Automatic Classification of Early Parkinsons Disease with Multi-Modal MR Imaging 英文参考文献.docVIP

AutomaticClassificationofEarlyParkinson’sDisease withMulti-ModalMRImaging DanLong1.,JinweiWang2.,MinXuan1,QuanquanGu1,XiaojunXu1,DexingKong3*,MinmingZhang1* 1Department of Radiology, the Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China, 2Center of Mathematical Sciences, Zhejiang University,Hangzhou,China,3DepartmentofMathematics,ZhejiangUniversity,Hangzhou,China Abstract Background: In recent years, neuroimaging has been increasingly used as an objective method for the diagnosis of Parkinson’sdisease(PD).Mostpreviousstudieswerebasedoninvasiveimagingmodalitiesoronasinglemodalitywhich wasnotanidealdiagnostictool.Inthisstudy,wedevelopedanon-invasivetechnologyintendedforuseinthediagnosisof earlyPDbyintegratingtheadvantagesofvariousmodals. MaterialsandMethods:NineteenearlyPDpatientsandtwenty-sevennormalvolunteersparticipatedinthisstudy.Foreach subject, we collected resting-state functional magnetic resonance imaging (rsfMRI) and structural images. For the rsfMRI images, we extracted the characteristics at three different levels: ALFF (amplitude of low-frequency fluctuations), ReHo (regionalhomogeneity)andRFCS(regionalfunctionalconnectivitystrength).Forthestructuralimages,weextractedthe volumecharacteristicsfromthegraymatter(GM),thewhitematter(WM)andthecerebrospinalfluid(CSF).Atwo-samplet- testwasusedforthefeatureselection,andthentheremainingfeatureswerefusedforclassification.Finallyaclassifierfor earlyPDpatientsandnormalcontrolsubjectswasidentifiedfromsupportvectormachinetraining.Theperformanceofthe classifierwasevaluatedusingtheleave-one-outcross-validationmethod. Results: Using the proposed methods to classify the data set, good results (accuracy =86.96%, sensitivity =78.95%, specificity =92.59%)wereobtained. Conclusions: This method demonstrates a promising diagnosis performance by the integration of information from a varietyofimagingmodalities,anditshowspotentialforimprovingtheclinicaldiagnosisandtreatmentofPD. Citation:LongD,Wan