AI-Powered Enhancement of CBCT for Brain Adaptive Radiotherapy

AI-Powered Enhancement of CBCT for Brain Adaptive Radiotherapy (Graduation Project)

Sep 2024 – Jan 2025

Collaborated with Baheya Institute for clinical data and Bibliotheca Alexandrina for Linux supercomputer resources.

Compared GAN models (pix2pix, CycleGAN, MedSR-GAN) using PyTorch, TensorFlow, and MLflow.

Applied volumetric registration using B-spline interpolation and Mattes Mutual Information Loss.

Tuned the final model architecture to improve PSNR from 22.3 to 26.4 dB and SSIM from 0.72 to 0.85.