2025
Improving the Robustness of Deep-Learning Models in Predicting Hematoma Expansion from Admission Head CT.
Tran A, Karam G, Zeevi D, Qureshi A, Malhotra A, Majidi S, Murthy S, Park S, Kontos D, Falcone G, Sheth K, Payabvash S. Improving the Robustness of Deep-Learning Models in Predicting Hematoma Expansion from Admission Head CT. American Journal Of Neuroradiology 2025, ajnr.a8650. PMID: 39794133, DOI: 10.3174/ajnr.a8650.Peer-Reviewed Original ResearchFast Gradient Sign MethodDeep learning modelsRobustness of deep learning modelsAdversarial attacksAdversarial imagesAdversarial trainingSign MethodModel robustnessDeploying deep learning modelsDeep learning model performanceConvolutional neural networkImprove model robustnessAcute intracerebral hemorrhageHematoma expansionMulti-threshold segmentationReceiver operating characteristicIntracerebral hemorrhageGradient descentType attacksData perturbationNeural networkProjected GradientTraining setAntihypertensive Treatment of Acute Cerebral HemorrhageThreshold-based segmentationEstimation of leading multi-block canonical correlation directions via ℓ1-norm constrained proximal gradient descent
Guan L. Estimation of leading multi-block canonical correlation directions via ℓ1-norm constrained proximal gradient descent. Electronic Journal Of Statistics 2025, 19 DOI: 10.1214/25-ejs2351.Peer-Reviewed Original Research
2024
Efficient standardization of clinical T2‐weighted images: Phase‐conjugacy e‐CAMP with projected gradient descent
Zhang H, Elsaid N, Sun H, Tagare H, Galiana G. Efficient standardization of clinical T2‐weighted images: Phase‐conjugacy e‐CAMP with projected gradient descent. Magnetic Resonance In Medicine 2024, 92: 2723-2733. PMID: 38988054, DOI: 10.1002/mrm.30214.Peer-Reviewed Original ResearchData fidelity termSignal evolution modelFidelity termGradient descentProjected GradientEfficient algorithmVirtual conjugate coilAlgorithmObjective functionMapping errorsTunable parametersLinear constraintsSampling schemeTSE dataLong echo train lengthMapsTrain lengthVirtualTurbo spin echoEcho train lengthHigh-resolutionSchemeDataBackground phaseError rangeMedical image registration via neural fields
Sun S, Han K, You C, Tang H, Kong D, Naushad J, Yan X, Ma H, Khosravi P, Duncan J, Xie X. Medical image registration via neural fields. Medical Image Analysis 2024, 97: 103249. PMID: 38963972, DOI: 10.1016/j.media.2024.103249.Peer-Reviewed Original ResearchLearning-based methodsNeural fieldsNeural networkImage registrationMedical image analysis tasksMini-batch gradient descentImage analysis tasksDeep neural networksMedical image registrationDiffeomorphic image registrationImage registration frameworkOptimization-based methodDomain shiftAnalysis tasksGradient descentCompetitive performanceImage pairsRegistration taskOptimal deformationShort computation timeRegistration frameworkDesign choicesDisplacement vector fieldComputation timeModel optimizationEfficient Standardization of Clinical T2-Weighted Images: Phase-Conjugacy e-CAMP with Projected Gradient Descent
Zhang H, Elsaid N, Sun H, Tagare H, Galiana G. Efficient Standardization of Clinical T2-Weighted Images: Phase-Conjugacy e-CAMP with Projected Gradient Descent. Proceedings Of The International Society For Magnetic Resonance In Medicine ... Scientific Meeting And Exhibition. 2024 DOI: 10.58530/2024/2733.Peer-Reviewed Original ResearchProjected Gradient DescentGradient descentMachine learningLarge-scale machine learningMassive data sourcesMap reconstructionParameter tuningRobust implementationParameter choicesEfficient enforcementData sourcesModel constraintsVirtual conjugate coilLearningImagesRoutine clinical imagingVirtualClinical imagesAlgorithmDatasetClinical scansParameter valuesMachineDescentQuantitative imaging
2023
Comparing Detection Schemes for Adversarial Images against Deep Learning Models for Cancer Imaging
Joel M, Avesta A, Yang D, Zhou J, Omuro A, Herbst R, Krumholz H, Aneja S. Comparing Detection Schemes for Adversarial Images against Deep Learning Models for Cancer Imaging. Cancers 2023, 15: 1548. PMID: 36900339, PMCID: PMC10000732, DOI: 10.3390/cancers15051548.Peer-Reviewed Original ResearchAdversarial imagesDeep learning modelsDL modelsDetection modelLearning modelConvolutional neural networkDetection schemeAdversarial detectionDefense techniquesMachine learningMedical imagesAdversarial perturbationsInput imageAdversarial trainingNeural networkArt performanceMagnetic resonance imagingGradient descentPixel valuesHigh accuracyImagesBrain magnetic resonance imagingAbsence of malignancyClassificationScheme
2021
Use of Machine Learning Models to Predict Death After Acute Myocardial Infarction
Khera R, Haimovich J, Hurley NC, McNamara R, Spertus JA, Desai N, Rumsfeld JS, Masoudi FA, Huang C, Normand SL, Mortazavi BJ, Krumholz HM. Use of Machine Learning Models to Predict Death After Acute Myocardial Infarction. JAMA Cardiology 2021, 6: 633-641. PMID: 33688915, PMCID: PMC7948114, DOI: 10.1001/jamacardio.2021.0122.Peer-Reviewed Original ResearchConceptsMachine learning modelsMeta-classifier modelLearning modelNeural networkGradient descent boostingAcute myocardial infarctionContemporary machineGradient descentXGBoost modelXGBoostHospital mortalityCohort studyLogistic regressionMyocardial infarctionNetworkChest Pain-MI RegistryPrecise classificationIndependent validation dataInitial laboratory valuesNovel methodLarge national registryHigh-risk individualsData analysisValidation dataResolution of riskGradient Descent and Newton's Method with Backtracking Line Search in Linear Regression
Liang J. Gradient Descent and Newton's Method with Backtracking Line Search in Linear Regression. 2021, 00: 394-397. DOI: 10.1109/cds52072.2021.00073.Peer-Reviewed Original ResearchBacktracking line searchLine searchNewton methodBacktracking line search methodLine search methodEfficiency of optimization algorithmsOptimization algorithmSupervised machine learningRegression problemIteration stepNewtonGradient descentExecution timeMachine learningRate alphaOptimal parameter valuesSearch methodParameter valuesAlgorithmLearningSearchIteration
This site is protected by hCaptcha and its Privacy Policy and Terms of Service apply