Automatic Liver Segmentation on Volumetric CT Images Using Supervoxel-Based Graph Cuts

Table 1

Overview of liver segmentation methods for CT images: auto = automatic; semi = semiautomatic; VOE = volumetric overlap error; RVD = relative absolute volume difference; MaxD = maximum symmetric surface distance; DSC = dice similarity coefficient; RG = region growing; DM = deformable model; SSM = statistical shape model; PA = probabilistic atlas; GC = graph cuts; local = from local hospitals; Sliver07 = MICCAI 2007 grand challenge in segmentation of liver datasets.

Study	Year	Method	Accuracy	Dataset	Auto	Time (s)	CPU (GHz)
Kumar et al. [7]	2013	RG	DSC = 98%	local	Auto	40/slice	—
Goryawala et al. [8]	2014	Clustering + RG	DSC = 92%; RVD = 2.78%	local	Semi	10.96/slice	—
Peng et al. [10]	2014	DM	VOE = 6.10%; RVD = −0.00%; MaxD = 16.80 mm	Sliver07	Semi	180	3.16 GHz; 4 GB RAM
Kainmüller  et al. [15]	2007	SSM + DM	VOE = 6.09%; RVD = −2.86%; MaxD = 18.69 mm	Sliver07	Auto	900	Intel 3.2 GHz
Linguraru et al. [16]	2010	PA + DM	DSC = 96.2%; VOE = 2.20%; ASD = 1.20 mm	Sliver07	Auto	—	—
Platero and Tobar [18]	2014	PA + GC	VOE = 7.60%; RVD = −0.50%; MaxD = 24.70 mm	Sliver07	Auto	261.35	Intel Xeon E5520 2.27 GHz
Massoptier and Casciaro [19]	2007	GC	DSC = 95%	local	Auto	—	—
Li et al. [20]	2015	SSM + GC	VOE = 6.24%; RVD = 1.18%; MaxD = 18.82 mm	Sliver07	Auto	284.95	Core(TM) i5 3.1 GHz; 4 GB RAM
Chen et al. [23]	2012	GC	VOE = 4.16%; RVD = 3.53%; MaxD = 16.70 mm	Sliver07	Semi	60–180	Intel Core 2; 2.66 GHz; 3.25 RAM