Effective semantic segmentation in Cataract Surgery: What matters most?

TL;DR

This paper introduces neural network design strategies that significantly improve semantic segmentation performance in cataract surgery videos, especially for rare tool classes, by addressing class imbalance.

Contribution

It presents a novel approach combining data oversampling and loss functions to enhance segmentation accuracy on imbalanced surgical datasets.

Findings

Achieved state-of-the-art results on the CaDIS benchmark.

Significant improvements in rare tool class segmentation.

Effective handling of class imbalance in surgical videos.

Abstract

Our work proposes neural network design choices that set the state-of-the-art on a challenging public benchmark on cataract surgery, CaDIS. Our methodology achieves strong performance across three semantic segmentation tasks with increasingly granular surgical tool class sets by effectively handling class imbalance, an inherent challenge in any surgical video. We consider and evaluate two conceptually simple data oversampling methods as well as different loss functions. We show significant performance gains across network architectures and tasks especially on the rarest tool classes, thereby presenting an approach for achieving high performance when imbalanced granular datasets are considered. Our code and trained models are available at https://github.com/RViMLab/MICCAI2021_Cataract_semantic_segmentation and qualitative results on unseen surgical video can be found at https://youtu.be/twVIPUj1WZM.