Contrast-Enhanced Cardiac Magnetic Resonance Image Based on Enhanced Deep Residual Networks Modified

Contrast is very important in analyzing medical images, and some contrast enhances the methods that could be very helpful when the image contrast is not good. The enhancement of Deep Super-Resolution (EDSR) is the deep learning (DL) technique that improves the quality and resolution of low-resolution images. Within this study, a transfer learning method grounded in a convolutional neural network (EDSR-Modified) is leveraged for enhancing contrast in cardiac magnetic resonance imaging (CMR). The study conducts experiments at the S7 ISP and ACDC datasets, utilizing the S7 ISP dataset for training and the ACDC dataset for testing the assessment enhancement. This study aims to combine EDSR-Modified with transfer learning to improve CMR image quality, which is used to diagnose cardiovascular diseases. Qualitative comparisons are performed using mean squared error (MSE), peak signal-to-noise ratio (PSNR), and SSIM evaluation metrics. It is observed that the proposed technique outperforms the other techniques.