BRAIN TUMOUR DETECTION: ASSESSMENT AND COMPARISON OF PERFORMANCE METRICS USING ARTIFICIAL INTELLIGENCE.

Abstract

This paper focuses on the essential issue of effective and precise identification of the brain tumors through MRI analysis. The main study goal was to design a CAD system that increases the efficiency of tumour segmentation and decreases diagnostic time with the help of the altered K-means clustering algorithm. As regarding the methodology, it entailed gathering of multiple MRI datasets, initialization of the images, and applying the adaptive K-means algorithm with subsequent enhancements achieved through the use of conventional classifiers, as well as the segmentation techniques. To compare the proposed method with the standard adaptive K-means method, distance variance, distance mean, accuracy, area, and perimeter were calculated. The results revealed better values: the accuracy of the proposed system – 80. 13% compared to 52. 75% of the traditional technique, as well as more accurate tumor contouring. The rationality of the proposed system in enhancing the segmentation capability was justified and verified through the visual and statistical analysis of the results, which point towards the framework’s ability to transform the diagnostic and treatment approaches to brain tumor patients. In conclusion, the study justifies the approach with establishing the possibility and directions for enhancing diagnostic accuracy and efficiency in clinical neuro-oncology with the help of the integration of advanced computational strategies with the basic clustering algorithms.