Initialization matters in few-shot adaptation of vision-language models for histopathological image classification

TL;DR

This paper introduces ZS-MIL, a novel initialization method for MIL classifiers in histopathological image classification that leverages VLM text embeddings, improving few-shot learning performance and robustness.

Contribution

The paper proposes ZS-MIL, a new initialization approach using class-level text embeddings to enhance few-shot MIL classification in histopathology.

Findings

ZS-MIL outperforms traditional weight initialization methods.

ZS-MIL demonstrates increased robustness and consistency.

Significant improvement in subtyping prediction accuracy.

Abstract

Vision language models (VLM) pre-trained on datasets of histopathological image-caption pairs enabled zero-shot slide-level classification. The ability of VLM image encoders to extract discriminative features also opens the door for supervised fine-tuning for whole-slide image (WSI) classification, ideally using few labeled samples. Slide-level prediction frameworks require the incorporation of multiple instance learning (MIL) due to the gigapixel size of the WSI. Following patch-level feature extraction and aggregation, MIL frameworks rely on linear classifiers trained on top of the slide-level aggregated features. Classifier weight initialization has a large influence on Linear Probing performance in efficient transfer learning (ETL) approaches based on few-shot learning. In this work, we propose Zero-Shot Multiple-Instance Learning (ZS-MIL) to address the limitations of random classifier initialization that underperform zero-shot prediction in MIL problems. ZS-MIL uses the class-level embeddings of the VLM text encoder as the classification layer's starting point to compute each sample's bag-level probabilities. Through multiple experiments, we demonstrate the robustness of ZS-MIL compared to well-known weight initialization techniques both in terms of performance and variability in an ETL few-shot scenario for subtyping prediction.