Abstract

Rich temporal information and variations in viewpoints makes 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 Sonography (USG) video captures different 2D slices of an organ. Hence, there is almost no similarity between the temporally distant frames of even the same USG 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 USG videos. We also construct the first large-scale USG video dataset containing 64 videos and 15,800 frames for learning GB representations. We show that 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 USG image dataset of COVID-19 pathologies and show an improvement of 1% compared to SOTA.

Key Results

BibTeX (Citation)

@@inproceedings{basu2022unsupervised,
  title={Unsupervised Contrastive Learning of Image Representations from Ultrasound Videos with Hard Negative Mining},
  author={Basu, Soumen and Singla, Somanshu and Gupta, Mayank and Rana, Pratyaksha and Gupta, Pankaj and Arora, Chetan},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={423--433},
  year={2022},
  organization={Springer}
}