A comparison of methods for three-class mammograms classification

Abstract

Obtained experimental results indicate that the proposed three-class SVM classifier is more suitable for practical use than the other two methods.

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

“…The proposed ML method gives good results in multi‐classifying masses in comparison with the existing works 21,46 and achieves accuracy, F 1‐score, precision, recall, specificity, and AUC scores of 75.42%, 76.39%, 75.18%, 75.67%, 87.71%, and 0.819, respectively.…”

“…In addition, our binary classification scores surpass those of recent works that use deep learning models to classify breast lesions in digital mammograms. [43][44][45] The proposed ML method gives good results in multi-classifying masses in comparison with the existing works 21,46 and achieves accuracy, F1-score, precision, recall, specificity, and AUC scores of 75.42%, 76.39%, 75.18%, 75.67%, 87.71%, and 0.819, respectively.…”

“…Last, the obtained feature vectors were given to a modified SVM classifier, which worked in two‐stage by classifying the masses into normal and abnormal, then into benign and malignant. Milosevic et al 21 also proposed a multi‐class classification of suspicious breast masses into normal, benign, or malignant. The GLCM features were extracted first then were classified using SVM, KNN, and Naïve Bayes classifiers.…”

DNA Methylation as a Noninvasive Epigenetic Biomarker for the Detection of Cancer

“…Currently, there are numerous noninvasive techniques for the detection of specific cancer types including computed tomography (CT) in lung cancer, mammography in breast cancer, or positron emission tomography (PET) scan and digital rectal examination (DRE) for prostate cancer [ 4 – 6 ]. Many imaging techniques have a high sensitivity for the detection of abnormal neoplasms but lack the capacity to classify tumour subtypes or distinguish benign from malignant tumours [ 7 ].…”

“…Many imaging techniques have a high sensitivity for the detection of abnormal neoplasms but lack the capacity to classify tumour subtypes or distinguish benign from malignant tumours [ 7 ]. In breast cancer screening, the use of mammography is limited to the detection of larger tumours, resulting in the neglect of smaller tumours [ 8 ], and diagnosis is dependent on the level of expertise possessed by the clinician [ 6 ]. In prostate cancer, as part of the physical examination of the prostate gland, the DRE is semi-invasive and can cause injury and bleeding to the patient [ 4 ].…”