2024
Prior knowledge-guided vision-transformer-based unsupervised domain adaptation for intubation prediction in lung disease at one week
Yang J, Henao J, Dvornek N, He J, Bower D, Depotter A, Bajercius H, de Mortanges A, You C, Gange C, Ledda R, Silva M, Dela Cruz C, Hautz W, Bonel H, Reyes M, Staib L, Poellinger A, Duncan J. Prior knowledge-guided vision-transformer-based unsupervised domain adaptation for intubation prediction in lung disease at one week. Computerized Medical Imaging And Graphics 2024, 118: 102442. PMID: 39515190, DOI: 10.1016/j.compmedimag.2024.102442.Peer-Reviewed Original ResearchUnsupervised domain adaptationSpatial prior informationDomain adaptationLabeled dataData-driven approachUnsupervised domain adaptation modelMedical image analysis tasksImage analysis tasksTransformer-based modelsMedical image analysisPrior informationOutcome prediction tasksAdversarial trainingDistribution alignmentDomain shiftAttention headsClass tokenPoor generalizationAnalysis tasksTarget domainPrediction taskData distributionKnowledge-guidedLocal weightsMedical images
2023
Unsupervised Domain Adaptation by Cross-Prototype Contrastive Learning for Medical Image Segmentation
Cai Z, Xin J, Dong S, You C, Shi P, Zeng T, Zhang J, Onofrey J, Zheng N, Duncan J. Unsupervised Domain Adaptation by Cross-Prototype Contrastive Learning for Medical Image Segmentation. 2023, 00: 819-824. DOI: 10.1109/bibm58861.2023.10386055.Peer-Reviewed Original ResearchUnsupervised domain adaptationDistribution alignmentDomain adaptationContrastive learningUnsupervised domain adaptation methodsMedical image segmentation tasksDomain distribution alignmentGlobal distribution alignmentContrastive learning methodDomain adaptation performanceIntra-class distancePixel-level featuresImage segmentation tasksInter-class distancePublic cardiac datasetsCategory centroidDiscrimination of classesClass prototypesSegmentation taskSource domainTarget domainCardiac datasetsLearning methodsGlobal prototypesCentroid alignment
2022
Incremental Learning Meets Transfer Learning: Application to Multi-site Prostate MRI Segmentation
You C, Xiang J, Su K, Zhang X, Dong S, Onofrey J, Staib L, Duncan J. Incremental Learning Meets Transfer Learning: Application to Multi-site Prostate MRI Segmentation. Lecture Notes In Computer Science 2022, 13573: 3-16. PMID: 37415747, PMCID: PMC10323962, DOI: 10.1007/978-3-031-18523-6_1.Peer-Reviewed Original ResearchIncremental learningMedical image segmentation tasksMulti-site datasetImage segmentation tasksMedical image segmentationProstate MRI SegmentationComputation resourcesMedical datasetsSegmentation taskImage segmentationSegmentation frameworkEmbedding featuresBenchmark datasetsMRI segmentationTraining dataTarget domainLearning approachPractical deploymentDomain-specific expertiseCompetitive performanceDatasetTraining schemePrior workSegmentationSingle model
2019
Unsupervised Domain Adaptation via Disentangled Representations: Application to Cross-Modality Liver Segmentation
Yang J, Dvornek NC, Zhang F, Chapiro J, Lin M, Duncan JS. Unsupervised Domain Adaptation via Disentangled Representations: Application to Cross-Modality Liver Segmentation. Lecture Notes In Computer Science 2019, 11765: 255-263. PMID: 32377643, PMCID: PMC7202929, DOI: 10.1007/978-3-030-32245-8_29.Peer-Reviewed Original ResearchDice similarity coefficientDomain adaptationContent spaceDomain shiftTarget domainCross-modality domain adaptationUnsupervised domain adaptation methodsDiverse image generationLiver segmentation taskDeep learning modelsDifferent target domainUnlabeled target dataFeature-level informationUnsupervised domain adaptationDomain adaptation methodsMulti-phasic MRISegmentation taskSegmentation performanceSegmentation modelImage generationLiver segmentationStyle transferDisentangled representationsBetter generalizationSource domainDomain-Agnostic Learning with Anatomy-Consistent Embedding for Cross-Modality Liver Segmentation
Yang J, Dvornek NC, Zhang F, Zhuang J, Chapiro J, Lin M, Duncan JS. Domain-Agnostic Learning with Anatomy-Consistent Embedding for Cross-Modality Liver Segmentation. ICCV Workshops 2019, 00: 323-331. PMID: 34676308, PMCID: PMC8528125, DOI: 10.1109/iccvw.2019.00043.Peer-Reviewed Original ResearchDomain adaptationDisentangled representationsLiver segmentationTarget domainSource domainDeep learning modelsGenerative adversarial networkHuman interpretabilityLearning frameworkAdversarial networkDownstream tasksArt methodsSegmentation consistencyLearning modelAgnostic learningMeaningful representationCycleGANNew tasksAblation analysisDA taskDifferent modalitiesTaskSegmentationEmbeddingLearning