Multimodal MRI Brain Tumor Image Fusion Pipeline | AI & Bioinformatics

This project focuses on medical image fusion using multi-modal brain MRI scans (T1ce, T2, FLAIR) from the BraTS2020 dataset. I built a full pipeline for preprocessing, fusion (PCA, Wavelet Transform, Laplacian Pyramid), and evaluation (SSIM, PSNR, Entropy, Mutual Information). The best-performing method (Sym4 wavelet + Max Rule) achieved SSIM: 0.82 and PSNR: 25.27 dB.

Tools Used:

Python, NumPy, MONAI, PyWavelets, Nibabel, Matplotlib, Scikit-Image

Contribution:

Image normalization and spatial preprocessing

Multiple fusion strategies (Weighted Avg, PCA, Wavelet)

Quality evaluation and visualization

Exported results for 150 patients

Ready-to-use metrics for research or diagnostic use

Outcome:

Produced clear, information-rich MRI images to assist doctors and researchers in brain tumor diagnosis.