Table 6.

Summary of classical machine learning-based methods.

Study	Year	Datasets	No. of Patients	Performance						Image Modalities	Time/ Equipment	Category
				Dice (%)	ASD (mm)	HD (mm)	Jac (%)	FPVF/FNVF (%)	MSE/RMSE (mm)
Ji [64]	2013	In-house	12	$97.90\pm 1.10$	$0.20\pm 0.13$	—	$95.80\pm 2.00$	—	—	MRI	128 s/CPU	CML
Wang [65]	2015	In-house	30 60	$94.00\pm 2.00$ $95.00\pm 2.00$	$0.42\pm 0.15$ $0.33\pm 0.11$	$0.74\pm 0.25$ $0.41\pm 0.20$	—	—	—	CBCT CT	20 min/—	CML
Udupa [66]	2014	In-house	15	—	—	$3.30\pm 0.56$	—	$1.00\pm 0.00$/ $49.00\pm 8.00$	—	MRI	54 s/CPU	CML
Linares [19]	2019	In-house	16	$92.88$	—	—	$86.48$	—	—	CBCT	5 min/CPU	CML
Barandiaran [67]	2009	In-house	12	—	—	—	—	—	—	CT	10 s/CPU	CML
Orbes-Arteaga [68]	2015	PDDCA	40	$93.08\pm 2.36$	—	—	—	—	—	CT	—	CML
Wang [69]	2016	PDDCA	48	$94.40\pm 1.30$	$0.43\pm 0.12$	—	—	—	—	CT	—/CPU	CML
Qazi [70]	2010	In-house	25	$90.19$	—	—	—	—	—	CT	3 min/CPU	CML
Torosdagli [71]	2017	PDDCA	40	$91.00$	—	<1.00	—	—	—	CT	—/CPU	CML
Wu [72]	2018	In-house	216	$89.00$	—	$1.60$	—	—	—	CT	—	CML
Tam [36]	2018	In-house	56	$85.20\pm 5.30$	—	—	—	—	$0.10$/$3.16$	CT	1 s/CPU	CML
Tong [73]	2018	In-house	246	—	—	—	—	—	—	CT	—	CML
Wu [74]	2019	In-house	216	$89.00$	—	—	—	—	—	CT	30 s/CPU	CML
Gacha [75]	2018	PDDCA	30	$80.49$	—	—	—	—	—	CT	—	CML
Spampinato [76]	2012	In-house	10	—	—	—	—	—	—	CT	—	CML