Breast Masses in Dynamic Contrast-Enhanced Magnetic Resonance Imaging Classification Based on Combining Deep Learning and Local Binary Pattern Features
Rehab Kadhim, Hussain S. Hasan, Ali Majeed Hasan · 2024
The project’s objective is to create a novel method for separating benign from malignant tumors in breast DCE-MRI scans. The proposed local binary pattern (LBP), kinetic curve, and convolutional neural network (CNN) techniques were used to extract textural information from magnetic resonance imaging (MRI). After that, the traits were categorized as benign or malignant using the support vector machines (SVMs) model. The achieved classification accuracy was 97.3%.