2024
Dose-aware Diffusion Model for 3D Low-count Cardiac SPECT Image Denoising with Projection-domain Consistency
Xie H, Gan W, Chen X, Zhou B, Liu Q, Xia M, Guo X, Liu Y, An H, Kamilov U, Wang G, Sinusas A, Liu C. Dose-aware Diffusion Model for 3D Low-count Cardiac SPECT Image Denoising with Projection-domain Consistency. 2024, 00: 1-1. DOI: 10.1109/nss/mic/rtsd57108.2024.10655170.Peer-Reviewed Original ResearchImage denoisingImage denoising performanceDeep learning techniquesNoise-levelDenoising performanceDenoising resultsNeural networkLearning techniquesSPECT imagesLow count levelsSPECT scansDenoisingSampling stepIterative reconstructionNoise amplitudeImagesInjected dosePatient studiesDiffusion modelRadiation exposureCardiology studiesSPECTNetworkStochastic natureMLEMDuDoCFNet: Dual-Domain Coarse-to-Fine Progressive Network for Simultaneous Denoising, Limited-View Reconstruction, and Attenuation Correction of Cardiac SPECT
Chen X, Zhou B, Guo X, Xie H, Liu Q, Duncan J, Sinusas A, Liu C. DuDoCFNet: Dual-Domain Coarse-to-Fine Progressive Network for Simultaneous Denoising, Limited-View Reconstruction, and Attenuation Correction of Cardiac SPECT. IEEE Transactions On Medical Imaging 2024, 43: 3110-3125. PMID: 38578853, PMCID: PMC11539864, DOI: 10.1109/tmi.2024.3385650.Peer-Reviewed Original ResearchMulti-task learning methodCross-domainLimited-viewLearning methodsCoarse-to-fine estimationProgressive networkDual domainCross-modal feature fusionDual-domain networkProgressive learning strategyCross-modal informationSimultaneous denoisingFeature fusionSingle-photon emission computed tomographyImage domainCardiac single-photon emission computed tomographyReconstruction accuracyDenoisingHardware expenseFusion mechanismAccelerated scansImage noiseM-mapSuperior accuracyNetworkTAI-GAN: A Temporally and Anatomically Informed Generative Adversarial Network for early-to-late frame conversion in dynamic cardiac PET inter-frame motion correction
Guo X, Shi L, Chen X, Liu Q, Zhou B, Xie H, Liu Y, Palyo R, Miller E, Sinusas A, Staib L, Spottiswoode B, Liu C, Dvornek N. TAI-GAN: A Temporally and Anatomically Informed Generative Adversarial Network for early-to-late frame conversion in dynamic cardiac PET inter-frame motion correction. Medical Image Analysis 2024, 96: 103190. PMID: 38820677, PMCID: PMC11180595, DOI: 10.1016/j.media.2024.103190.Peer-Reviewed Original ResearchGenerative adversarial networkAdversarial networkMotion estimation accuracyInter-frame motionIntensity-based image registration techniqueAll-to-oneSegmentation masksImage registration techniquesOriginal frameTemporal informationDiagnosis accuracyMyocardial blood flowEstimation accuracyFrame conversionPositron emission tomographyNovel methodImage qualityPET datasetsRegistration techniqueNetworkCardiac positron emission tomographyBlood flowDynamic cardiac positron emission tomographyMotion correctionCoronary artery disease
2023
Unified Noise-Aware Network for Low-Count PET Denoising With Varying Count Levels
Xie H, Liu Q, Zhou B, Chen X, Guo X, Wang H, Li B, Rominger A, Shi K, Liu C. Unified Noise-Aware Network for Low-Count PET Denoising With Varying Count Levels. IEEE Transactions On Radiation And Plasma Medical Sciences 2023, 8: 366-378. PMID: 39391291, PMCID: PMC11463975, DOI: 10.1109/trpms.2023.3334105.Peer-Reviewed Original ResearchLarge-scale dataDeep learningDynamic PET imagesLow-count dataNeural networkMultiple networksSpecific noise levelDifferent vendorsDifferent noise levelsDenoised resultsNoisy counterpartDynamic frameInput noise levelNetworkData availabilityHigher image noiseImage qualityImage noiseSuperior performanceImportant topicAdditional challengesNoise levelPET imagesLimited data availabilityVendorsDual-Domain Iterative Network with Adaptive Data Consistency for Joint Denoising and Few-Angle Reconstruction of Low-Dose Cardiac SPECT
Chen X, Zhou B, Xie H, Guo X, Liu Q, Sinusas A, Liu C. Dual-Domain Iterative Network with Adaptive Data Consistency for Joint Denoising and Few-Angle Reconstruction of Low-Dose Cardiac SPECT. Lecture Notes In Computer Science 2023, 14307: 49-59. DOI: 10.1007/978-3-031-44917-8_5.Peer-Reviewed Original ResearchIterative networkAuxiliary modulesJoint denoisingLow reconstruction accuracySource codeData consistencyNetwork performanceAblation studiesReconstruction accuracyCardiac SPECTConsistency moduleHardware expensePrediction accuracyAngle reconstructionNetworkDenoisingImage noiseAngle projectionsModuleADC moduleAccuracyReconstructionImagesMPI dataCodeFedFTN: Personalized federated learning with deep feature transformation network for multi-institutional low-count PET denoising
Zhou B, Xie H, Liu Q, Chen X, Guo X, Feng Z, Hou J, Zhou S, Li B, Rominger A, Shi K, Duncan J, Liu C. FedFTN: Personalized federated learning with deep feature transformation network for multi-institutional low-count PET denoising. Medical Image Analysis 2023, 90: 102993. PMID: 37827110, PMCID: PMC10611438, DOI: 10.1016/j.media.2023.102993.Peer-Reviewed Original ResearchConceptsFederated learning processFederated learning algorithmFederated learning strategyLarge domain shiftDifferent data distributionsTransformation networkLarge-scale datasetsDeep learningDomain shiftLearning algorithmDownstream tasksNetwork weightsFeature outputFeature transformationSecurity concernsData distributionCollaborative trainingPersonalized modelPET image qualityReconstructed imagesReconstruction methodImage qualityNetworkEfficient wayLocal dataTransformer-Based Dual-Domain Network for Few-View Dedicated Cardiac SPECT Image Reconstructions
Xie H, Zhou B, Chen X, Guo X, Thorn S, Liu Y, Wang G, Sinusas A, Liu C. Transformer-Based Dual-Domain Network for Few-View Dedicated Cardiac SPECT Image Reconstructions. Lecture Notes In Computer Science 2023, 14229: 163-172. DOI: 10.1007/978-3-031-43999-5_16.Peer-Reviewed Original ResearchDual-domain networkSPECT image reconstructionImage reconstructionDeep learning methodsPrevious baseline methodsCardiac SPECT imagesHigh-quality imagesReconstruction networkIterative reconstruction processView reconstructionBaseline methodsReconstruction outputLearning methodsClinical softwareReconstruction processImaging problemsProjection dataImage qualityNetworkImagesStationary dataSPECT scannerDiagnosis of CVDLimited amountSoftwareSegmentation-Free PVC for Cardiac SPECT Using a Densely-Connected Multi-Dimensional Dynamic Network
Xie H, Liu Z, Shi L, Greco K, Chen X, Zhou B, Feher A, Stendahl J, Boutagy N, Kyriakides T, Wang G, Sinusas A, Liu C. Segmentation-Free PVC for Cardiac SPECT Using a Densely-Connected Multi-Dimensional Dynamic Network. IEEE Transactions On Medical Imaging 2023, 42: 1325-1336. PMID: 36459599, PMCID: PMC10204821, DOI: 10.1109/tmi.2022.3226604.Peer-Reviewed Original ResearchDSFormer: A Dual-domain Self-supervised Transformer for Accelerated Multi-contrast MRI Reconstruction
Zhou B, Dey N, Schlemper J, Salehi S, Liu C, Duncan J, Sofka M. DSFormer: A Dual-domain Self-supervised Transformer for Accelerated Multi-contrast MRI Reconstruction. 2023, 00: 4955-4964. DOI: 10.1109/wacv56688.2023.00494.Peer-Reviewed Original ResearchReconstruction networkSelf-supervised learning strategyHigh-fidelity reconstructionConvolutional architectureMRI reconstructionTraining dataFull supervisionInformation sharingTime costMulti-contrast MRINetworkLearning strategiesMultiple acquisitionsArchitectureSharingFine anatomical detailsRedundancyCostMultiple contrastsMRI sequencesReconstructionComplementary imaging modalities
2022
An Adaptive Patch Sampling Scheme for Deep Learning Based PET Image Denoising
Wu J, Tan H, Liu H, Liu C, Onofrey J. An Adaptive Patch Sampling Scheme for Deep Learning Based PET Image Denoising. 2022, 00: 1-3. DOI: 10.1109/nss/mic44845.2022.10399313.Peer-Reviewed Original ResearchOver-smoothing effectSignal-to-noise ratioU-NetImage denoisingDeep learning-based approachPET image denoisingL1 loss functionPatch-based strategyLearning-based approachSampling schemeMean square errorHigh signal-to-noise ratioDenoising performanceLow-dose PET imagesNetwork trainingWeight mapData augmentationDenoisingLoss functionNetworkImage noiseSquare errorPatch samplesSampling rateSchemeDual-Branch Squeeze-Fusion-Excitation Module for Cross-Modality Registration of Cardiac SPECT and CT
Chen X, Zhou B, Xie H, Guo X, Zhang J, Sinusas A, Onofrey J, Liu C. Dual-Branch Squeeze-Fusion-Excitation Module for Cross-Modality Registration of Cardiac SPECT and CT. Lecture Notes In Computer Science 2022, 13436: 46-55. DOI: 10.1007/978-3-031-16446-0_5.Peer-Reviewed Original ResearchConvolutional neural networkCross-modality registrationFeature fusionPrevious convolutional neural networkEarly feature fusionCross-modality informationMultiple convolutional layersMedical image registrationLow registration errorCardiac SPECTConvolutional layersCNN moduleImage featuresLate fusionSource codeNeural networkExcitation moduleInput modalitiesImage registrationSpatial featuresMultiple modalitiesRegistration errorPrevious methodsRigid registrationNetworkDeep-Learning-Based Few-Angle Cardiac SPECT Reconstruction Using Transformer
Xie H, Thorn S, Liu Y, Lee S, Liu Z, Wang G, Sinusas A, Liu C. Deep-Learning-Based Few-Angle Cardiac SPECT Reconstruction Using Transformer. IEEE Transactions On Radiation And Plasma Medical Sciences 2022, 7: 33-40. PMID: 37397179, PMCID: PMC10312390, DOI: 10.1109/trpms.2022.3187595.Peer-Reviewed Original ResearchConvolutional neural networkLimitations of CNNMedical imaging tasksDeep U-NetImage reconstruction taskCardiac SPECT imagesComputer visionVision TransformerConvolutional kernelsTransformer networkAttention blockInput imageU-NetNeural networkMemory burdenImage sizeInductive biasInformative featuresImage volumesImaging tasksTesting dataNetworkWhole 3D volumeNetwork structureCardiac single photon emissionDeep-learning-based methods of attenuation correction for SPECT and PET
Chen X, Liu C. Deep-learning-based methods of attenuation correction for SPECT and PET. Journal Of Nuclear Cardiology 2022, 30: 1859-1878. PMID: 35680755, DOI: 10.1007/s12350-022-03007-3.Peer-Reviewed Original ResearchConceptsHigh computational complexityAC strategyNeural networkRaw emission dataComputational complexityLearning methodsCT imagesΜ-mapsPET imagesLow accuracySuperior performanceImagesAttenuation correctionPromising resultsMR imagesAttenuation mapPET/CT scannerHigh noise levelsArtifactsNetworkCT artifactsPET/MRI scannerIntermediate stepComplexityScanner
2021
Automatic Inter-Frame Patient Motion Correction for Dynamic Cardiac PET Using Deep Learning
Shi L, Lu Y, Dvornek N, Weyman CA, Miller EJ, Sinusas AJ, Liu C. Automatic Inter-Frame Patient Motion Correction for Dynamic Cardiac PET Using Deep Learning. IEEE Transactions On Medical Imaging 2021, 40: 3293-3304. PMID: 34018932, PMCID: PMC8670362, DOI: 10.1109/tmi.2021.3082578.Peer-Reviewed Original ResearchConceptsConvolutional neural networkRegistration-based methodMotion correctionDynamic frameTracer distribution changeDynamic image dataPatient motion correctionPatient scansDeep learningPatient motionMotion estimationImage dataLSTM networkNeural networkRealistic patient motionTemporal informationMotion correction methodMotion detectionCardiac PETClinical workflowRigid translational motionFlow estimationNetworkPatient datasetsSuperior performanceMDPET: A Unified Motion Correction and Denoising Adversarial Network for Low-Dose Gated PET
Zhou B, Tsai YJ, Chen X, Duncan JS, Liu C. MDPET: A Unified Motion Correction and Denoising Adversarial Network for Low-Dose Gated PET. IEEE Transactions On Medical Imaging 2021, 40: 3154-3164. PMID: 33909561, PMCID: PMC8588635, DOI: 10.1109/tmi.2021.3076191.Peer-Reviewed Original ResearchConceptsMotion estimationPyramid networkAdversarial networkAccurate motion estimationMotion correctionLow-noise reconstructionGated positron emission tomographyMotion correction methodMotion estimation networkGated PET dataEstimation networkRecurrent layersDenoising NetworkRespiratory motion blurringExperimental resultsLow-noise imagesMotion blurringNoise levelCorrection methodNetworkPET reconstructionPrevious methodsImage qualityImagesEstimationSuper-resolution PET Brain Imaging using Deep Learning
Ren S, Liu J, Xie H, Toyonaga T, Mirian N, Chen M, Aboian M, Carson R, Liu C. Super-resolution PET Brain Imaging using Deep Learning. 2021, 00: 1-6. DOI: 10.1109/nss/mic44867.2021.9875548.Peer-Reviewed Original ResearchDeep learning networkPET image resolutionData augmentation methodImage resolutionSuper-resolution approachMedical imaging modalitiesClinical brain imagesDeep learningLearning networkAugmentation methodPET image qualityBrain imagesImage qualityNetworkImagesMedical diagnostic technologyPET imagesHRRT imagesData generalizabilityLearningSubstantial improvementScannerTechnologyPET brain imagingAccuracyAnatomy-Constrained Contrastive Learning for Synthetic Segmentation Without Ground-Truth
Zhou B, Liu C, Duncan J. Anatomy-Constrained Contrastive Learning for Synthetic Segmentation Without Ground-Truth. Lecture Notes In Computer Science 2021, 12901: 47-56. DOI: 10.1007/978-3-030-87193-2_5.Peer-Reviewed Original ResearchSegmentation networkContrastive learningManual segmentationSuperior segmentation performanceObject of interestSynthetic SegmentationManual effortSegmentation performanceTraining dataUnsupervised adaptationImaging dataSource modalitySegmentationNetworkPrevious methodsLearningLarge amountSuccessful applicationPET imaging dataImagesObjectsCodeDataNew imaging modalityLimited View Tomographic Reconstruction Using a Cascaded Residual Dense Spatial-Channel Attention Network With Projection Data Fidelity Layer
Zhou B, Zhou S, Duncan JS, Liu C. Limited View Tomographic Reconstruction Using a Cascaded Residual Dense Spatial-Channel Attention Network With Projection Data Fidelity Layer. IEEE Transactions On Medical Imaging 2021, 40: 1792-1804. PMID: 33729929, PMCID: PMC8325575, DOI: 10.1109/tmi.2021.3066318.Peer-Reviewed Original ResearchConceptsAttention networkView reconstructionGrand challenge datasetLimited angle reconstructionHigh-quality reconstructionNeural network methodSparse-view reconstructionExperimental resultsLimited angle acquisitionArchitecture issuesSparse viewsChallenge datasetLimited view dataView dataNeural architectureQuality reconstructionNetwork methodTomographic reconstructionReconstructed imagesProjection viewsPrevious methodsAngle reconstructionDatasetNetworkAngle acquisition
2019
Deep Learning based Respiratory Pattern Classification and Applications in PET/CT Motion Correction
Guo Y, Dvornek N, Lu Y, Tsai Y, Hamill J, Casey M, Liu C. Deep Learning based Respiratory Pattern Classification and Applications in PET/CT Motion Correction. 2019, 00: 1-5. DOI: 10.1109/nss/mic42101.2019.9059783.Peer-Reviewed Original ResearchDeep learningNeural networkMotion correction methodDeep neural networksDeep learning modelsHybrid neural networkConvolutional layersHigh prediction accuracyRecurrent layersGeneralization capabilityData preprocessingLearning modelPattern classificationRespiratory motionAnzai systemLoss functionLinear classifierPrediction accuracyIntra-gate motionRPM systemMotion correctionTumor detectionNetworkIrregular breathersCT images