The present-day computer-aided diagnosis requires clustering, a segmentation technique that can link a single pixel to many clusters of the medical images. The existing clustering-based segmentation employs classical optimization technique, which may yield sub-optimal cluster centroids. This chapter endeavours to overcome the drawback of landing at sub-optimal cluster centroids by employing an improved Most Valuable Player Algorithm (MVPA). The MVPA was inspired from the social behaviour of athletes in sports games in grabbing the most valuable player trophy, and developed for solving optimization problems. This algorithm might fall into less-than-ideal pitfalls. By taking into account the bad players, the MVPA is improved to avoid sub-optimal solutions. The improved MVPA is then employed as an optimization tool in fuzzy-based clustering of medical images with a view of obtaining the global best cluster centroids for medical images. The developed algorithm is then studied on six medical images and the results are compared with those of classical approaches.