Table 7.

Summary of liver vessel segmentation approaches using supervised learning (segmentation metrics are given in Table 2). CT: computed tomography, MR: magnetic resonance, USG: ultrasonography.

Author	Testing Dataset	Synthetic Data Used	Metrics Results
Ibragimov et al. [34]	72 CT images	No	$DSC=0.83,e_{Dsym}=1.08$ mm
Kitrungrotsakul et al. [29]	1 CT volume (3D-IRCADb-01)	Yes	$Sens=0.9,DSC=0.92,PPV=0.84,$
	(http://www.ircad.fr/research/3dircadb/)		$VOE=17.2\%$
	(accessed on 5 May 2019)
Keshwani et al. [28]	20 CT volumes from internal dataset	Yes	$Sens=0.96,Spec=0.94,DSC=0.94$
	and 20 CT (3D-IRCADb-01) volumes		$Sens=0.96,Spec=0.91,DSC=0.92$
	(http://www.ircad.fr/research/3dircadb/)
	(accessed on 5 May 2019)
Zhang et al. [31]	20 CT (3D-IRCADb-01) volumes	No	$Ac=0.96,Sens=0.73,Spec=0.97,$
	(http://www.ircad.fr/research/3dircadb/)		$DSC=0.67$
	(accessed on 5 May 2019)
	20 CT (SLIVER07) datasets [51]		$Ac=0.96,Sens=0.89,Spec=0.97,$
			$DSC=0.71$
Huang et al. [30]	10 CT volumes from internal dataset and	Yes	$Ac=0.97,Sens=0.76,Spec=0.98,$
	20 CT (SLIVER07) datasets [51] and		$DSC=0.75$
	10 CT (3D-IRCADb-01) volumes
	(http://www.ircad.fr/research/3dircadb/)
	(accessed on 5 May 2019)
Thomson et al. [26]	203 USG volumes	No	$DSC=0.66$
Mishra et al. [32]	132 USG images	No	$JSC=0.69$
Yan et al. [25]	10 CT volumes from internal dataset and	No	$Sens=0.85,PPV=0.78,DSC=0.8$
	20 CT (3D-IRCADb-01) volumes		$Sens=0.93,PPV=0.99,DSC=0.9$
	(http://www.ircad.fr/research/3dircadb/)
	(accessed on 5 May 2019)
Xu et al. [27]	20 CT (3D-IRCADb-01) volumes	No	$Ac=0.99,Sens=0.78,Spec=0.99,$
	(http://www.ircad.fr/research/3dircadb/)		$DSC=0.68$
	(accessed on 5 May 2019)
Nazir et al. [24]	30 CTA internal datasets and	No	$Ac=$ up to $98.90$
	10 CT (SLIVER07) datasets [51]		$Ac=$ up to $98.89$