Table 2.

Number of correctly predicted bladder cancer treatment response assessment of the test set at an operating point determined using the training set.

	DL-CNN	RF-SL	RF-ROI	Radiologist 1	Radiologist 2
Complete Response (Sensitivity)	6/12 (50%)	6/12 (50%)	8/12 (66.7%)	11/12 (91.7%)	11/12 (91.7%)
Non-complete Response (Specificity)	34/42 (81.0%)	33/42 (78.6%)	23/42 (54.8%)	18/42 (42.9%)	16/42 (38.1%)

DL-CNN: Deep-learning convolution neural network. RF-SL: Radiomics features extracted from segmented lesions. RF-ROI: Radiomics features extracted from pre- and post-treatment paired ROIs.