Table 3.

Comparison of proposed method with existing methods.

Dataset details	Method	Feature set	Testing Accuracy/precision /recall/ sensitivity /AUC-ROC/ specificity
PPMI (3T T1 MRI scans with 3D images, 906 subjects) [30]	Support vector machine	Textural, morphological, and statistical features	81.74% accuracy, 82.41% precision, 80.71% recall, 81.78% f1-score
MRI scans – structural & functional, 213 subjects) [37]	Support vector machine	Intensity histogram, texture, wavelet	78.07% accuracy, 78.80% sensitivity, 76.08% specificity, 0.85 AUC
MRI Dataset (T1- weighted images, 103 subjects) [50]	Support vector machine	Morphological features	65% accuracy, 66.7% precision, 62.2% recall, 0.69 AUC
MRI dataset (Resting state images, 120 subjects) [51]	Random forest	Regional homogeneity, resting-state functional connectivity and gray matter volume	82.61% accuracy 0.90 AUC
MRI images- susceptibility weighted imaging (SWI) images, 190 subjects) [52]	Random forest	Intensity, shape, gray level co-occurrence matrix, run length matrix, gray level size zone matrix features	69% accuracy, 73% specificity 64% sensitivity
Proposed method (LBP variant-I)	Support vector machine	Advanced LBP histogram features	83.33% accuracy, 84.62% precision, 91.67% recall, 88% f1-score, 0.86 AUC