Abstract

Hala Moustafa*, Metwally Kotb and Diana El-Sherif

ABSTRACT

Objective: As artificial intelligence methods for the diagnosis of disease advance, we aimed to evaluate machine learning in the predictive task of distinguishing between malignant and benign breast lesions on an independent clinical magnetic resonance imaging (MRI) dataset within a single institution for subsequent use as a computer aid for radiologists. Methods: Between December 2015 and December 2019, 60 female breast patients were enrolled in this study, in addition to 10 control female with a confirmed diagnosis of breast cancer and dense breasts underwent bilateral breast magnetic resonance imaging. In the physics MRI, is effective in the identification of additional masses in dense breasts that are not visualized on mammography. The results: Sensitivity and specificity are the two most important indicators in selection of medical imaging devices for cancer screening. Breast images taken by mammography, and MRI were collected from patients. The statistical features extracted from the histograms of the regions of interest revealed that MRI modality registered the highest scores, and ended with mammography, in the differentiation between normal, benign, and malignant breast tissues. They were then studied and compared for sensitivity and specificity results. The sensitivity and specificity, it is clear that, sensitivity increases on the expense of specificity, and vice versa. The data of this study, revealed that, both mammography and MRI has high sensitivity. In conclusion: The behavior and the general shape of the gray-level histogram describe specific behavior with each category of tissue, namely; normal, benign and malignant, because each modality gives specific shape for each imaged ROI of each tissue category.