A Classification Algorithm by Combination of Feature Decomposition and Kernel Discriminant Analysis (KDA) for Automatic MR Brain Image Classification and AD Diagnosis

Table 8

Comparison between the proposed classification and previous results.

Methods	Subjects	Modalities	AD vs. NC (%)	pMCI vs. sMCI (%)	pMC vs. NC (%)	pMCI vs. AD (%)
Baseline and also longitudinal patterns of the brain [40]	27 pMCI, 76 sMCI	MRI	—	81.5	—	—
Pattern classification using baseline measurements [42]	53 AD, 53 NC, 237 MCI	MRI	—	—	—	53.3
Voxel_stand_D GM features and SVM classifier [1]	76 pMCI, 134 sMCI	MRI	—	70	70.40	—
ROI GM feature and via SVM classifier [43]	51 AD, 52 NC, 99 MCI	MRI	62
ROI GM feature and via SVM [44]	198 AD, 231 NC, 167 pMCI, 238 sMCI	MRI		64.68	82.76	—
Koikkalainen et al. [45]	54 pMCI, 115 sMCI	MRI	86	72	—	—
BrainAGE framework [46]	188 NC, 171 NC, 133pMCI, 62 sMCI	MRI	—	75	—	—
Separating pMCI subjects from different individuals [47]	61 pMCI, 134 sMCI	MRI		66.7
Casanova et al. [48]	188 NC, 171AD, 153 pMCI, 182 sMCI	MRI	81.4	61.5	63	—
Data-driven ROI [49]	97 AD, 128 NC, 117 pMCI, 117 sMCI	MRI	—	73.69	—	—
Tong and Gao [50]	191 AD, 229 NC, 161 pMCI, 100 sMCI	MRI	76	—	—	—
Combining MRI data with cognitive test results MRI [24]	53 AD, 53 NC 237 MCI	MRI	—	—	—	61
Discriminative multitask feature selection method [51]	51AD, 52 NC, 99 MCI	MRI	87.2	53.68	68.02	—
Inherent structure-based multiview learning method [52]	97AD, 128 NC, 117 pMCI, 175 sMCI	MRI	92.51	78.88	—	—
Explicitly modeling structural information in the multitemplate data [53]	97 AD, 128 NC, 117 pMCI, 175 sMCI	MRI	93.6	79.25	—	—
Proposed method	98 AD, 229 NC, 167 pMCI, 236 sMCI	MRI	90.40	65.04	84.33	58.93