Image segmentation is a vital part of many applications because it makes possible for the information extraction and analysis of image contents. For image segmentation process, many approaches have been proposed earlier. The image segmentation using normal standard methods is well only for simple image contents. In existing approaches image segmentation is done through multi-resolution stochastic level set method (MSLSM), but the topology changes are undesirable and it presented nonparametric topology-constrained segmentation model. To improve the image segmentation more effective, in this work, we plan to do image segmentation using level set method with geodesic active contour to analyze medical image disease diagnosis. With level set segmentation based on geodesic active contour, active cluster objects are segregated. Cluster object formation is done with fuzzification of growing active contours with the previously known contours of diseased image portions. With segment portions new similar regions can be traced out with automatic seeded growing method. Experimentation is conducted with bench mark data sets obtained from UCI repository and proved that the proposed GAC work will be 80% efficiency for image segmentation compared to an existing MSLSM. The parametric evaluations are carried over in terms of segment size and contour growth, Contour objects in the cluster, Similarity ratio of known and unknown contours, Seed size of the detected contours.