DL-MRI: A Unified Framework of Deep Learning-Based MRI Super Resolution

Table 3

PSNR/SSIM of MRI dataset on different networks.

Dataset	Scale	Bicubic	RDN	EDSR	SRGAN	D-DBPN	SRFBN
Head and neck	×2	25.44/0.6928	29.05/0.8824	37.34/0.9861	37.69/0.9868	35.74/0.9800	38.03/0.9879
	×3	23.04/0.6289	27.41/0.8326	31.14/0.8693	—/—	31.61/0.9474	32.79/0.9595
	×4	21.58/0.5442	24.53/0.7144	29.25/0.9164	28.78/0.9093	29.28/0.9182	30.23/0.9304
Breast	×2	28.56/0.8848	33.50/0.9373	34.57/0.9447	34.06/0.9332	31.94/0.9243	34.28/0.9428
	×3	21.85/0.6130	30.61/0.8876	31.25/0.8949	—/—	28.56.0.8568	32.14/0.9040
	×4	20.50/0.4830	23.68/0.6118	30.00/0.8631	27.99/0.8137	28.13/0.8351	30.68/0.8735
Knee	×2	30.60/0.9185	36.34/0.9641	37.49/0.9719	36.93/0.9670	37.67/0.9727	37.95/0.9744
	×3	24.36/0.6202	32.13/0.9103	32.85/0.9230	—/—	33.75/0.9360	33.82/0.9377
	×4	22.69/0.5649	30.02/0.8597	28.34/0.8968	29.63/0.8643	31.29/0.8940	31.35/0.8964
Skull	×2	20.320.6025	38.05/0.9565	38.03/0.9565	35.80/0.9234	33.82/0.9062	39.19/0.9600
	×3	19.23/0.5843	28.81/0.7790	34.16/0.9211	—/—	31.38/0.8620	35.65/0.9300
	×4	18.60/0.5623	29.70/0.7962	32.22/0.8886	30.95/0.8500	29.39/0.7906	33.60/0.9089