Title: Generation of Cardiac MRI Images from Echocardiographic Scans Using a Deep Learning-Based Generative Adversarial Network (GAN) Model
Authors: Ibrohimjon Abdullayev
Volume: 9
Issue: 8
Pages: 41-44
Publication Date: 2025/08/28
Abstract:
Early and accurate diagnosis of cardiovascular diseases remains one of the most critical goals of modern medicine. Although echocardiography is a widely used, safe, and cost-effective real-time imaging modality, its spatial resolution is significantly lower compared to high-precision techniques such as magnetic resonance imaging (MRI). Therefore, the development of a deep learning-based approach for generating high-resolution MRI-like images from echocardiographic inputs using a Generative Adversarial Network (GAN) holds both scientific and clinical significance. The fundamental components of a GAN-namely, the generator and discriminator neural networks-enable the model to learn the differences between modalities and synthesize novel MRI-style images. In this study, we employ a CycleGAN architecture to convert ultrasound images of the heart into MRI-like representations. The paper outlines the training methodology, performance evaluation metrics (SSIM, PSNR, MAE), and the potential diagnostic implications of the approach. The experimental results demonstrate that the generated MRI images closely resemble actual MRI scans and suggest that this model may significantly enhance diagnostic accuracy in clinical practice by bridging the resolution gap between echocardiography and MRI.