Table 12-.

Mean ± standard deviation of WMH loads in small (<5cm³), medium (5–20cm³), and large (>20cm³) groups. Statistically significant differences from manual segmentations after corrections for multiple comparisons using false discovery rate (FDR) correction are indicated with *

Sequences	T1-FLAIR			T1–T2-PD
Method	Small	Medium	Large	Small	Medium	Large
Manual	1.85±1.39	12.07±4.58	40.02±23.33	1.47±1.14	10.81±4.27	47.62±25.74
Naïve Bayes	18.65±12.52*	33.45±16.79*	68.31±37.88*	91.29±98.89*	89.85±33.56*	148.31±48.49*
Logistic Regression	2.53±2.59*	9.83±4.98*	34.03±25.35*	2.18±5.68	4.44±8.32*	32.80±27.80*
LDA	3.97±3.09*	12.48±5.76	39.49±30.83	22.44±8.16*	28.72±12.14*	53.47±19.22
QDA	17.26±11.21*	32.53±16.05*	73.04±42.40*	121.10±89.30*	129.28±43.98*	205.15±69.82*
KNN	2.42±2.46*	11.00±5.81	41.53±30.49	4.74±21.28	8.74±8.53	45.50±25.54
Decision Trees	3.44±3.24*	12.64±6.35	41.63±25.35	5.10±4.41*	14.03±9.86*	54.30±26.60*
Random Forests	2.62±2.61*	11.47±5.87	40.79±25.37	2.58±3.29	11.51±9.72	51.57±27.38
Bagging	1.46±2.41	7.13±6.23*	28.13±21.69*	3.12±14.95	5.16±7.17*	32.86±20.56*
SVM	2.76±3.23*	11.05±5.90	41.17±30.55	5.49±8.53*	11.98±14.87	49.28±29.49
AdaBoost	2.85±2.78*	11.93±5.97	41.08±25.60	3.42±4.05*	11.94±10.25	52.47±29.93