2025
Cybersecurity Threats and Mitigation Strategies for Large Language Models in Health Care.
Akinci D'Antonoli T, Tejani A, Khosravi B, Bluethgen C, Busch F, Bressem K, Adams L, Moassefi M, Faghani S, Gichoya J. Cybersecurity Threats and Mitigation Strategies for Large Language Models in Health Care. Radiology Artificial Intelligence 2025, 7: e240739. PMID: 40366259, DOI: 10.1148/ryai.240739.Peer-Reviewed Original ResearchConceptsMalicious actorsLanguage modelArtificial intelligenceRobust security measuresSensitive patient informationProtect patient privacyMalicious attacksUnauthorized manipulationCybersecurity challengesPrivacy breachesCybersecurity threatsCybersecurity risksTraining dataSecurity measuresDeployment stageAI modelsPatient privacyPoisoning dataCybersecurityPrivacyHealth carePatient informationPatient dataImprove medical practiceLanguageEfficient few-shot medical image segmentation via self-supervised variational autoencoder
Zhou Y, Zhou F, Xi F, Liu Y, Peng Y, Carlson D, Tu L. Efficient few-shot medical image segmentation via self-supervised variational autoencoder. Medical Image Analysis 2025, 104: 103637. PMID: 40449308, DOI: 10.1016/j.media.2025.103637.Peer-Reviewed Original ResearchConceptsFew-shot medical image segmentationMedical image segmentationUnlabeled imagesVariational autoencoderImage segmentationMulti-modality medical image datasetEnd-to-end modelDice scoreFully-supervised methodsMedical image datasetsSelf-supervised learningImproving feature extractionEnd-to-endSecond-best methodSegmentation taskFeature extractionImage datasetsData augmentationSource codePrevent overfittingTraining dataReconstruction taskStructural priorsSegmentation qualityLabeled atlasesA Current Review of Generative AI in Medicine: Core Concepts, Applications, and Current Limitations
Rouzrokh P, Khosravi B, Faghani S, Moassefi M, Shariatnia M, Rouzrokh P, Erickson B. A Current Review of Generative AI in Medicine: Core Concepts, Applications, and Current Limitations. Current Reviews In Musculoskeletal Medicine 2025, 18: 246-266. PMID: 40304941, PMCID: PMC12185825, DOI: 10.1007/s12178-025-09961-y.Peer-Reviewed Original ResearchGenerative AIArtificial intelligenceAI modelsSynthetic medical imagesEnhance information retrievalInformation retrievalGenerative artificial intelligenceAI agentsLanguage modelTraining dataModel reasoningComplex workflowsMedical imagesSynthetic dataData typesDecision supportMultiple data typesDiscriminant modelMultimodal modelClinical documentationMedical fieldModel familyEnhanced capabilitiesSpecialized applicationsCore conceptsReporting of Fairness Metrics in Clinical Risk Prediction Models Used for Precision Health: Scoping Review
Rountree L, Lin Y, Liu C, Salvatore M, Admon A, Nallamothu B, Singh K, Basu A, Bu F, Mukherjee B. Reporting of Fairness Metrics in Clinical Risk Prediction Models Used for Precision Health: Scoping Review. Online Journal Of Public Health Informatics 2025, 17: e66598. PMID: 39962044, PMCID: PMC11966066, DOI: 10.2196/66598.Peer-Reviewed Original ResearchClinical risk prediction modelsRisk prediction modelFairness metricsSex-stratified modelsEthnicity dataPrecision healthClinical risk predictionSensitive featuresStudy populationCardiovascular diseaseRisk predictionEvaluate potential disparitiesTraining dataPotential disparitiesEmpirical evaluationPrediction modelPrimary preventionInformatics systemsHigh-impact publicationsCOVID-19Metrics usageStudy cohortGoogle ScholarImplementation frameworkCOVID-19 modelBroken time-reversal symmetry in visual motion detection
Wu N, Zhou B, Agrochao M, Clark D. Broken time-reversal symmetry in visual motion detection. Proceedings Of The National Academy Of Sciences Of The United States Of America 2025, 122: e2410768122. PMID: 40048271, PMCID: PMC11912477, DOI: 10.1073/pnas.2410768122.Peer-Reviewed Original ResearchConceptsNeural network modelTraining dataNetwork modelTrained neural network modelFlexible neural networkBiological motion detectorsMotion estimationTime-reversal symmetryNeural networkMotion detectionVisual motion detectionMotion detectorsVisual systemSymmetry breakingTime-reversal symmetry breakingContrast distributionPerception of motionReversal symmetryTrainingMovieMotion perceptionSceneTime reversalIntuitionNetworkCT-Less Whole-Body Bone Segmentation of PET Images Using a Multimodal Deep Learning Network
Bao N, Zhang J, Li Z, Wei S, Zhang J, Greenwald S, Onofrey J, Lu Y, Xu L. CT-Less Whole-Body Bone Segmentation of PET Images Using a Multimodal Deep Learning Network. IEEE Journal Of Biomedical And Health Informatics 2025, 29: 1151-1164. PMID: 40030243, DOI: 10.1109/jbhi.2024.3501386.Peer-Reviewed Original ResearchPositron emission tomographyMultimodal fusion modulePositron emission tomography imagingMultimodal fusion networkAttenuation mapDice similarity coefficientFusion moduleFusion networkEncoder RepresentationsEncoder branchesCT imagesTraining dataComputed tomographyTumor analysisTracer activityModality imagesMultimodal deep learning networkPET imagingBone segmentsSqueeze-and-excitationBone cancerPositron emission tomography informationCT-based approachDeep learning networkImprove segmentation performanceDICOM LUT is a Key Step in Medical Image Preprocessing Towards AI Generalizability
Dapamede T, Li F, Khosravi B, Purkayastha S, Trivedi H, Gichoya J. DICOM LUT is a Key Step in Medical Image Preprocessing Towards AI Generalizability. Journal Of Imaging Informatics In Medicine 2025, 1-9. PMID: 39890738, DOI: 10.1007/s10278-025-01418-5.Peer-Reviewed Original ResearchDeep learning modelsHistogram equalizationInformation lossPreprocessing techniquesPre-processingLearning modelsPerformance of deep learning modelsMachine learning practitionersRisk of information lossDeep learning classifierImage preprocessing techniquesImprove model robustnessImage pre-processingTraining dataCXR datasetPreprocessed informationLearning classifiersDataset curationLearning practitionersModel performancePotential overfittingDatasetModel robustnessPreprocessingSharing dataset
2024
Self-supervised Pre-training Tasks for an fMRI Time-Series Transformer in Autism Detection
Zhou Y, Duan P, Du Y, Dvornek N. Self-supervised Pre-training Tasks for an fMRI Time-Series Transformer in Autism Detection. Lecture Notes In Computer Science 2024, 15266: 145-154. PMID: 40160559, PMCID: PMC11951341, DOI: 10.1007/978-3-031-78761-4_14.Peer-Reviewed Original ResearchSelf-supervised pre-training tasksPre-training tasksFunctional magnetic resonance imagingPre-training stepTransformer-based modelsTime-series fMRI dataTraining data availabilityAutism spectrum disorderTime series transformationsTransformation modelMachine learning methodsFunctional magnetic resonance imaging time-series dataClassification taskPublic datasetsTraining dataOver-fittingComputed functional connectivityLearning methodsModel performanceMasking strategyAutism detectionCross-validationTaskDatasetAverage improvementLessons learned from the IMMREP23 TCR-epitope prediction challenge
Nielsen M, Eugster A, Jensen M, Goel M, Tiffeau-Mayer A, Pelissier A, Valkiers S, Martínez M, Meynard-Piganeeau B, Greiff V, Mora T, Walczak A, Croce G, Moreno D, Gfeller D, Meysman P, Barton J. Lessons learned from the IMMREP23 TCR-epitope prediction challenge. ImmunoInformatics 2024, 16: 100045. DOI: 10.1016/j.immuno.2024.100045.Peer-Reviewed Original ResearchData leakageIssue of data leakageT cell receptorPerformance of proposed methodsPrediction challengeTraining dataBenchmarking CompetitionPMHC targetsRandom performanceTCR-pMHC interactionsInteraction of T-cell receptorsParticipating teamsCellular immune systemData setsTest dataImmune systemPerformanceNarrow spaceEfficient deep learning-based automated diagnosis from echocardiography with contrastive self-supervised learning
Holste G, Oikonomou E, Mortazavi B, Wang Z, Khera R. Efficient deep learning-based automated diagnosis from echocardiography with contrastive self-supervised learning. Communications Medicine 2024, 4: 133. PMID: 38971887, PMCID: PMC11227494, DOI: 10.1038/s43856-024-00538-3.Peer-Reviewed Original ResearchSelf-supervised learningTransfer learningTraining dataEchocardiogram videosPortion of labelled dataStandard transfer learning approachContrastive self-supervised learningSelf-supervised learning approachLearning approachImage recognition tasksState-of-the-artContrastive learning approachFine-tuningTransfer learning approachMedical image diagnosisCardiac disease diagnosisContrastive learningVideo framesLabeled datasetLabeled dataExpert labelsClassification performanceMedical imagesRecognition taskVideoEvaluating Augmentation Approaches for Deep Learning-based Major Depressive Disorder Diagnosis with Raw Electroencephalogram Data*
Ellis C, Miller R, Calhoun V. Evaluating Augmentation Approaches for Deep Learning-based Major Depressive Disorder Diagnosis with Raw Electroencephalogram Data*. Annual International Conference Of The IEEE Engineering In Medicine And Biology Society (EMBC) 2024, 00: 1-5. PMID: 40039441, DOI: 10.1109/embc53108.2024.10782103.Peer-Reviewed Original ResearchConceptsAugmented training setData augmentationTraining setDA methodsDeep learning methodsDA approachNeuropsychiatric disorder diagnosisModel performanceTraining dataDeep learningEEG datasetDataset sizeLearning methodsAugmentation approachImprove model performanceDepressive disorder diagnosisDA efficacyDatasetDisorder diagnosisCompare performanceMajor depressive disorder diagnosisPerformanceBaseline setDeepChannelCalibrating Multi-modal Representations: A Pursuit of Group Robustness without Annotations
You C, Min Y, Dai W, Sekhon J, Staib L, Duncan J. Calibrating Multi-modal Representations: A Pursuit of Group Robustness without Annotations. 2015 IEEE Conference On Computer Vision And Pattern Recognition (CVPR) 2024, 00: 26140-26150. PMID: 39640960, PMCID: PMC11620289, DOI: 10.1109/cvpr52733.2024.02470.Peer-Reviewed Original ResearchDiverse downstream tasksVision-language modelsPre-trained modelsRepresentation of samplesContrastive learningDownstream tasksFeature reweightingTraining dataFeature patternsModel generalizationGroup annotationsPain pointsGroup labelsAnnotationRobustnessClassifierClipsFeaturesDeepDeploymentBenchmarksTime-intensiveCodeTaskLearningComb EMI: a hardware-free, training-free approach to EMI correction
Sun H, Sun C, Ha Y, Samardzija A, Gross R, Galiana G, Constable R. Comb EMI: a hardware-free, training-free approach to EMI correction. Proceedings Of The International Society For Magnetic Resonance In Medicine ... Scientific Meeting And Exhibition. 2024 DOI: 10.58530/2024/2691.Peer-Reviewed Original ResearchElectromagnetic interferenceTraining-freeNoise-dominated regionTraining dataDesign complexityExternal hardwareHardware-freeExtra hardwareHardwareDesign possibilitiesPoint-of-careExperimental dataMRI systemWhite noiseCancellationImprove costSparsitySampling windowSNRDesignTrainingPassive shieldingPortabilityCostMethodExploring Backdoor Attacks in Off-the-Shelf Unsupervised Domain Adaptation for Securing Cardiac MRI-Based Diagnosis
Liu X, Xing F, Gaggin H, Kuo C, El Fakhri G, Woo J. Exploring Backdoor Attacks in Off-the-Shelf Unsupervised Domain Adaptation for Securing Cardiac MRI-Based Diagnosis. 2011 IEEE International Symposium On Biomedical Imaging: From Nano To Macro 2024, 00: 1-5. PMID: 39421190, PMCID: PMC11483644, DOI: 10.1109/isbi56570.2024.10635403.Peer-Reviewed Original ResearchUnsupervised domain adaptationTarget domain modelBackdoor attacksDomain adaptationTraining dataLabeled source domain dataSusceptible to backdoor attacksAccurate pseudo labelsDomain modelSource domain dataPatient data privacyTarget training dataOff-the-shelfPseudo-labelsData privacySource domainMulti-vendorRandom initializationTraining phaseDomain dataDiagnosis modelTarget modelMulti-diseaseAttacksAuxiliary modelImproving large language models for clinical named entity recognition via prompt engineering
Hu Y, Chen Q, Du J, Peng X, Keloth V, Zuo X, Zhou Y, Li Z, Jiang X, Lu Z, Roberts K, Xu H. Improving large language models for clinical named entity recognition via prompt engineering. Journal Of The American Medical Informatics Association 2024, 31: 1812-1820. PMID: 38281112, PMCID: PMC11339492, DOI: 10.1093/jamia/ocad259.Peer-Reviewed Original ResearchClinical NER tasksNER taskTask-specific promptsEntity recognitionLanguage modelTraining samplesState-of-the-art modelsFew-shot learningState-of-the-artMinimal training dataTask-specific knowledgeF1-socreAnnotated samplesConcept extractionModel performanceAnnotated datasetsTraining dataF1 scoreTask descriptionFormat specificationsComplex clinical dataOptimal performanceTaskEvaluation schemaGPT modelBiometric contrastive learning for data-efficient deep learning from electrocardiographic images
Sangha V, Khunte A, Holste G, Mortazavi B, Wang Z, Oikonomou E, Khera R. Biometric contrastive learning for data-efficient deep learning from electrocardiographic images. Journal Of The American Medical Informatics Association 2024, 31: 855-865. PMID: 38269618, DOI: 10.1093/jamia/ocae002.Peer-Reviewed Original ResearchLabeled training dataContrastive learningECG imagesLabeled dataTraining dataDeep learningProportions of labeled dataArtificial intelligenceSelf-supervised contrastive learningTraditional supervised learningConvolutional neural networkHeld-out test setSupervised learningPretraining strategyBiometric signatureImageNet initializationPretraining approachNeural networkImageNetAI modelsImage objectsTest setLearningDetect atrial fibrillationEquivalent performanceEMPATHIC: Emulating Human-Like Multimodal Personality Architecture Through Thoughtful Human-AI Conversation
Devi V, Oviya I, Raja K. EMPATHIC: Emulating Human-Like Multimodal Personality Architecture Through Thoughtful Human-AI Conversation. 2024, 00: 79-85. DOI: 10.1109/confluence60223.2024.10463330.Peer-Reviewed Original ResearchHigh-quality training dataReal-time faceHuman-computer interactionNatural language processingChatbot systemGeneral queriesConversational agentsTraining dataObject countingHuman instructionsEmotion recognitionLow-rankLanguage processingDialogue performanceChatbotFine-tuningInstruction templatesLanguage dataPersonality architectureFrameworkCustomer service experienceOperational efficiencyTuning methodologyLoRaQueryBias and Fairness in Chatbots: An Overview
Xue J, Wang Y, Wei C, Liu X, Woo J, Kuo C. Bias and Fairness in Chatbots: An Overview. APSIPA Transactions On Signal And Information Processing 2024, 13: e102. DOI: 10.1561/116.00000064.Peer-Reviewed Original Research
2023
Using annotation for computerized support for fast skimming of cardiology electronic health record notes
Dehkordi M, Einstein A, Zhou S, Elhanan G, Perl Y, Keloth V, Geller J, Liu H. Using annotation for computerized support for fast skimming of cardiology electronic health record notes. 2023, 00: 4043-4050. DOI: 10.1109/bibm58861.2023.10385289.Peer-Reviewed Original ResearchElectronic health recordsElectronic health record notesNamed Entity RecognitionTraining dataMining conceptsInterface terminologyMachine learningSNOMED CTEntity recognitionHealth recordsHealthcare professionalsTransfer learningPatient careCurrent healthcareMining techniquesMining phrasesArt techniquesRecord notesSNOMED conceptsMedical specialtiesComputer-SupportedMedical professionalsReference terminologyCritical informationAnnotationTeacher’s PET: Semi-supervised Deep Learning for PET Head Motion Correction
Zeng T, You C, Cai Z, Lieffrig E, Zhang J, Chen F, Lu Y, Onofrey J. Teacher’s PET: Semi-supervised Deep Learning for PET Head Motion Correction. 2023, 00: 1-1. DOI: 10.1109/nssmicrtsd49126.2023.10337834.Peer-Reviewed Original ResearchMotion tracking methodHead motion correctionMotion trackingExtra hardwareMotion estimatesTracking methodSemi-supervised deep learningSupervised deep learning methodsQuality training dataDeep learning methodsMean teacher modelSemi-supervised mannerMotion correctionMotion detectionHead motionCorrection networkDeep learningInaccurate quantitative resultsTraining dataLearning methodsBetter generalizationMotionLow resolutionCorrection resultsPerformance
This site is protected by hCaptcha and its Privacy Policy and Terms of Service apply