فهرست نشریات علمی دسترسی آزاد ایران

Background and Objectives: Accurate segmentation of breast tumors in Magnetic Resonance Imaging (MRI) is crucial for early detection, diagnosis, and treatment planning in breast cancer management. Optimizing thresholding and segmentation techniques can significantly improve the accuracy of tumor delineation. Thus, this study aims to develop and evaluate a novel approach for breast tumor segmentation in MRI images.   Methods: The data were collected from the RIDER Breast MRI public dataset. We propose a framework consisting of: (i) 3D to 2D MRI image conversion, (ii) BA-optimized tri-level thresholding using Otsu and Kapur functions, (iii) tumor segmentation using Watershed and Chan-Vese techniques, and (iv) performance evaluation against ground truth. The study utilized 250 2D MRI slices from 5 volunteers, obtained from the RIDER database, across axial, coronal, and sagittal planes.   Results: The proposed method achieved an overall segmentation accuracy exceeding 95% when tested on the benchmark breast MRI slices. The BA-optimized thresholding effectively enhanced image quality, facilitating improved tumor region extraction by both Watershed and Chan-Vese segmentation algorithms. Comparative analysis revealed specific strengths of each segmentation technique across different MRI planes.   Conclusion: Our integrated approach demonstrates high accuracy in breast tumor segmentation from 2D MRI slices. The combination of BA-optimized tri-level thresholding with Watershed and Chan-Vese segmentation techniques shows promise for improving the precision of breast tumor delineation. This method could potentially enhance the accuracy of breast cancer diagnosis and treatment planning. Further research with larger datasets and comparison with other state-of-the-art techniques is needed to validate these findings.