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

Table 3

The performance summary of thyroid ultrasound CAD system.

Reference	Dataset	Features	Classifiers	Performance
[25]	48 benign; 223 malignant	Textural features	Decision tree: C4.5	Accuracy: 94.3%
[26]	10 benign; 10 malignant	Textural features	AdaBoost	Accuracy: 100%; Sensitivity: 100%; Specificity: 100%
[27]	71 benign; 357 malignant	GoogLeNet		Accuracy: 99.13%; Sensitivity: 99.70%; Specificity: 95.80%
[28]	465 normal; 9957 thyroid nodular lesions	CNN (15 convolutional layers)	CNN (4 convolutional layers)	AUC: 0.986