Fractal and edge-based techniques for kidney enhancement and segmentation on Magnetic Resonance Images (MRI) / Alaá Rateb Mahmoud Al-Shamasneh

• Main Author: Alaá Rateb Mahmoud , Al-Shamasneh
• Format: Thesis
• Published: 2020
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://studentsrepo.um.edu.my/14425/1/Alaa_Rateb.pdf; http://studentsrepo.um.edu.my/14425/2/Ala%C3%A1_Rateb.pdf; http://studentsrepo.um.edu.my/14425/

Recently, many rapid developments in digital medical imaging have made further contributions to healthcare systems. However, the segmentation of regions of interest in medical images plays a vital role in assisting doctors in their medical diagnoses and for the early detection of disease. Since health issues related to the kidneys are increasing exponentially, this thesis focused on developing methods for the segmentation of MRI images of the kidney. Kidney images frequently suffer from low contrast, low resolution and noise, and are blur. Hence, it is necessary to enhance the images in order to improve the segmentation. Therefore, the current thesis focused on enhancing the fine details of the kidney region and the segmentation of the kidney images. To solve the above issues, the proposed work introduced a new model for enhancing low-contrast MRI kidney images based on fractional entropy. It is true that fractional entropy is able to handle complex situations such as images that are affected by the above challenges, and as such, the proposed work explored the same in this thesis to find solutions. However, sometimes, due to the presence of neighbouring organs and other regions in the background, the enhancement model must be one that can sharpen those details, thereby making the segmentation problem a challenging one. Therefore, this thesis was aimed at proposing a new method for kidney segmentation based on an active contour model driven by fractional-based energy minimization. Since the special characteristic of fractional calculus is its ability to preserve high-frequency contours regardless of contrast variations and noise, the proposed work explored this characteristic for the segmentation of kidney images. However, it should be noted that this method is said to be computationally expensive. Therefore, the thesis proposed a new method based on edge information for the segmentation of kidney images. It is true that the pixels representing the contours of the kidney share a unique spatial relationship. The proposed work used the same basis for the detection of the pixels in the edge domain, which represented the contours of the kidney in the enhanced images. Overall, this study made three contributions, namely, a fractional entropy-based method for the enhancement of kidney images, a fractional-based minimization function for kidney image segmentation, and an edge-based method for kidney image segmentation. The developed methods were tested on datasets using standard measures to evaluate the methods. The results of the proposed methods were compared with existing methods to show that the proposed methods are effective and useful.