A new method for automatically segmenting brain tumor images has been developed. By improving how seed points are chosen and how the tumor grows, this method accurately identifies tumor boundaries. It uses a combination of MRI images to select seed points and creates a special texture feature to preserve tumor details. The resulting algorithm, based on region growing with intensity and spatial texture constraints, outperforms other methods in accuracy and computational efficiency. This method is a valuable tool for accurately segmenting brain tumor images.