Artificial neural network to aid differentiation of malignant and benign breast masses by ultrasound imaging

Abstract

The goal of this study is to evaluate an Artificial Neural Network (ANN) for differentiating benign and malignant breast masses on ultrasound scans. The ANN was designed with three layers (input, hidden and output layer), where a sigmoidal (hyperbolic tangent) response function is used as an activation function at each unit. Data from 54 patients with biopsy-proven malignant (N=20) and benign (N=34) masses were used to evaluate the diagnostic performance of the ANN. Of the seven quantitative features extracted from ultrasound images, only four showed statistically significant difference between the two categories. These features were margin sharpness, margin echogenicity, angular continuity, and age of patients. The diagnostic performance was evaluated by round-robin substitution to negate bias due to small sample size. All the input features were standardized to zero-mean and unit-variance to prevent non-uniform learning, which can generate unwanted error. The outputs of the network were analyzed by Receiver Operating Characteristics (ROC). The resulting area under the ROC curve A z was 0.856 ± 0.058 with 95% confidence limit from 0.734 to 0.936, providing 76.5% specificity at 95% sensitivity. The performance of the ANN was comparable to the performance by logistic regression analysis reported by our group earlier. These results suggest that an ANN when combined with sonography can effectively classify malignant and benign breast lesions.

Automated breast cancer detection and classification using ultrasound images: A survey

“…Sometimes, other information can be integrated to help the diagnosis. Patient's age is proved to be an effective feature to diagnose malignancy [14,17,122,123]. Also, family disease history is another useful feature for the diagnosis.…”

“…As much as 65-90% of the biopsies turned out to be benign, therefore, a crucial goal of breast cancer CAD systems is to distinguish benign and malignant lesions to reduce FPs. Many techniques such as linear discriminant analysis (LDA), support vector machine (SVM) and artificial neural network (ANN) [5,10,17,18,20] have been studied for mass detection and classification. Most of the CAD systems need a large number of samples to construct the models or rules, but [22] proposed a novel diagnosis system requiring very few samples.…”

Phase portrait analysis for automatic initialization of multiple snakes for segmentation of the ultrasound images of breast cancer

“…Numerous reviews (e.g., 1 [1,2]) present methods to treat segmentation of medical images as a general image processing problem, while others use a priori information relevant to the specific type of the images. Conventional segmentation methods include thresholding [3][4][5][6][7], neural networks [8][9][10][11][12][13][14][15], mode-based methods (such as expectation-maximization) [16,17], clustering [18,19], region growing [20], deformable active contours (snakes) [3,[21][22][23][24][25] and level set methods [26]. The segmentation is usually followed by feature extraction to distinguish malignant and benign masses.…”

Ultrasound based computer aided diagnosis of breast cancer: Evaluation of a new feature of mass central regularity degree

“…Sonograph in breast cancer diagnosis field has been used for differentiating solid from cyst masses. The role of ultrasound image has been expanded by improving the quality of the image and now ultrasounds have become a complementary test to mammographs for differentiating benign from malignant masses (Lee et al, 2008;Song et al, 2005). Unlike mammography, ultrasound can deal with dense breast tissue.…”

“…The A z values achieved by the study were between 0.8 and 0.86. Also, Artificial Neural Network (ANN) has been evaluated Song et al(2005) using age and three ultrasound features (margin sharpness, intensity of absorbed sound waves by the mass margin and angular continuity of the margin). The study obtained an accuracy 0.856 ± 0.058 under ROC curve.…”