Involvement of Machine Learning for Breast Cancer Image Classification: A Survey

Table 6

Neural Network for breast image classification.

Reference	Descriptor	Image type	Number of images	Key findings
Alharbi et al. [48]	49 features have been utilized.	Mammogram	1100	Five feature selection methods: Fisher score, Minimum Redundancy-Maximum Relevance, Relief-f, Sequential Forward Feature Selection, and Genetic Algorithm have been used.
				Achieved Accuracy, Sensitivity, and specificity are 94.20%, 98.36%, and 99.27%, respectively
Peng et al. [49]	Haralick and Tamura features have been utilized	Mammogram	322	Feature reduction has been performed by Rough-Set theory and selected 5 prioritized features.
				The best Accuracy, Sensitivity, and Specificity achieved were 96.00%, 98.60%, and 89.30%
Jalalian et al. [50]	GLCM	Mammogram		The obtained classifier Accuracy, Sensitivity, and Specificity are 95.20%, 92.40%, and 98.00%, respectively.
	Compactness
Li et al. [51]	Four feature vectors have been calculated	Mammogram	322	2D contour of breast mass in mammography has been converted into 1D signature.
				NN techniques achieved Accuracy is 99.60% when RMS slope is utilized.
Chen et al. [52]	Autocorrelation features	Ultrasound	242	The overall achieved Accuracy, Sensitivity, and Specificity are 95.00%, 98.00%, and 93%, respectively.
Chen et al. [53]	Autocorrelation features	Ultrasound	1020	The obtained ROC area is 0.9840 ± 0.0072.