Analysis on cerebral diffusion tensor imaging automatic fiber quantification of patients with Alzheimer's disease / 中华行为医学与脑科学杂志

ABSTRACT

Objective:To investigate the application value of magnetic resonance (MR) diffusion tensor imaging (DTI) automatic fiber quantification (AFQ) in the diagnosis and prediction of Alzheimer's disease (AD).Methods:Clinical and MR data of 21 patients with AD (AD group) and 33 normal controls (NC group) were collected.AFQ software was used to analyze DTI data, track 20 white matter fiber bundles in the brain, and compare the differences of fractional anisotropy (FA) and mean diffusivity (MD) value of each bundle between groups.Each fiber bundle was divided into 100 equal parts along the direction of travel, and the FA or MD value of each part was taken as a characteristic.Screening the characteristics with statistic differences between groups for classification of AD and NC by support vector machine (SVM) with leave one method for cross validation.Classification effectiveness was evaluated using the receiver operating characteristic (ROC) curve.Results:Eleven (left/right anterior thalamic radiation (ATR), left/right corticospinal tract (CST), genu of corpus callosum (CC Genu), right inferior longitudinal fasciculus (ILF), right superior longitudinal fasciculus (SLF), left/right uncinated fasciculus (UF), and left/right arcuate fasciculus (AF)) of the 20 fiber bundles were successfully tracked in all subjects.Compared with NC group, the FA values of 2 fiber bundles (left/right UF) in AD group were significantly decreased( t=-2.532, -2.391, both P<0.05), and the MD values of 7 fiber bundles (left ATR, left/right CST, right ILF, left/right UF, and left AF) were significantly increased ( t=2.569, 2.411, 2.108, 2.357, 3.773, 3.796, 3.492, all P<0.05). Among the 2 200 characteristics of 11 fiber bundles, 412 classification characteristics with inter-group differences were selected.Among which, 78 FA characteristics were distributed in 7 fiber bundles (left ATR, left/right CST, CC Genu, right ILF, left/right UF), and 334 MD characteristics were distributed in 9 fiber tracts (left/right ATR, left/right CST, CC Genu, right ILF, left/right UF, and left AF). The accuracy of SVM classification was 85.19%, sensitivity was 80.95%, specificity was 87.88%, and area under ROC curve was 0.894 7. Conclusion:AFQ analysis based on DTI has a high application value in the diagnosis and prediction of AD.