Research Progress of Machine Learning and Deep Learning in Intelligent Diagnosis of the Coronary Atherosclerotic Heart Disease

Table 3

AI applications in intravascular imaging.

Application	Method	Tasks	Data	Measure	Value	Calculate time or cost	Paper
IVUS	ML	Lumen image segmentation	435	Jaccard measure		—	[36]
		Prediction of progression to vulnerable plaque	748	Accuracy	91.47%	—	[37]
	DL	Plaque image segmentation	12325	AUC	0.911	3,584 CUDA cores and 12GB of GPU memory	[38]
		Lumen image segmentation	435	Jaccard measure	0.869	Run in 0.03 seconds	[39]
		Extraction of coronary plaque parameters and prediction of functional parameters	1328	AUC	0.84-0.87	—	[40]
IVOCT	ML	Plaque image segmentation and composition classification	300	Accuracy		Under 4 seconds when run on a standard 12-core CPU	[41]
	DL	Fully automated semantic segmentation of plaques	4892	Sensitivity/specificity	87.4%/89.5%; 85.1%/94.2%	0.27 seconds of each image	[43]
		Feature extraction and classification of fibroatheromas	360	Accuracy	76.39%	—	[44]