{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,24]],"date-time":"2026-09-24T01:01:47Z","timestamp":1790211707247,"version":"4.1.0"},"reference-count":51,"publisher":"SPIE-Intl Soc Optical Eng","issue":"01","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["J. Med. Imag."],"published-print":{"date-parts":[[2024,1,2]]},"abstract":"<jats:p>PurposeThe hippocampus is organized in subfields (HSF) involved in learning and memory processes and widely implicated in pathologies at different ages of life, from neonatal hypoxia to temporal lobe epilepsy or Alzheimer\u2019s disease. Getting a highly accurate and robust delineation of sub-millimetric regions such as HSF to investigate anatomo-functional hypotheses is a challenge. One of the main difficulties encountered by those methodologies is related to the small size and anatomical variability of HSF, resulting in the scarcity of manual data labeling. Recently introduced, capsule networks solve analogous problems in medical imaging, providing deep learning architectures with rotational equivariance. Nonetheless, capsule networks are still two-dimensional and unassessed for the segmentation of HSF.ApproachWe released a public 3D Capsule Network (3D-AGSCaps, https:\/\/github.com\/clementpoiret\/3D-AGSCaps) and compared it to equivalent architectures using classical convolutions on the automatic segmentation of HSF on small and atypical datasets (incomplete hippocampal inversion, IHI). We tested 3D-AGSCaps on three datasets with manually labeled hippocampi.ResultsOur main results were: (1)\u00a03D-AGSCaps produced segmentations with a better Dice Coefficient compared to CNNs on rotated hippocampi (p=0.004, cohen\u2019s d=0.179); (2)\u00a0on typical subjects, 3D-AGSCaps produced segmentations with a Dice coefficient similar to CNNs while having 15 times fewer parameters (2.285M versus 35.069M). This may greatly facilitate the study of atypical subjects, including healthy and pathological cases like those presenting an IHI.ConclusionWe expect our newly introduced 3D-AGSCaps to allow a more accurate and fully automated segmentation on atypical populations, small datasets, as well as on and large cohorts where manual segmentations are nearly intractable.<\/jats:p>","DOI":"10.1117\/1.jmi.11.1.014003","type":"journal-article","created":{"date-parts":[[2024,1,2]],"date-time":"2024-01-02T10:01:01Z","timestamp":1704189661000},"page":"1-17","source":"Crossref","is-referenced-by-count":3,"title":["Attention-gated 3D CapsNet for robust hippocampal segmentation"],"prefix":"10.1117","volume":"11","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-1571-2161","authenticated-orcid":false,"given":"Clement","family":"Poiret","sequence":"first","affiliation":[{"name":"CEA-Paris-Saclay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Antoine","family":"Bouyeure","sequence":"additional","affiliation":[{"name":"CEA Paris-Saclay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sandesh","family":"Patil","sequence":"additional","affiliation":[{"name":"CEA Paris-Saclay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"C\u00e9cile","family":"Boniteau","sequence":"additional","affiliation":[{"name":"CEA Paris-Saclay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Edouard","family":"Duchesnay","sequence":"additional","affiliation":[{"name":"CEA Paris-Saclay"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Antoine","family":"Grigis","sequence":"additional","affiliation":[{"name":"NeuroSpin"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Frederic","family":"Lemaitre","sequence":"additional","affiliation":[{"name":"Universit\u00e9 de Rouen"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2832-0332","authenticated-orcid":false,"given":"Marion","family":"Noulhiane","sequence":"additional","affiliation":[{"name":"CEA Paris-Saclay"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"189","reference":[{"key":"r1","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1014391107"},{"key":"r2","doi-asserted-by":"publisher","DOI":"10.1038\/s41380-018-0036-2"},{"key":"r3","doi-asserted-by":"publisher","DOI":"10.1155\/2015\/450341"},{"key":"r4","doi-asserted-by":"publisher","DOI":"10.3174\/ajnr.A2013"},{"key":"r5","doi-asserted-by":"publisher","DOI":"10.3389\/fnana.2015.00160"},{"key":"r6","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2015.04.042"},{"key":"r7","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2017.09.049"},{"key":"r8","doi-asserted-by":"publisher","DOI":"10.1002\/hbm.22627"},{"key":"r9","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-319-24574-4_28","article-title":"U-Net: convolutional networks for biomedical image segmentation","author":"Ronneberger","year":"2015"},{"key":"r10","article-title":"Exploring the landscape of spatial robustness","author":"Engstrom","year":"2019"},{"key":"r11","article-title":"Geometric robustness of deep networks: analysis and improvement","author":"Kanbak","year":"2017"},{"key":"r12","article-title":"Spatially transformed adversarial examples","author":"Xiao","year":"2018"},{"key":"r13","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijrobp.2019.06.2177"},{"key":"r14","doi-asserted-by":"publisher","DOI":"10.3389\/fninf.2019.00030"},{"key":"r15","article-title":"Examining the benefits of capsule neural networks","author":"Punjabi","year":"2020"},{"key":"r16","first-page":"2012","article-title":"Learning rotation invariant convolutional filters for texture classification","author":"Marcos","year":"2016"},{"key":"r17","first-page":"9","article-title":"Comparing data augmentation strategies for deep image classification","author":"O\u2019Gara","year":"2019"},{"key":"r18","doi-asserted-by":"publisher","DOI":"10.1186\/s40537-019-0197-0"},{"key":"r19","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-642-21735-7_6"},{"key":"r20","article-title":"Dynamic routing between capsules","author":"Sabour","year":"2017"},{"key":"r21","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2020.101889"},{"key":"r22","doi-asserted-by":"publisher","DOI":"10.1007\/s10489-020-01895-x"},{"key":"r23","doi-asserted-by":"publisher","DOI":"10.1016\/j.jksuci.2019.09.014"},{"key":"r24","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-020-64824-5"},{"key":"r25","doi-asserted-by":"crossref","DOI":"10.1109\/ICIP.2018.8451379","article-title":"Brain tumor type classification via capsule networks","author":"Afshar","year":"2018"},{"key":"r26","doi-asserted-by":"publisher","DOI":"10.1016\/j.artmed.2021.102072"},{"key":"r27","doi-asserted-by":"publisher","DOI":"10.1109\/ACSSC.2018.8645256"},{"key":"r28","doi-asserted-by":"publisher","DOI":"10.1109\/WACV.2019.00132"},{"key":"r29","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2020.2971950"},{"key":"r30","article-title":"Attention U-Net: learning where to look for the pancreas","author":"Oktay","year":"2018"},{"key":"r31","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2019.2908039"},{"key":"r32","doi-asserted-by":"publisher","DOI":"10.1007\/978-981-15-5199-4_20"},{"key":"r33","doi-asserted-by":"crossref","DOI":"10.1109\/ICCVW.2019.00247","article-title":"Attention routing between capsules","author":"Choi","year":"2019"},{"key":"r34","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-020-68453-w"},{"key":"r35","doi-asserted-by":"publisher","DOI":"10.1038\/s41598-021-93977-0"},{"key":"r36","article-title":"Capsules with inverted dot-product attention routing","author":"Tsai","year":"2020"},{"key":"r37","article-title":"A large annotated medical image dataset for the development and evaluation of segmentation algorithms","author":"Simpson","year":"2019"},{"key":"r38","doi-asserted-by":"publisher","DOI":"10.1038\/sdata.2015.59"},{"key":"r39","doi-asserted-by":"publisher","DOI":"10.1002\/hipo.23385"},{"key":"r40","doi-asserted-by":"publisher","DOI":"10.1177\/2398212817701448"},{"key":"r41","article-title":"Differentiable learning-to-normalize via switchable normalization","author":"Luo","year":"2019"},{"key":"r42","article-title":"A novel focal Tversky loss function with improved attention U-Net for lesion segmentation","author":"Abraham","year":"2018"},{"key":"r43","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-319-46723-8_49","article-title":"3D U-Net: learning dense volumetric segmentation from sparse annotation","author":"\u00c7i\u00e7ek","year":"2016"},{"key":"r44","first-page":"218","article-title":"Brain tumor segmentation based on 3D residual U-Net","author":"Bhalerao","year":"2020"},{"key":"r45","first-page":"419","article-title":"Deep residual 3D U-Net for joint segmentation and texture classification of nodules in lung","author":"Rassadin","year":"2020"},{"key":"r46","article-title":"Capsule networks need an improved routing algorithm","author":"Paik","year":"2019"},{"key":"r47","doi-asserted-by":"publisher","DOI":"10.3174\/ajnr.A7845"},{"key":"r48","doi-asserted-by":"crossref","DOI":"10.1609\/aaai.v33i01.33019989","article-title":"The level weighted structural similarity loss: a step away from the MSE","author":"Lu","year":"2019"},{"key":"r49","doi-asserted-by":"publisher","DOI":"10.3389\/fninf.2023.1130845"},{"key":"r50","article-title":"How to represent part-whole hierarchies in a neural network","author":"Hinton","year":"2021"},{"key":"r51","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-021-26216-9"}],"container-title":["Journal of Medical Imaging"],"original-title":[],"link":[{"URL":"https:\/\/www.spiedigitallibrary.org\/journals\/journal-of-medical-imaging\/volume-11\/issue-01\/014003\/Attention-gated-3D-CapsNet-for-robust-hippocampal-segmentation\/10.1117\/1.JMI.11.1.014003.full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,24]],"date-time":"2026-09-24T00:18:31Z","timestamp":1790209111000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.spiedigitallibrary.org\/journals\/journal-of-medical-imaging\/volume-11\/issue-01\/014003\/Attention-gated-3D-CapsNet-for-robust-hippocampal-segmentation\/10.1117\/1.JMI.11.1.014003.full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,1,2]]},"references-count":51,"journal-issue":{"issue":"01","published-online":{"date-parts":[[2024,1,1]]}},"URL":"https:\/\/doi.org\/10.1117\/1.jmi.11.1.014003","relation":{},"ISSN":["2329-4302","2329-4310"],"issn-type":[{"value":"2329-4302","type":"print"},{"value":"2329-4310","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,1,2]]}}}