A combination of support vector machine and voxel-based morphometry in adult male alcohol use disorder patients with cognitive deficits
Xiyong Dai,Lei Gao,Haibo Zhang,Xinhua Wei,Zaiyi Liu
DOI: https://doi.org/10.1016/j.brainres.2021.147644
IF: 3.61
2021-11-01
Brain Research
Abstract:Cognitive performance develops with drinking. However, the neural basis of cognitive deficits in AUD is still absent. Here we examined the relationship between overall drinking, brain structural alterations and cognitive deficits in middle-aged AUD men by mediation analysis. A total of 40 patients and 40 healthy controls (HC) underwent highresolution T1-weighted imaging scans and neuroimaging data were applied to voxelbased morphometry, support vector machine (SVM) classifications and mediation analysis. The patients demonstrated reduced gray matter (GM) volumes in different brain regions, including left amygdala, thalamus, hippocampus, precentral gyrus, cerebellum, calcarine, right supplementary motor area and bilateral superior temporal gyrus (all p values < 0.0001, FWE). Decreased GM volumes in the bilateral cerebellum (left r = 0.501, p = 0.001; right r = 0.519, p = 0.001) and thalamus (left r = 0. 442, p = 0.004; right r = 0.418, p = 0.007), left anterior medial temporal Lobe (r = 0.425, p = 0.006), left nucleus ambiguus+parahippocampus gyrus (r = 0.433, p = 0.005), left fusiform gyrus (r = 0.453, p = 0.003), left lingual gyrus (r = 0.455, p = 0.003), left hippocampus (r = 0.434, p = 0.005), right nucleus accumbens volume (r = 0.533, p < 0.001) had positive correlations with MoCA scores. The mediation analysis showed that left Crus 1 partially mediated the association of overall drinking and MoCA scores (standardized beta coefficient = -0.0003, s.e. = 0.0002, 95% CI = (-0.0006, 0.0000)). The SVM results could distinguish patients from HC with satisfactory classification results (0.8275). Our findings showed the altered GM volumes within the extended amygdala circuits and cognitive deficits in patients. The left Crus 1 may be an informative biomarker to predict the risk of AUD and a possible neural basis of cognitive deficits in middle-aged AUD men.
neurosciences