Effective semantic segmentation in Cataract Surgery: What matters most?

TY - CHAP

T1 - Effective semantic segmentation in Cataract Surgery: What matters most?

AU - Pissas, Theodoros

AU - Ravasio, Claudio

AU - Da Cruz, Lyndon

AU - Bergeles, Christos

N1 - Funding Information: Acknowledgements. The authors would like to thank Martin Huber, Jeremy Birch and Joan M. Nunez Do Rio for their contributions in the EndoVIS challenge participation. This work was supported by the National Institute for Health Research NIHR (Invention for Innovation, i4i; II-LB-0716-20002). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care. Publisher Copyright: © 2021, Springer Nature Switzerland AG.

PY - 2021/7/10

Y1 - 2021/7/10

N2 - 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.

AB - 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.

UR - http://www.scopus.com/inward/record.url?scp=85116464637&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-87202-1_49

DO - 10.1007/978-3-030-87202-1_49

M3 - Conference paper

SN - 9783030872014

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 509

EP - 518

BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2021 - 24th International Conference, Proceedings

A2 - de Bruijne, Marleen

A2 - de Bruijne, Marleen

A2 - Cattin, Philippe C.

A2 - Cotin, Stéphane

A2 - Padoy, Nicolas

A2 - Speidel, Stefanie

A2 - Zheng, Yefeng

A2 - Essert, Caroline

ER -