Brain extraction from cerebral MRI volume using a hybrid level set based active contour neighborhood model

Table 1 Comparison of ACNM with existing brain extraction approaches on data set 1: BSE in Brainsuite software, BET in MRIcro software, WAT, HWA, GCUT and ROBEX using IBSR data set 1 (18 × 1.5 mm scans)

Method	DS mean (SD)[range]	JS mean (SD)[range]	FP_Rate(%) mean (SD)[range]	FN_Rate(%) mean (SD)[range]	HD mean (SD)[range]
BSE	0.94(0.04)	0.90(0.07)	7.8(6.2)	3.7(6.3)	16.5(6.7)
BSE	[0.83-0.97]	[0.70-0.95]	[1.6-18.6]	[0.5-28.5]	[8-30]
BET	0.88(0.03)	0.78(0.04)	5.7(1.3)	17.5(4.9)	19.6(4.4)
BET	[0.86-0.96]	[0.72-0.93]	[4.4-8.9]	[1.0-24.9]	[7-29]
WAT	0.91(0.08)	0.85(0.11)	18.8(29.7)	2.45(1.79)	_
WAT	[0.60-0.96]	[0.43-0.92]	[5.3-134.3]	[0.08-7.08]	_
HWA	0.88(0.03)	0.79(0.04)	27.1(6.8)	0.015(0.02)	_
HWA	[0.82-0.91]	[0.69-0.83])	[20.0-44.4])	[0-0.07]	_
GCUT	0.91(0.02)	0.84(0.03)	19.3(4.0)	0.029(0.04)	_
GCUT	[0.87-0.93]	[0.78-0.87])	[14.8-28.6]	[0-0.15]	_
ROBEX	0.96(0.08)	_	_	_	13.3(2.6)
ACNM One	0.95(0.02)	0.90(0.03)	7.04(3.68)	3.58(3.95)	12.4(4.5)
ACNM One	[0.91-0.97]	[0.83-0.94]	[3.0-15.1]	[0.58-14.28]	[7-21]
ACNM Two	0.96(0.01)	0.92 (0.03)	5.28(2.06)	3.41(2.73)	10.5(3.0)
ACNM Two	[0.93-0.98]	[0.88-0.95]	[2.81-10.35]	[0.82-8.64]	[7-18]
ACNM Three	0.94(0.03)	0.90(0.05)	4.66(1.6)	6.15(5.86)	14.9(4.4)
ACNM Three	[0.85-0.98]	[0.74-0.95]	[1.21-8.11]	[1.88-11.17]	[7-24]

“ACNM One” represents performing the ACNM with fixed parameters. “ACNM Two” represents performing the ACNM with varied parameters. “ACNM Three” represents initializing using a simple circle.

ISSN: 1475-925X