Light Weighted Healthcare CNN Model to Detect Prostate Cancer on Multiparametric MRI

Table 2

Comparison of four MP-MRI information fusion methods.

Fusion method	Convolutional network	DSC	Sensitivity	Specificity	PPV
Input image concatenation (Mask-RCNN)	Resnet50	0.756	0.756	0.846	0.741
Feature map addition	Resnet50	0.751	0.741	0.845	0.756
Feature map concatenation	Resnet50	0.745	0.754	0.842	0.746
Se - feature map concatenation	Resnet50	0.714	0.753	0.823	0.716
Input image concatenation	Se-Resnet	0.723	0.785	0.863	0.723
Feature map addition	Se-Resnet	0.756	0.795	0.845	0.781
Feature map concatenation	Se-Resnet	0.785	0.712	0.854	0.763
SE-feature map concatenation (SE-Mask-RCNN)	Se-Resnet	0.756	0.715	0.856	0.752