000280940 001__ 280940
000280940 005__ 20250918102642.0
000280940 037__ $$aDZNE-2025-01023
000280940 1001_ $$0P:(DE-HGF)0$$aRoy, Saikat$$b0
000280940 1112_ $$a5th International Conference on Medical Imaging with Deep Learning$$cZurich$$d2022-07-06 - 2022-07-08$$gMIDL 2022$$wSwitzerland
000280940 245__ $$aAre 2.5D approaches superior to 3D deep networks in whole brain segmentation?
000280940 260__ $$bML Research Press$$c2022
000280940 29510 $$aProceedings of Machine Learning Research
000280940 300__ $$a988 - 1004
000280940 3367_ $$2ORCID$$aCONFERENCE_PAPER
000280940 3367_ $$033$$2EndNote$$aConference Paper
000280940 3367_ $$2BibTeX$$aINPROCEEDINGS
000280940 3367_ $$2DRIVER$$aconferenceObject
000280940 3367_ $$2DataCite$$aOutput Types/Conference Paper
000280940 3367_ $$0PUB:(DE-HGF)8$$2PUB:(DE-HGF)$$aContribution to a conference proceedings$$bcontrib$$mcontrib$$s1758103921_31852
000280940 3367_ $$0PUB:(DE-HGF)7$$2PUB:(DE-HGF)$$aContribution to a book$$mcontb
000280940 4900_ $$v172
000280940 520__ $$aSegmentation of 3D volumes with a large number of labels, small convoluted structures, and lack of contrast between various structural boundaries is a difficult task. While recent methodological advances across many segmentation tasks are dominated by 3D architectures, currently the strongest performing method for whole brain segmentation is FastSurferCNN, a 2.5D approach. To shed light on the nuanced differences between 2.5D and various 3D approaches, we perform a thorough and fair comparison and suggest a spatially-ensembled 3D architecture. Interestingly, we observe training memory intensive 3D segmentation on full-view images does not outperform the 2.5D approach. A shift to training on patches even while evaluating on full-view solves these limitations of both memory and performance limitations at the same time. We demonstrate significant performance improvements over state-of-the-art 3D methods on both Dice Similarity Coefficient and especially average Hausdorff Distance measures across five datasets. Finally, our validation across variations of neurodegenerative disease states and scanner manufacturers, shows we outperform the previously leading 2.5D approach FastSurferCNN demonstrating robust segmentation performance in realistic settings. Our code is available online at github.com/Deep-MI/3d-neuro-seg.
000280940 536__ $$0G:(DE-HGF)POF4-354$$a354 - Disease Prevention and Healthy Aging (POF4-354)$$cPOF4-354$$fPOF IV$$x0
000280940 7001_ $$0P:(DE-2719)2814290$$aKügler, David$$b1$$udzne
000280940 7001_ $$0P:(DE-2719)2812134$$aReuter, Martin$$b2$$udzne
000280940 773__ $$y2022
000280940 8564_ $$uhttps://pub.dzne.de/record/280940/files/DZNE-2025-01023_Restricted.pdf
000280940 8564_ $$uhttps://pub.dzne.de/record/280940/files/DZNE-2025-01023_Restricted.pdf?subformat=pdfa$$xpdfa
000280940 909CO $$ooai:pub.dzne.de:280940$$pVDB
000280940 9101_ $$0I:(DE-588)1065079516$$6P:(DE-2719)2814290$$aDeutsches Zentrum für Neurodegenerative Erkrankungen$$b1$$kDZNE
000280940 9101_ $$0I:(DE-588)1065079516$$6P:(DE-2719)2812134$$aDeutsches Zentrum für Neurodegenerative Erkrankungen$$b2$$kDZNE
000280940 9131_ $$0G:(DE-HGF)POF4-354$$1G:(DE-HGF)POF4-350$$2G:(DE-HGF)POF4-300$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$aDE-HGF$$bGesundheit$$lNeurodegenerative Diseases$$vDisease Prevention and Healthy Aging$$x0
000280940 9201_ $$0I:(DE-2719)1040310$$kAG Reuter$$lArtificial Intelligence in Medicine$$x0
000280940 980__ $$acontrib
000280940 980__ $$aVDB
000280940 980__ $$acontb
000280940 980__ $$aI:(DE-2719)1040310
000280940 980__ $$aUNRESTRICTED