TABLE 5.

Mean Accuracy Values and 95% Confidence Intervals of the Three Worst‐Performing Models Along the 10‐Fold Cross‐Validation for Each Sequence

		Accuracy (mean ± 95% CI) Individual Slice			Accuracy (mean ± 95% CI) Entire Sequence
Sequence	CNN Architecture	Global	Q0 Class	Q1 Class	Global	Q0 Class	Q1 Class
T2WI	SqueezeNet1‐1	71.07 ± 0.41	43.86 ± 0.54	88.68 ± 0.47	62.50 ± 0.00	0.00 ± 0.00	100.00 ± 0.00
	VGG16‐BN	72.11 ± 0.74	46.48 ± 0.82	88.66 ± 0.63	87.50 ± 0.29	66.67 ± 0.32	100.00 ± .00
	SqueezeNet1‐0	72.65 ± 0.61	57.03 ± 0.44	82.75 ± 0.75	87.50 ± 0.26	66.67 ± 0.28	100.00 ± 0.00
DWI	MobileNet‐V2	52.73 ± 0.45	55.12 ± 0.37	50.74 ± 0.51	55.57 ± 0.21	25.00 ± 0.17	80.00 ± 0.24
	MNASNet0‐5	59.12 ± 0.63	19.18 ± 0.61	91.68 ± 0.66	55.57 ± 0.18	0.00 ± .00	100.00 ± 0.00
	AlexNet	59.95 ± 0.66	14.09 ± 0.75	97.50 ± 0.62	66.67 ± 0.37	25.00 ± 0.00	100.00 ± 0.00
ADC	MNASNet0‐5	52.46 ± 0.76	24.51 ± 0.73	85.36 ± 0.74	53.86 ± 0.00	0.00 ± 0.00	100.00 ± 0.00
	ResNet28	54.25 ± 0.63	30.78 ± 0.65	81.90 ± 0.63	76.92 ± 0.39	50.00 ± 0.46	100.00 ± 0.00
	AlexNet	56.67 ± 0.69	40.43 ± 0.61	75.66 ± 0.76	84.63 ± 0.35	66.67 ± 0.32	100.00 ± 0.00
DCE	ResNet50	71.92 ± 0.74	85.81 ± 0.83	60.50 ± 0.72	89.00 ± 0.00	0.00 ± 0.00	100.00 ± 0.00
	VGG16‐BN	77.31 ± 0.85	79.71 ± 0.77	75.32 ± 0.85	92.85 ± 0.24	33.33 ± 0.25	100.00 ± 0.00
	GoogLeNet	78.55 ± 0.71	89.54 ± 0.66	69.45 ± .70	96.39 ± 0.32	66.67 ± 0.33	100.00 ± 000