Table 1. Evaluation of prostate cancer aggressiveness prediction using the proposed methods and comparison methods. Means and standard deviations are presented. The highest values are denoted in bold.

Image configurations	Self-supervised learning set	Accuracy(%)	Sensitivity(%)	Specificity(%)	PPV(%)	NPV(%)	AUC
T2wMR	ImageNet+ProstateX	65.00±8.47	58.93±12.40	70.00±15.92	64.29±14.89	67.63±6.78	0.68±0.11
DWI		60.64±9.57	52.86±13.66	67.06±19.55	61.74±18.75	63.00±7.67	0.60±0.10
ADCmap		62.26±9.98	44.64±25.16	76.76±17.51	61.63±18.04	64.53±10.27	0.55±0.11
Image fusion		62.42±7.57	46.43±22.9	75.59±15.32	62.52±12.6	64.76±10.30	0.59±0.08
Feature fusion		63.39±4.24	56.07±8.92	69.41±8.79	60.84±7.09	65.92±3.97	0.63±0.05
Prediction fusion (AV)		67.26±5.21	49.64±9.74	81.77±10.17	70.96±10.99	66.47±3.97	0.66±0.05
Prediction fusion (MV)		69.68±6.17	54.64±10.66	82.06±8.93	72.45±10.42	68.93±5.27	0.68±0.06
T2wMR	ImageNet+ProstateX+SNUBPCa	65.32±6.46	64.64±18.86	65.88±13.46	61.53±7.67	71.51±10.58	0.67±0.10
DWI		61.94±5.70	54.64±21.30	67.94±15.41	59.26±9.66	66.17±7.46	0.62±0.08
ADCmap		63.55±3.96	51.07±8.43	73.82±8.02	62.3±6.18	64.81±3.66	0.61±0.04
Image fusion		62.58±6.84	52.14±18.07	71.18±18.75	62.85±11.34	65.15±6.97	0.63±0.06
Feature fusion		63.71±4.70	57.14±15.25	69.12±13.53	64.24±6.70	67.06±5.50	0.64±0.06
Prediction fusion (AV)		67.90±5.66	56.07±7.72	77.65±10.30	68.51±9.78	68.22±3.85	0.67±0.05
Prediction fusion (MV)		71.29±5.63	58.93±7.39	81.47±8.88	73.29±9.09	70.70±4.36	0.70±0.05