Table 1.

Total count and percentage of regional uncorrected (p < 0.01) and FDR corrected morphometric abnormalities in MRIs of TLE patients.

		used datasets: 42 MRIs from 32 TLE patients			excluded datasets: 5 MRIs from 5 TLE patients
		test count	puncorr < 0.01	pFDR < 0.01	test count	puncorr < 0.01	pFDR < 0.01
raw	count	83454	4041	1514	9935	1197	610
	percentage		4.84%	1.81%		12.05%	6.14%
	p (healthy controls)		0	0		0	0
	p (QC passed)		n.a.	n.a.		0	0
normalized	count	83454	4109	1553	9935	1300	622
	percentage		4.92%	1.86%		13.09%	6.26%
	p (healthy controls)		0	0		0	0
	p (QC passed)		n.a.	n.a.		0	0
asymmetry	count	41412	1456	326	4930	544	180
	percentage		3.52%	0.79%		11.03%	3.65%
	p (healthy controls)		0	0		0	0
	p (QC passed)		n.a.	n.a.		0	0

A binomial model was used to compare the empirical detection rates to the expectations in healthy controls³². In patients they were highly significantly larger than in a leave-one-out cross-validation in healthy controls with p-values zero to machine precision. For datasets that did not pass the quality checks we also compared to the detection rate in high quality datasets of TLE patients. In excluded datasets the detection rates were highly significantly elevated.

Abbreviations: FDR, false discovery rate; n.a., not applicable; QC, quality check; uncorr, uncorrected.