Abstract
Abstract In contemporary Ethiopia, a notable surge in cancer-related fatalities, particularly attributed to brain tumor tissues, has become a concerning trend. The insidious nature of these tumors, often remaining undetected at their nascent stages, has led to a substantial loss of life. The early identification of such tumors presents a formidable challenge for medical professionals. This research endeavors to address this critical issue by introducing enhancement and detection techniques for brain tumors from medical MRI images. To augment image quality, we propose the adoption of a pioneering, nature-inspired algorithm known as the Water Cycle Algorithm (WCA) to mitigate complications encountered in grayscale medical MRI images. The WCA algorithm emulates the natural water cycle process, drawing inspiration from the flow of rivers and streams into the ocean. The method's foundation is rooted in the observation of the water cycle in nature. In this paper, we conduct a comparative analysis to underscore the efficacy of WCA in terms of image entropy and fitness, relative to established optimization methods. Furthermore, for tumor tissue detection and noise reduction in MRI images, we introduce a segmentation technique based on the Relevance Vector Machine (RVM). Image enhancement primarily centers on maximizing the information content through intensity transformation functions. We employ a parameterized transformation function that harnesses both local and global image data. To attain optimal image enhancement, we fine-tune the transformation function's parameters using the WCA. The parameters of the proposed Relevance Vector Machine are also harnessed for image segmentation. The results achieved through these methodologies align consistently and are assessed through performance graphs and image-based enhancement results. Simulation outcomes substantiate the superiority of the WCA-based image enhancement and the Relevance Vector Machine-based segmentation algorithm over conventional methods. Comparisons with other optimization techniques, such as Particle Swarm Optimization and Accelerated Particle Swarm Optimization, highlight the advantages of the WCA, while the segmentation comparison demonstrates the merits of the proposed Relevance Vector Machine over Support Vector Machine (SVM) approaches.