PreMix: Label-Efficient Multiple Instance Learning via Non-Contrastive Pre-training and Feature Mixing

Multiple instance learning (MIL) has emerged as a powerful framework for weakly supervised whole slide image (WSI) classification, enabling slide-level predictions without requiring detailed patch-level annotations. Despite its success, a critical limitation of current MIL methods lies in the underutilization of pre-training for the MIL aggregator. Most existing approaches initialize the aggregator randomly and train it from scratch, making performance highly sensitive to the quantity of labeled WSIs and ignoring the abundance of unlabeled WSIs commonly available in clinical settings. To address this, we propose PreMix, a novel framework that leverages a non-contrastive pre-training method, Barlow Twins, augmented with the Slide Mixing approach to generate additional positive pairs and enhance feature learning, particularly under limited labeled WSI conditions. Fine-tuning with Mixup and Manifold Mixup further enhances robustness by effectively handling the diverse sizes of gigapixel WSIs. Experimental results demonstrate that integrating PreMix as a plug-in module into HIPT yields an average F1 improvement of 4.7% over the baseline HIPT across various WSI training sizes and datasets. These findings underscore its potential to advance WSI classification with limited labeled data and its applicability to real-world histopathology practices. The code is available at https://github.com/bryanwong17/PreMix