Early diagnosis of a brain tumor may increase life expectancy. Magnetic resonance imaging (MRI) accompanied by several segmentation algorithms is preferred as a reliable method for assessment. This manuscript presented the performance analysis of clustered based and fusion-based segmentation techniques intended to detect the tumor from human brain MRI images in an efficient manner. Four primary steps are involved in this work such as pre-processing, clustering, segmentation, and fusion techniques. The main clustering methods such as K-means and fuzzy c-means (FCM) were first applied to the pre-processed MRI images, and then, the clustered images were segmented directly using the active contour segmentation techniques such as chan-vese (C-V) and level set method (LSM). Then in the next step, the clustered images were fused by using the non-sub sampled contour transform (NSCT) and convolution neural network (CNN) fusion methods, and then, the fused images were segmented by using the C-V and LSM segmentation methods again. The results of both clustered based and fusion-based segmentation in terms of structural similarity index measure (SSIM), dice coefficient (DC), computational time, sensitivity, precision, and segmentation accuracy revealed that CNN fusion-based C-V segmentation performs better than without fusion (clustered based or direct segmentation) to detect the tumors from the MRI images. The results indicate that C-V performs better with CNN as compared with the LSM. Finally, the fusion-based segmentation is an efficient approach to detect the tumor from the MRI images with minimal information loss and high segmentation accuracy over the clustered based segmentation.