Classification of Mammograms Using Sigmoidal Transformation and SVM

Breast cancer: Three‐class masses classification in mammograms using Apriori dynamic selection

“…Machine learning‐based classification models are widely used for breast masses' classification as they give good results and do not require large mammogram datasets. In literature, breast masses' classification can be classified in three categories, a binary classification, 13,18,19,22,23,41 a two‐stage multi‐class classification (normal/ abnormal then benign/malignant), 12,20 or a one‐stage multi‐class classification (normal/benign/malignant) 42 . The ML‐based classification approaches require a feature extraction step prior to the classification process to obtain the characteristics of the suspicious masses 12,13,18–20,22,23,41,42 .…”

“…In literature, breast masses' classification can be classified in three categories, a binary classification, 13,18,19,22,23,41 a two‐stage multi‐class classification (normal/ abnormal then benign/malignant), 12,20 or a one‐stage multi‐class classification (normal/benign/malignant) 42 . The ML‐based classification approaches require a feature extraction step prior to the classification process to obtain the characteristics of the suspicious masses 12,13,18–20,22,23,41,42 . Most of the ML existing techniques use either texture 13,18,20 or shape 22,23 features to classify suspicious regions in the breast as they describe most of the grayscale variations of breast masses intensity values and their corresponding margins.…”

“…42 The ML-based classification approaches require a feature extraction step prior to the classification process to obtain the characteristics of the suspicious masses. 12,13,[18][19][20]22,23,41,42 Most of the ML existing techniques use either texture 13,18,20 or shape 22,23 features to classify suspicious regions in the breast as they describe most of the grayscale variations of breast masses intensity values and their corresponding margins. However, and as previously presented in Tables 3 and 4, both shape and texture features need to be combined for better binary and multi-class classifications' results.…”

“…The best classification results were achieved using the hybrid KNN and RBFSVM base estimator. Vikrant et al 19 also proposed a binary classification technique of suspicious breast masses in digital mammograms into normal and abnormal.…”

“…The best classification results were achieved using the hybrid KNN and RBFSVM base estimator. Vikrant et al 19 also proposed a binary classification technique of suspicious breast masses in digital mammograms into normal and abnormal. In the first stage, preprocessing, suspicious masses were extracted from full mammograms and were enhanced using a sigmoidal transformation function.…”