Table 1 Open in new tab Accuracy and...

Inference method	Accuracy %	Accuracy weighted %	Sensitivity %	Specificity %	Misinference error %
Single source binary
LSC (Fig. 3A)	33	5	100	31	92
LSM (Fig. 3B) LSM_vc	48 71	8 54	100 100	47 70	88 26
MAPP (Fig. 3C)	100	100	100	100	0
MSA (Fig. 3D)	100	100	100	100	0
Single source binary and 10% noise
LSC	77	11	100	78	88
LSM	74	11	100	73	88
MAPP	98	46	100	98	6
MSA	98	73	100	98	4
Double source binary synergistic
LSC	22	22	100	18	90
LSM	35	35	100	32	86
MAPP	100	100	100	100	0
MSA	100	100	100	100	0
Double source binary redundant
LSC	39	39	88	37	91
LSM	47	47	100	45	86
MAPP	98	98	99	98	1
MSA	93	92	100	92	12
Double source binary redundant with 10% noise
LSC	80	79	100	79	89
LSM	77	76	99	76	86
MAPP	100	100	100	100	95
MSA	98	98	100	98	23
Double source binary redundant with 50% noise
LSC	79	78	99	78	90
LSM	74	74	99	73	86
MAPP	100	100	100	100	96
MSA	97	96	97	97	53
Double source binary mutual inhibition
LSC	41	41	90	40	92
LSM	47	46	98	44	85
MAPP	95	95	98	95	1
MSA	95	95	96	95	7

Inference method

Accuracy %

Accuracy weighted %

Sensitivity %

Specificity %

Misinference error %

Single source binary

LSC (Fig. 3A)

100

LSM (Fig. 3B)
LSM_vc

48
71

8
54

100
100

47
70

88
26

MAPP (Fig. 3C)

100

MSA (Fig. 3D)

100

Single source binary and 10% noise

LSC

100

LSM

100

MAPP

100

MSA

100

Double source binary synergistic

LSC

100

LSM

100

MAPP

100

MSA

100

Double source binary redundant

LSC

LSM

100

MAPP

MSA

100

Double source binary redundant with 10% noise

LSC

100

LSM

MAPP

100

MSA

100

Double source binary redundant with 50% noise

LSC

LSM

MAPP

100

MSA

Double source binary mutual inhibition

LSC

LSM

MAPP

MSA

Inference method	Accuracy %	Accuracy weighted %	Sensitivity %	Specificity %	Misinference error %
Single source binary
LSC (Fig. 3A)	33	5	100	31	92
LSM (Fig. 3B) LSM_vc	48 71	8 54	100 100	47 70	88 26
MAPP (Fig. 3C)	100	100	100	100	0
MSA (Fig. 3D)	100	100	100	100	0
Single source binary and 10% noise
LSC	77	11	100	78	88
LSM	74	11	100	73	88
MAPP	98	46	100	98	6
MSA	98	73	100	98	4
Double source binary synergistic
LSC	22	22	100	18	90
LSM	35	35	100	32	86
MAPP	100	100	100	100	0
MSA	100	100	100	100	0
Double source binary redundant
LSC	39	39	88	37	91
LSM	47	47	100	45	86
MAPP	98	98	99	98	1
MSA	93	92	100	92	12
Double source binary redundant with 10% noise
LSC	80	79	100	79	89
LSM	77	76	99	76	86
MAPP	100	100	100	100	95
MSA	98	98	100	98	23
Double source binary redundant with 50% noise
LSC	79	78	99	78	90
LSM	74	74	99	73	86
MAPP	100	100	100	100	96
MSA	97	96	97	97	53
Double source binary mutual inhibition
LSC	41	41	90	40	92
LSM	47	46	98	44	85
MAPP	95	95	98	95	1
MSA	95	95	96	95	7

Table 1

Open in new tab

Accuracy and misinference errors of uni- and multivariate evaluations of binary ground-truth models of brain function of single source region and double source regions

Inference method	Accuracy %	Accuracy weighted %	Sensitivity %	Specificity %	Misinference error %
Single source binary
LSC (Fig. 3A)	33	5	100	31	92
LSM (Fig. 3B) LSM_vc	48 71	8 54	100 100	47 70	88 26
MAPP (Fig. 3C)	100	100	100	100	0
MSA (Fig. 3D)	100	100	100	100	0
Single source binary and 10% noise
LSC	77	11	100	78	88
LSM	74	11	100	73	88
MAPP	98	46	100	98	6
MSA	98	73	100	98	4
Double source binary synergistic
LSC	22	22	100	18	90
LSM	35	35	100	32	86
MAPP	100	100	100	100	0
MSA	100	100	100	100	0
Double source binary redundant
LSC	39	39	88	37	91
LSM	47	47	100	45	86
MAPP	98	98	99	98	1
MSA	93	92	100	92	12
Double source binary redundant with 10% noise
LSC	80	79	100	79	89
LSM	77	76	99	76	86
MAPP	100	100	100	100	95
MSA	98	98	100	98	23
Double source binary redundant with 50% noise
LSC	79	78	99	78	90
LSM	74	74	99	73	86
MAPP	100	100	100	100	96
MSA	97	96	97	97	53
Double source binary mutual inhibition
LSC	41	41	90	40	92
LSM	47	46	98	44	85
MAPP	95	95	98	95	1
MSA	95	95	96	95	7

Inference method	Accuracy %	Accuracy weighted %	Sensitivity %	Specificity %	Misinference error %
Single source binary
LSC (Fig. 3A)	33	5	100	31	92
LSM (Fig. 3B) LSM_vc	48 71	8 54	100 100	47 70	88 26
MAPP (Fig. 3C)	100	100	100	100	0
MSA (Fig. 3D)	100	100	100	100	0
Single source binary and 10% noise
LSC	77	11	100	78	88
LSM	74	11	100	73	88
MAPP	98	46	100	98	6
MSA	98	73	100	98	4
Double source binary synergistic
LSC	22	22	100	18	90
LSM	35	35	100	32	86
MAPP	100	100	100	100	0
MSA	100	100	100	100	0
Double source binary redundant
LSC	39	39	88	37	91
LSM	47	47	100	45	86
MAPP	98	98	99	98	1
MSA	93	92	100	92	12
Double source binary redundant with 10% noise
LSC	80	79	100	79	89
LSM	77	76	99	76	86
MAPP	100	100	100	100	95
MSA	98	98	100	98	23
Double source binary redundant with 50% noise
LSC	79	78	99	78	90
LSM	74	74	99	73	86
MAPP	100	100	100	100	96
MSA	97	96	97	97	53
Double source binary mutual inhibition
LSC	41	41	90	40	92
LSM	47	46	98	44	85
MAPP	95	95	98	95	1
MSA	95	95	96	95	7

LSM_vc: LSM volume corrected; TP: true positive; TN: true negative; FP: false positive; FN: false negative. Accuracy = (TP + TN)/(TP + TN + FP + FN); Accuracy_weighted = (TP*(1-var(diag))+TN*(1-sum(off_diag)))/(TP + TN + FP + FN). Sensitivity = TP/(TP + FN); Specificity = TN/(TN + FP); Mis-inference error = sum of all inferred regional contributions beyond the ones of the target/s. All the quantities are computed on the normalized significant values in the matrices (figures instead show all the values in the matrices, without excluding the nonsignificant values). For the single source binary simulation with noise we used a SVM predictor with linear kernel and cost function = 1, while for all other simulations, we used a regression tree predictor (default parameters). For all simulations without noise, we considered all the values for the regions not expected to be found by the method (FP and TN), whereas for the simulations with noise, we only considered the regions with a contribution higher than 0.001 to avoid all kinds of spurious results.

This Feature Is Available To Subscribers Only