Kidney MRI Segmentation for Lesion Detection Using Clustering with Slime Mould Algorithm

Both the incidence and mortality rates of kidney cancer are increasing worldwide. Imaging examinations followed by effective systemic therapies can reduce the mortality rate. In this article, a new method to segment the kidney MRI for lesion detection is developed using a hard-clustering technique with Slime Mould Algorithm (SMA). First, a new partitional or hard clustering technique is developed using SMA which searches the optimal cluster centers for segmentation. In the preprocessing steps of the proposed method, the noise and intensity inhomogeneities are removed from the MR images as these artifacts affect the segmentation process. Region of Interests (ROIs) are selected and the clustering process is carried out using the SMA-based clustering technique. After the clustering, i.e., segmentation, the lesions are separated from the segmented images and finally, localized in the MR images as the postprocessing steps. The quantitative results are measured in terms of a well-known cluster validity index named Dunn-index and compared with that of the K-means algorithm. Both the quantitative and qualitative (i.e., visual) results show that the proposed method performs better than K-means.