TABLE 2.
The demographic and scan parameters of the looser datasets.
| Class | Database | Num (train/test) | Age (mean ± std) | Male/female | Flip angel (degrees) | Slice thickness (mm) |
| AD | ADNI | 334/88 | 75.5 ± 7.79 | 230/192 | 9.00 ± 0 | 1.20 ± 0 |
| FTD | NIFD | 440/112 | 65.1 ± 7.48 | 332/220 | 9.00 ± 0 | 1.20 ± 0 |
| NC | ADNI | 381/88 | 75.3 ± 6.19 | 224/245 | 9.00 ± 0 | 1.20 ± 0 |
| NIFD | 282/72 | 64.9 ± 7.85 | 151/203 | 9.00 ± 0 | 1.20 ± 0 | |
| ADl | ADNI | 1,051/263 | 75.4 ± 8.02 | 729/585 | 9.16 ± 1.20 | 1.18 ± 0.05 |
| FTDl | NIFD | 1,001/249 | 64.9 ± 7.63 | 744/506 | 8.93 ± 0.26 | 1.20 ± 0 |
| NCl | ADNI | 749/189 | 75.7 ± 6.64 | 458/480 | 9.08 ± 1.08 | 1.16 ± 0.08 |
| NIFD | 475/122 | 64.4 ± 8.17 | 240/357 | 9.00 ± 0 | 1.20 ± 0 |
The superscript “l” indicates that we removed the restriction of scanning from MPRAGE sequence and distinguish from scenarios 1–4.