In the field of medical images, automatic segmentation of ROI region accuracy plays the most important role. For tumor segmentation, a network should be optimized with precise weights for the respective features. Till yet researchers used many features like intensity and area with various optimization techniques. Accuracy mostly depends upon the features. It is directly proportional to the number of features. Our proposed algorithm going to use energy feature like LESH with BAT colony optimization to enhances the optimization problem for weights. The proposed work will include all the features of time and spatial domain so is more sensitive to optimize the weights in spite of using only time domain features.