Unsupervised Contrastive Learning of Image Representations from Ultrasound Videos with Hard Negative Mining

TL;DR

This paper introduces a novel unsupervised contrastive learning framework that uses hard negative mining within ultrasound videos to improve image representations, significantly enhancing medical diagnosis accuracy.

Contribution

It proposes a hardness-sensitive negative mining curriculum for UCL in ultrasound videos, utilizing intra- and cross-video negatives, and constructs a large-scale US video dataset for gallbladder malignancy detection.

Findings

Improved accuracy in GB malignancy detection by 2-6% over SOTA methods.

Enhanced generalization on COVID-19 lung US dataset with 1.5% improvement.

First large-scale US video dataset with 15,800 frames for GB analysis.

Abstract

Rich temporal information and variations in viewpoints make video data an attractive choice for learning image representations using unsupervised contrastive learning (UCL) techniques. State-of-the-art (SOTA) contrastive learning techniques consider frames within a video as positives in the embedding space, whereas the frames from other videos are considered negatives. We observe that unlike multiple views of an object in natural scene videos, an Ultrasound (US) video captures different 2D slices of an organ. Hence, there is almost no similarity between the temporally distant frames of even the same US video. In this paper we propose to instead utilize such frames as hard negatives. We advocate mining both intra-video and cross-video negatives in a hardness-sensitive negative mining curriculum in a UCL framework to learn rich image representations. We deploy our framework to learn the representations of Gallbladder (GB) malignancy from US videos. We also construct the first large-scale US video dataset containing 64 videos and 15,800 frames for learning GB representations. We show that the standard ResNet50 backbone trained with our framework improves the accuracy of models pretrained with SOTA UCL techniques as well as supervised pretrained models on ImageNet for the GB malignancy detection task by 2-6%. We further validate the generalizability of our method on a publicly available lung US image dataset of COVID-19 pathologies and show an improvement of 1.5% compared to SOTA. Source code, dataset, and models are available at https://gbc-iitd.github.io/usucl.