Published Paper Details:

Because of the complicated structure of brain tumours, indistinct boundaries, and external influences such as noise, inferring tumour and edoema areas from brain magnetic resonance imaging (MRI) data remains difficult. This work proposes an effective hybrid clustering technique combined with morphological procedures for segmenting brain tumours to reduce noise sensitivity and increase segmentation stability. The following are the paper's main contributions: To begin, adaptive Wiener filtering is employed for denoising, then morphological processes are used to remove non-brain tissue, effectively lowering the method's noise sensitivity. Second, to segment images, K-means++ clustering is merged with the Gaussian kernel-based fuzzy C-means method. This clustering enhances the algorithm's stability while simultaneously lowering the sensitivity of the clustering settings. Finally, to acquire realistic representations of brain tumours, the retrieved tumour images are post processed utilising morphological procedures and median filtering. The suggested technique was also compared to other existing segmentation algorithms. The suggested algorithm outperforms the competition in terms of accuracy, sensitivity, specificity, and recall.