Brain Tumor Segmentation

Academic research project conducted at SRM University AP under the supervision of Dr. Anabik Pal (Sep 2023 – Dec 2023). Focused on automated segmentation of brain tumors across multi-modal MRI scans (FLAIR, T1, T1ce, T2) using deep convolutional neural networks.

Formulated preprocessing pipelines (skull stripping, N4 bias field correction, intensity normalization), developed custom U-Net architectures with residual skip connections, and optimized combined loss functions (Dice + Focal Loss).

Python, PyTorch/TensorFlow, OpenCV, and medical imaging libraries (Nibabel, SimpleITK). Trained on the multimodal BraTS benchmark dataset with spatial data augmentation.

Achieved high segmentation fidelity across necrotic cores and active tumor structures, validating deep learning as an effective diagnostic aid for medical imaging.