ROI-Guided Brain MRI Segmentation and Tumor-Region Visualization Using MATLAB

Abstract A MATLAB workflow and graphical user interface (GUI) are presented for region-of-interest (ROI)-guided segmentation and visualization of a suspected tumor region in brain magnetic resonance imaging. The method is applied to one axial FLAIR image from the REMBRANDT collection. A normalized 5 × 5 binomial kernel smooths the image. Central-difference filters provide a separate edge-magnitude view, while three fixed ROIs guide intensity-interval thresholding. A brightness cutoff and coordinate-based exclusions refine pixel selection. Binary hole filling is then applied to the mask used for the boundary overlay. The mathematical formulation defines each processing stage and its role in constructing the segmentation mask. The GUI displays the input, smoothed image, edge magnitude, isolated region, and boundary overlay in one window. These outputs show the selected region both in isolation and within the surrounding anatomy. The implementation brings image processing, visualization, and user interaction into a single MATLAB workflow.