Deep Multiple Instance Learning with Distance-Aware Self-Attention

TL;DR

This paper introduces DAS-MIL, a novel multiple instance learning model that incorporates continuous, distance-aware self-attention to explicitly model spatial relationships between image patches, improving performance in medical image analysis.

Contribution

The paper presents the first application of continuous, distance-dependent relative position representations in MIL, enhancing patch interaction modeling with spatial awareness.

Findings

Achieved 0.91 AUROC on CAMELYON16 dataset.

Outperformed existing MIL models with absolute and relative positional encodings.

Demonstrated effectiveness on a custom MNIST-based MIL dataset.

Abstract

Traditional supervised learning tasks require a label for every instance in the training set, but in many real-world applications, labels are only available for collections (bags) of instances. This problem setting, known as multiple instance learning (MIL), is particularly relevant in the medical domain, where high-resolution images are split into smaller patches, but labels apply to the image as a whole. Recent MIL models are able to capture correspondences between patches by employing self-attention, allowing them to weigh each patch differently based on all other patches in the bag. However, these approaches still do not consider the relative spatial relationships between patches within the larger image, which is especially important in computational pathology. To this end, we introduce a novel MIL model with distance-aware self-attention (DAS-MIL), which explicitly takes into account relative spatial information when modelling the interactions between patches. Unlike existing relative position representations for self-attention which are discrete, our approach introduces continuous distance-dependent terms into the computation of the attention weights, and is the first to apply relative position representations in the context of MIL. We evaluate our model on a custom MNIST-based MIL dataset that requires the consideration of relative spatial information, as well as on CAMELYON16, a publicly available cancer metastasis detection dataset, where we achieve a test AUROC score of 0.91. On both datasets, our model outperforms existing MIL approaches that employ absolute positional encodings, as well as existing relative position representation schemes applied to MIL. Our code is available at https://anonymous.4open.science/r/das-mil.