Breast DCE-MRI Segmentation for Lesion Detection by Multilevel Thresholding Using Arithmetic Optimization Algorithm

Abstract

Patra, Dipak Kumar Si, Tapas Mondal, Sukumar Mukherjee, PrakashIt is becoming more common to employ Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) to diagnose breast disorders. DCE-MRI, which is assisted by Computer-Aided Design (CAD) methods, shows its effectiveness in detecting breast cancer. In image segmentation, multilevel thresholding is an important and easy to implement technique. This paper proposes breast DCE-MRI segmentation by multilevel thresholding using the Arithmetic Optimization Algorithm (AOA). First, the anisotropic diffusion filter is used to denoise MR images, and then, the correction of Intensity Inhomogeneities (IIHs) is performed. The lesions are then retrieved from the segmented images and located in the initial MR images. 50 Sagittal T2-Weighted DCE-MRI images are used to test the suggested approach. Particle Swarm Optimizer (PSO) and Hidden Markov Random Field (HMRF) are compared to the suggested AOA technique. The devised technique achieves a high level of accuracy of 99.80%, sensitivity of 98.06%, and Dice Similarity Coefficient (DSC) of 85.52%. The devised method outperforms the two compared methods.