Table 3.

Benchmarking table showing ML/DL methods for cIMT/PA measurements

SN	Gen	Author (year)	Method	Cross-validation protocol	Data size	cIMT/PA error	Performance (cIMT error and PA errors)
1	First	Liang et al. [124] (2000)	Dynamic programming	NA	50	cIMT	$0.042\pm 0.02\mathrm{mm}$
2		Stein et al. [125] (2005)	Edge-detection	NA	50	cIMT	$0.012\pm 0.006\mathrm{mm}$
3		Faita et al. [126] (2008)	Edge-detection	NA	150	cIMT	$0.010\pm 0.038\mathrm{mm}$
4		Ikeda et al. [48] (2017)	BEP* detection	NA	649	cIMT	$0.0106\pm 0.00310\mathrm{mm}$
5	First + second	Molinari et al. [72] (2012)	Level set, morphological image processing	NA	200	cIMT	$0.144\pm 0.179\mathrm{mm}$
6		Molinari et al. [127] (2011)	CARES	NA	647	cIMT	$0.172\pm 0.222\mathrm{mm}$
7		Molinari et al. [24] (2011)	CAMES 1.0	NA	657	cIMT	$0.154\pm 0.227\mathrm{mm}$
8		Molinari et al. [128] (2011)	CAUDLES	NA	630	cIMT	$0.224\pm 0.252\mathrm{mm}$
9		Molinari et al. [129] (2011)	FOAM	NA	665	cIMT	$0.150\pm 0.169\mathrm{mm}$
10	Second	Gutierrez et al. [130] (2002)	Active contours	NA	180	cIMT	$0.090\pm 0.060\mathrm{mm}$
11		Molinari et al. [131] (2009)	Snakes	NA	200	cIMT	$0.01\pm 0.01\mathrm{mm}$
12		Molinari et al. [132] (2014)	CALEX 1.0	NA	665	cIMT	$0.191\pm 0.217\mathrm{mm}$
13	Third	Rosa et al. [92] (2013)	ANN	60% training, 40% testing	60	cIMT	$0.03763\pm 0.02518$ mm (mean) $0.03670\pm 0.02429$ mm (polyline) $0.03683\pm 0.02450$ mm (center)
14		Rosa et al. [93] (2014)	RBNN	Jackknifing (Leave one out)	25	cIMT	$0.065\pm 0.046$ mm
15		Molinari et al. [23] (2010)	FKMC	K13 (92% training, 8% testing)	200	cIMT	$0.054\pm 0.035$ mm (polyline)
16		Rosa et al. [110] (2015)	ANN + autoencoder	K2 (50% Training, 50% Testing)	55	cIMT	$0.0499\pm 0.0498$ mm
17		Biswas et al. [111] (2018)	FCN	K10 (90% training, 10% testing)	396	cIMT	$0.126\pm 0.134$ mm (DL1) $0.124\pm 0.10$ mm (DL2)
18		Elisa et al. [103] (2018)	FCN	K10 (90% training, 10% testing)	396	TPA	$20.52$ mm2 (DL1) $19.44$ mm2 (DL2)
19		Biswas et al. [104] (2020)	CNN + FCN	K10 (90% training, 10% testing)	250	cIMT, TPA value, TPA	$0.0935\pm 0.0637$ mm (cIMT error), $21.5794\pm 7.9975$ mm2 (DL-TPA), $2.7939\pm 2.3702$ mm2 (PA error)
20		Del Mar et al. [112] 2020	FCN	K10 (90% training, 10% testing)	331	cIMT error	$0.02$ mm