Table 3.

Results of optimizing network and network hyperparameters using the particle swarm algorithm (discussed in the supplement).

	LR	M	Batch Size	γ	α	Accuracy	AUC
VGG	0.009	0.73	10	1.560	0.6	84.7%	0.896
CT Test						88.3%	0.901

Computed tomography angiography (CTA) test performance with the common network having been retrained only with the entire magnetic resonance imaging (MRI) dataset for 500 epochs. LR, learning rate; M, momentum; AUC, Area under receiver operator characteristic curve.