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

Table 2

The performance summary of liver ultrasound CAD system.

Reference	Dataset	Features	Classifiers	Performance
[20]	50 normal; 50 fatty liver disease (FLD)	Textural features	ANN	Accuracy: 98%; Sensitivity: 100%; Specificity: 96%
[21]	15 normal; 16 cirrhotic; 25 hepatocellular carcinoma (HCC)	Textural features	SVM	Accuracy: 88.8%
[22]	44 cyst; 18 hemangioma; 30 HCC; 16 normal	Sparse autoencoder		Accuracy: 90.50%; Sensitivity: 91.60%; Specificity: 88.50%
[23]	79 normal; 89 early-stage fibrosis; 111 late-stage fibrosis	VGGNet	FCN	Accuracy: 93.90%; Sensitivity: 88.6%; Specificity: 97.1%
[24]	47 cirrhosis; 44 normal	CNN	SVM	Accuracy: 86.9%