and Random Cropping for Breast Cancer Subtype Classification
This work addresses the problem of limited labeled data for breast cancer subtype classification, which is crucial for medical professionals, but it is incremental as it builds on existing self-supervised and autoencoder methods.
The authors tackled breast cancer subtype classification from histopathological images by developing a self-supervised framework using masked autoencoders and random cropping, achieving strong performance on the BRACS dataset compared to existing benchmarks.
This work contributes to breast cancer sub-type classification using histopathological images. We utilize masked autoencoders (MAEs) to learn a self-supervised embedding tailored for computer vision tasks in this domain. This embedding captures informative representations of histopathological data, facilitating feature learning without extensive labeled datasets. During pre-training, we investigate employing a random crop technique to generate a large dataset from WSIs automatically. Additionally, we assess the performance of linear probes for multi-class classification tasks of cancer sub-types using the representations learnt by the MAE. Our approach aims to achieve strong performance on downstream tasks by leveraging the complementary strengths of ViTs and autoencoders. We evaluate our model's performance on the BRACS dataset and compare it with existing benchmarks.