# Effective semantic segmentation in Cataract Surgery: What matters most?

Source: [Effective semantic segmentation in Cataract Surgery: What matters most?](<https://www.sota2.com/research/paper/68fa6a921e209131cea28fbb>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-04T03:34:04.118Z

Authors: Theodoros Pissas, Claudio Ravasio, Lyndon Da Cruz, Christos Bergeles

Published: 2021-08-13T08:27:54.000Z

## 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.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Semantic Segmentation on CaDIS t1 1.0 (val)](<https://www.sota2.com/research/sota/semantic-segmentation-on-cadis-t1-1-0-val.md>): Semantic segmentation; CaDIS t1 1.0 (val); mIoU: 88.97.

- [Semantic Segmentation on CaDIS t1 1.0 (test)](<https://www.sota2.com/research/sota/semantic-segmentation-on-cadis-t1-1-0-test.md>): Semantic segmentation; CaDIS t1 1.0 (test); mIoU: 0.864.

- [Semantic Segmentation on CaDIS t2 1.0 (val)](<https://www.sota2.com/research/sota/semantic-segmentation-on-cadis-t2-1-0-val.md>): Semantic segmentation; CaDIS t2 1.0 (val); mIoU: 83.25.

- [Semantic Segmentation on CaDIS t2 1.0 (test)](<https://www.sota2.com/research/sota/semantic-segmentation-on-cadis-t2-1-0-test.md>): Semantic segmentation; CaDIS t2 1.0 (test); mIoU: 79.09.

- [Semantic Segmentation on CaDIS t3 1.0 (val)](<https://www.sota2.com/research/sota/semantic-segmentation-on-cadis-t3-1-0-val.md>): Semantic segmentation; CaDIS t3 1.0 (val); mIoU: 79.4.

- [Semantic Segmentation on CaDIS t3 1.0 (test)](<https://www.sota2.com/research/sota/semantic-segmentation-on-cadis-t3-1-0-test.md>): Semantic segmentation; CaDIS t3 1.0 (test); mIoU: 71.94.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2108.06119>)

[PDF](<https://arxiv.org/pdf/2108.06119.pdf>)

[Implementation](<https://github.com/RViMLab/MICCAI2021_Cataract_semantic_segmentation>)