Brain Tumor Segmentation

MATLAB-based automated brain tumor segmentation pipeline from MRI scans, combining classical image processing and deep learning methods.

Brain Tumor Segmentation is a MATLAB-based pipeline for automated detection and delineation of brain tumors from MRI images, developed as part of the biomedical signal and image processing curriculum at KUET.

Pipeline Overview

  1. Pre-processing: MRI volume loading, skull stripping, intensity normalization, and noise reduction.
  2. Segmentation: Combination of classical methods (thresholding, morphological operations, region growing) with CNN-based feature extraction for tumor boundary delineation.
  3. Post-processing: Connected component analysis and volume quantification.
  4. Visualization: 2D slice-by-slice overlay of segmentation masks with tumor volume estimation.

Tools & Technologies

  • MATLAB Image Processing Toolbox
  • Deep Learning Toolbox (CNN feature extraction)
  • BraTS dataset format for benchmarking

Outcome

Produced accurate tumor boundary masks that assist in pre-surgical planning and treatment response monitoring. Demonstrated the value of combining classical morphological operations with learned features.