CLDICE - A novel topology-preserving loss function for tubular structure segmentation

Suprosanna Shit, Johannes C. Paetzold, Anjany Sekuboyina, Ivan Ezhov, Alexander Unger, Andrey Zhylka, Josien P.W. Pluim, Ulrich Bauer, Bjoern H. Menze

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

138 Citaten (Scopus)

Samenvatting

Accurate segmentation of tubular, network-like structures, such as vessels, neurons, or roads, is relevant to many fields of research. For such structures, the topology is their most important characteristic; particularly preserving connectedness: in the case of vascular networks, missing a connected vessel entirely alters the blood-flow dynamics. We introduce a novel similarity measure termed centerlineDice (short clDice), which is calculated on the intersection of the segmentation masks and their (morphological) skeleta. We theoretically prove that clDice guarantees topology preservation up to homotopy equivalence for binary 2D and 3D segmentation. Extending this, we propose a computationally efficient, differentiable loss function (soft-clDice) for training arbitrary neural segmentation networks. We benchmark the soft-clDice loss on five public datasets, including vessels, roads and neurons (2D and 3D). Training on soft-clDice leads to segmentation with more accurate connectivity information, higher graph similarity, and better volumetric scores.

Originele taal-2Engels
TitelProceedings - 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2021
UitgeverijIEEE Computer Society
Pagina's16555-16564
Aantal pagina's10
ISBN van elektronische versie9781665445092
DOI's
StatusGepubliceerd - 2021
Evenement2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2021 - Virtual, Online, Verenigde Staten van Amerika
Duur: 19 jun. 202125 jun. 2021

Congres

Congres2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2021
Land/RegioVerenigde Staten van Amerika
StadVirtual, Online
Periode19/06/2125/06/21

Bibliografische nota

Publisher Copyright:
© 2021 IEEE

Financiering

Acknowledgement: J. C. Paetzold. and S. Shit. are sup-portedbytheGCBandTranslatum,TUMunich. S.Shit., A. Zhylka. and I. Ezhov. are supported by TRABIT (EU Grant: 765148). We thank Ali Ertuerk, Mihail I. Todorov, Nils Börnerand Giles Tetteh. J. C. Paetzold. and S. Shit. are supported by the GCB and Translatum, TU Munich. S.Shit., A. Zhylka. and I. Ezhov. are supported by TRABIT (EU Grant: 765148). We thank Ali Ertuerk, Mihail I. Todorov, Nils B?rner and Giles Tetteh.

FinanciersFinanciernummer
Technical University of Munich
European Union’s Horizon Europe research and innovation programme765148

    Vingerafdruk

    Duik in de onderzoeksthema's van 'CLDICE - A novel topology-preserving loss function for tubular structure segmentation'. Samen vormen ze een unieke vingerafdruk.

    Citeer dit