Deep Learning Super-Resolution reconstruction for fast cardiac MRI protocol：A Comparative Study with Conventional cardiac MR

Motivation: Deep Learning (DL) Super-Resolution reconstruction methods combined with Compressed SENSE for cardiac MRI(CMR) have not been well studied. There is a need to reduce scan times without compromising image quality. Goal(s): To evaluate a DL Compressed SENSE artificial intelligence (CSAI) in reconstructing common CMR sequences and compare its performance with standard sequences. Approach: 100 patients were prospectively recruited for conventional SENSE CMR sequences and CSAI-accelerated CMR sequences between March and August 2024. Two readers assessed image quality qualitatively and quantitatively. Quantitative measurements of biventricular function, myocardial edema, and fibrosis were obtained. Results: CSAI-CMR reduced acquisition time by 57.4% while significantly enhancing image quality. Impact: This study demonstrates that CSAI-CMR improves image quality and significantly reduces scan time, enhancing patient comfort and clinical efficiency, it supports advancing cardiac MRI toward more precise, efficient, and patient-friendly practices, potentially increasing its clinical adoption.