Novel AI-Enhanced MRI Technique Improves Cardiac Imaging for Arrhythmia Patients

A recent study has demonstrated that AI-enhanced MRI significantly enhances image quality and provides reliable ventricular measurements in patients with arrhythmia, according to findings published in Radiology: Cardiothoracic Imaging. This advancement is crucial for evaluating left ventricular function, a key factor in managing heart failure.

Traditional cardiac MRI is the standard for assessing left ventricular function, but it often requires multiple episodes of breath-holding, posing challenges for patients with arrhythmias. A new AI-based approach, using a deep-learning-enhanced Compressed SENSE (AI-CS) single-shot cine sequence, offers a compelling solution by reducing the need for breath-holds and minimizing motion artifacts.

Nan Zhang, MS, and colleagues from Zhongshan Hospital of Fudan University conducted the study involving 25 healthy volunteers and 45 patients with suspected arrhythmias. Participants underwent cine imaging using both conventional MRI and AI-CS sequences. The AI-enhanced method excelled in image quality, especially in individuals with severe arrhythmias, and effectively reduced mistriggering and image distortions.

In terms of ventricular assessment, AI-CS matched the standard cine MRI in measuring end-diastolic and end-systolic volumes, stroke volume, ejection fraction, and peak strain across multiple orientations. When conventional cine methods faced limitations, AI-CS provided ejection fraction values comparable to those from echocardiography.

The research assembled data from three independent cardiovascular radiologists, who analyzed the images while blind to clinical and previous imaging information. Their evaluation confirmed the improved visibility of cardiac structures with the AI technique, highlighting the endocardial and epicardial borders as well as papillary muscles.

The study findings reveal a 100% success rate in obtaining usable images with the AI-CS approach, compared to 88% with traditional cine sequences. “The AI-CS sequence effectively avoided cardiac motion artifacts commonly caused by mistriggering in conventional cine,” Zhang noted, emphasizing its potential for clinical application.

The AI-CS framework marks a promising advancement for cardiac MRI, especially in clinical settings where reducing acquisition time is essential. Further optimization may enhance its utility by improving image contrast and reducing artifacts.