The Differences Of White Matter Microstructure And Texture Features Based On 3D Structure Of ROI In Patients Of Early Stage Parkinson’s Disease And Multiple System Atrophy

Objective1.Through the Tract-based spatial statistics(TBSS)analysis of diffusion tensor imaging(DTI)in patients with early stage Parkinson’s disease(PD)and multiple system atrophy(MSA),normal control(NC),we extracted the microstructure characteristics of white matter in patients,to identified patients in early stage PD and MSA and NC,and to explored the similarities and differences of neurophysiological mechanism between patients with early stage PD and MSA,for making early and accurate diagnosis.2.To explore the feasibility of texture analysis in extracting features from ROI of 3D-T1WI images to identify patients with early stage PD and MSA,NC,for providing a reliable for early and accurate clinical diagnosis.MethodsAccording to the inclusion and exclusion criteria,50 early PD patients,50 early MSA patients and 50 NC were collected.All subjects were scanned by GE 3.0T(signa excite Ⅱ HD)MRI scanner,including conventional sequence,DTI and 3D-T1WI sequence.All DTI data were analyzed by TBSS method,and the fractional anisotropy(FA),medium diffusion(MD),radial diffusion(RD)and axial diffusion(AD)of white matter fiber bundles 50 ROIs in each subject were obtained.All 3D-T1WI images were used 3D slicer to extract 6 ROIs of bilateral caudate nucleus,bilateral putamen,midbrain and pons,including 11 texture features for each ROI.Spearman correlation test was used to select texture features to reduce the dimension.The FA,MD,RD and AD values of white matter fiber bundles 50 ROIs and the differences of texture features were compared by ANOVA test.Receiver operator characteristic curve(ROC)was used to evaluate the ability of DTI parameters and their combination and texture features to identify diseases among groups.Results1.There were significant differences in FA,MD,RD and AD between patients with early stage PD and MSA,NC of several supratentorial and infratentorial white matter fiber tracts.White matter injury of patients with early stage MSA was more serious in infratentorial structure.Compared with NC,only lower FA values in right anterior radial and lower AD values in right posterior colliculus was in found patients with early stage PD.The FA,MD,RD and AD of middle cerebellar peduncle were significantly different among the three groups.2.The FA,MD,RD and AD and their combination of middle cerebellar peduncle had better diagnostic efficacy in distinguishing patients with early stage PD and MSA,NC,and the corresponding area of curve(AUC),were greater than 0.5,except for distinguishing PD from MSA-P,and PD from NC.The diagnositic efficacy of combination was better than that of single parameter.3.Texture analysis showed that the sum average in the right caudate nucleus,left caudate nucleus and pons,the entropy in the bilateral caudate nucleus,the difference variance was in the right caudate nucleus,the correlation in the left caudate nucleus,and the mean absolute deviation in pons were significantly different among groups.4.In addition to sum average in the left caudate nucleus distinguishing NC from MSA-C and PD from MSA-C,and entropy in the right caudate nucleus distinguishing NC from MSA-C,the AUC of other textrue featrues of corresponding ROI were greater than 0.5.The best diagnositic efficacy was sum average in the right caudate nucleus,with AUC greater than 0.75;each feature had good diagnositic efficacy in distinguishing NC from MSA,NC from MSA-P.Conclusion1.There was injury on different degree in the microstructure of white matter in patients with early stage PD and MSA.The white matter injury in patients with MSA was more extensive.DTI parameters of middle cerebellar peduncle could be used to identify early PD and MSA,and their combination was more effective.2.Texture analysis based on 3D-T1WI showed great potentinally for patients with early stage PD and MSA in the differential diagnosis.and sum average in the right caudate nucleus was the best texture feature in the differential diagnosis.It was suggested that texture feature could be used as a marker for early differential diagnosis of PD and MSA.