Detecting MRI-Invisible Prostate Cancers Using a Weakly Supervised Deep Learning Model

Table 2

MIPCa detection performance of WSUNet.

PI-RADS	Training set		Testing set
	AUC	Sensitivity	AUC	Sensitivity
All (1-5)	0.798 (0.775-0.819)	0.817 [272/333]; (0.781-0.850)	0.764 (0.728-0.798)	0.797 [94/118]; (0.737-0.856)
≥3	0.794 (0.773-0.815)	0.814 [263/323]; (0.780-0.848)	0.762 (0.725-0.795)	0.803 [94/117]; (0.744-0.863)

Note. Data in parentheses are 95% CIs. AUC = area under the receiver operating characteristic curve. Data are percentages, with number of participants in brackets. PI-RADS refers to the maximum PI-RADS for all outlined lesions in one examination. Since most of the patient’s PI-RAD is greater than 3, so this part of the data has been presented alone. MIPCa detection refers to the prediction of systematic biopsy.