MRI to CT Style Transfer Using Generative Adversarial Networks

This thesis explores image synthesis and style transfer from MRI to CT images using state-of-the-art Generative Adversarial Network (GAN) architectures, specifically CycleGAN and Pix2Pix. The study focuses on optimizing these models through data normalization, parameter fine-tuning, and architectural enhancements. Performance is evaluated using key metrics such as Mean Absolute Error (MAE), Structural Similarity Index Measure (SSIM), and Peak Signal-to-Noise Ratio (PSNR). The optimized model achieves competitive results, with an SSIM value surpassing benchmarks for the dataset and MAE and PSNR values remaining strong. This work lays the foundation for further advancements and practical applications in the field.