Table 1.

Comparison of prostate segmentation results with the literature

	2D/3D	Modalities	Dataset size	Approach	Reported performance	Prostate segmentation	Registration	Validation protocol
Rampun et al. [30]	2D	Mono-modal	45	Engineered features	0.93 AUC	Manual	N/A	9-fold CV
Lemaitre	3D	Multi-modal	17a	Engineered	0.834 AUC	Manual	Automatic	Leave one out
et al. [19, 21]				features			(elastic)
Trigui et al. [41]	2D	Bi-modal	34a	Engineered	0.72 sensitivity	Manual	N/A	10-fold CV
				features	0.88 specificity
Wang and	3D	Multi-modal	17a	Texton	0.884 AUC	Manual	Automatic	Leave one out
Zwiggelaar [46]				dictionary			(rigid)
Kohl et al. [17]	2D	Bi-modal	152	GANs	0.55 sensitivity 0.98 specificity	Implicit	N/A	4-fold CV
Kiraly et al. [16]	2D	Multi-modal	202	Convolutional encoder-decoder	0.843 AUC	Automatic (elastic)	Level-set	5-fold CV
Yang et al. [50]	2D	Bi-modal	360	Dual path CNN	0.89 sensitivity	Implicit	Implicit	5-fold CV

^a From the I2CVB dataset