Machine Learning in Ultrasound Computer-Aided Diagnostic Systems: A Survey

Table 1

The performance summary of breast ultrasound CAD system.

Reference	Dataset	Features	Classifiers	Performance
[14]	88 benign; 90 malignant	Textural features; +; morphologic features	ANN (BPNN)	Accuracy: 95.86%; Sensitivity: 95.14%; Specificity: 96.58%
[15]	70 benign; 50 malignant	Textural features; +; morphologic features	SVM	Accuracy: 95.83%; Sensitivity: 96%; Specificity: 95.71%
[16]	4254 benign; 3154 malignant	GoogLeNet		Accuracy: 91.23%; Sensitivity: 84.29%; Specificity: 96.07%
[17]	135 benign; 92 malignant	Boltzmann; machine		Accuracy: 93.4%; Sensitivity: 88.6%; Specificity: 97.1%
[18]	275 benign; 245 malignant	Stacked denoising; Autoencoder; (SDAE)		Accuracy: 82.4%; Sensitivity: 78.7%; Specificity: 85.7%
[19]	100 benign; 100 malignant	Deep polynomial network	SVM	Accuracy: 92.40%; Sensitivity: 92.67%; Specificity: 91.36%