Please use this identifier to cite or link to this item: https://doi.org/10.21256/zhaw-25502
Full metadata record
DC FieldValueLanguage
dc.contributor.authorSager, Pascal-
dc.contributor.authorSalzmann, Sebastian-
dc.contributor.authorBurn, Felice-
dc.contributor.authorStadelmann, Thilo-
dc.date.accessioned2022-08-25T08:24:31Z-
dc.date.available2022-08-25T08:24:31Z-
dc.date.issued2022-08-19-
dc.identifier.issn2313-433Xde_CH
dc.identifier.urihttps://digitalcollection.zhaw.ch/handle/11475/25502-
dc.description.abstractA variety of medical computer vision applications analyze 2D slices of computed tomography (CT) scans, whereas axial slices from the body trunk region are usually identified based on their relative position to the spine. A limitation of such systems is that either the correct slices must be extracted manually or labels of the vertebrae are required for each CT scan to develop an automated extraction system. In this paper, we propose an unsupervised domain adaptation (UDA) approach for vertebrae detection and identification based on a novel Domain Sanity Loss (DSL) function. With UDA the model’s knowledge learned on a publicly available (source) data set can be transferred to the target domain without using target labels, where the target domain is defined by the specific setup (CT modality, study protocols, applied pre- and processing) at the point of use (e.g., a specific clinic with its specific CT study protocols). With our approach, a model is trained on the source and target data set in parallel. The model optimizes a supervised loss for labeled samples from the source domain and the DSL loss function based on domain-specific “sanity checks” for samples from the unlabeled target domain. Without using labels from the target domain, we are able to identify vertebra centroids with an accuracy of 72.8%. By adding only ten target labels during training the accuracy increases to 89.2%, which is on par with the current state-of-the-art for full supervised learning, while using about 20 times less labels. Thus, our model can be used to extract 2D slices from 3D CT scans on arbitrary data sets fully automatically without requiring an extensive labeling effort, contributing to the clinical adoption of medical imaging by hospitals.de_CH
dc.language.isoende_CH
dc.publisherMDPIde_CH
dc.relation.ispartofJournal of Imagingde_CH
dc.rightshttp://creativecommons.org/licenses/by/4.0/de_CH
dc.subjectUnsupervised domain adaptationde_CH
dc.subjectSemi-supervised learningde_CH
dc.subjectVertebrae detectionde_CH
dc.subjectVertebrae identificationde_CH
dc.subjectTransfer learningde_CH
dc.subjectSemantic segmentationde_CH
dc.subjectData centrismde_CH
dc.subjectDeep learningde_CH
dc.subject.ddc006: Spezielle Computerverfahrende_CH
dc.subject.ddc616: Innere Medizin und Krankheitende_CH
dc.titleUnsupervised domain adaptation for vertebrae detection and identification in 3D CT volumes using a domain sanity lossde_CH
dc.typeBeitrag in wissenschaftlicher Zeitschriftde_CH
dcterms.typeTextde_CH
zhaw.departementSchool of Engineeringde_CH
zhaw.organisationalunitCentre for Artificial Intelligence (CAI)de_CH
dc.identifier.doi10.3390/jimaging8080222de_CH
dc.identifier.doi10.21256/zhaw-25502-
zhaw.funding.euNode_CH
zhaw.issue8de_CH
zhaw.originated.zhawYesde_CH
zhaw.pages.start222de_CH
zhaw.publication.statuspublishedVersionde_CH
zhaw.volume8de_CH
zhaw.publication.reviewPeer review (Publikation)de_CH
zhaw.webfeedComputer Vision, Perception and Cognitionde_CH
zhaw.webfeedDatalabde_CH
zhaw.webfeedDigital Health Labde_CH
zhaw.webfeedZHAW digitalde_CH
zhaw.author.additionalNode_CH
zhaw.display.portraitYesde_CH
Appears in collections:Publikationen School of Engineering

Files in This Item:
File Description SizeFormat 
2022_Sager-etal_Unsupervised-domain-adaptation-vertebrae-detection-3D-CT.pdf32.91 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.