VLCD: Vision-Language Contrastive Distillation for Accurate and Efficient Automatic Placenta Analysis

TL;DR

This paper introduces VLCD, a novel vision-language contrastive distillation method that enhances accuracy and efficiency in automatic placenta analysis, making AI healthcare tools more accessible and deployable.

Contribution

It proposes a new text-anchored knowledge distillation strategy and unsupervised predistillation to improve medical vision-language models' performance and efficiency.

Findings

VLCD achieves model compression and acceleration.

Unsupervised predistillation improves robustness on low-quality images.

The approach surpasses previous methods in accuracy and efficiency.

Abstract

Pathological examination of the placenta is an effective method for detecting and mitigating health risks associated with childbirth. Recent advancements in AI have enabled the use of photographs of the placenta and pathology reports for detecting and classifying signs of childbirth-related pathologies. However, existing automated methods are computationally extensive, which limits their deployability. We propose two modifications to vision-language contrastive learning (VLC) frameworks to enhance their accuracy and efficiency: (1) text-anchored vision-language contrastive knowledge distillation (VLCD)-a new knowledge distillation strategy for medical VLC pretraining, and (2) unsupervised predistillation using a large natural images dataset for improved initialization. Our approach distills efficient neural networks that match or surpass the teacher model in performance while achieving model compression and acceleration. Our results showcase the value of unsupervised predistillation in improving the performance and robustness of our approach, specifically for lower-quality images. VLCD serves as an effective way to improve the efficiency and deployability of medical VLC approaches, making AI-based healthcare solutions more accessible, especially in resource-constrained environments.