This is a source code for classifying MRI scans of brain images if it has a tumour, and segmenting out the tumour if any exists.
- tumour_detection.mlx : Main Interface of programme
- classification.mlx : Experiment with different values for training tree
- Script files (.m) : Functions used in the programme
- Excel files (.xlsx) : Values calculated using feature extraction to train classification model
- Image Processing Toolbox
- Fuzzy Logic Toolbox
- Main Interface is in the file tumour_detection.mlx. For a cleaner interface, hide the code. Only the output will be shown.
- Change the path to the input image in line 1 of the file.
- To allow the script to find thresholds or other parameters automatically,
each of the parameters MUST be set to the following values:
- threshold for skull stripping must be set to 0
- levels for Otsu's thresholding must be set to 1
- percentage for Watershed Algorithm must be set to 0 The recommended number of clusters for Fuzzy C-Means Algorithm is 4.
- Run the script.
- If any of the results of skull stripping, Otsu's thresholding, Fuzzy C-Means algorithm and Watershed algorithm is not satisfactory, use the slider to change the parameters to perhaps get a better results.
Amber Song Xin Ying (29612330) - Morphological Operations, Feature Extraction & Classification Ho Yuan Ai (29566061) - Segmentation (Otsu, Fuzzy C-Means & Watershed) & Combination of results Liow Gian Hao (30666910) - Pre-processing, Skull stripping
Dr. Anuja and Ms. Najini for their supervision and help throughout the project.