New research shows that a new AI-enhanced MRI technique can significantly improve the image quality and reliability of cardiac imaging in patients with cardiac arrhythmias.
Accurate assessment of left ventricular function is essential for treating and monitoring patients with heart failure. This is typically done using cardiac MRI, which captures moving images of the heart using so-called cine sequences.
However, this technique requires patients to hold their breath several times during the scan. This is particularly difficult for patients with cardiac arrhythmias, which can lead to reduced image quality or even failed examinations. “For patients with severe arrhythmias, holding their breath is often a challenge, which can result in reduced image quality, failed scans or inaccurate measurements,” says lead researcher Nan Zhang of the Zhongshan Hospital of Fudan University.
AI speeds up and improves imaging
The researchers, whose findings have been published in Radiology: Cardiothoracic Imaging, tested a new method in which an AI-assisted technique, deep learning enhanced Compressed SENSE (AI-CS), is applied to so-called single-shot cine MRI. This technique captures the entire cardiac cycle in just two heartbeats, significantly shortening the required breath-hold.
The study examined 25 healthy volunteers and 45 patients with suspected arrhythmia. All participants underwent both a traditional MRI and the new AI-assisted scan. The results show that the AI-CS technique delivers significantly better image quality, with fewer motion artefacts and fewer errors in the timing of the scan. This advantage proved particularly significant in patients with arrhythmia.
High success rate
Furthermore, it was found that measurements of cardiac function, such as ejection fraction and ventricular volumes, corresponded well with those from conventional MRI. In cases where traditional MRI failed, the AI technique was still able to deliver reliable results.
It is striking that the success rate of the AI-assisted method in the study was 100 per cent, compared to 88 per cent for the conventional technique. According to Zhang, the technology offers a promising alternative to existing MRI methods: “The AI-CS sequence effectively avoided the cardiac motion artefacts commonly caused by mistriggering in conventional cine.”
Reducing scan time
The researchers emphasise that the technique not only improves image quality but also reduces scan time. This can contribute to more efficient workflows and better diagnostics, particularly for patients in whom traditional MRI is difficult to perform.
Further research is needed to optimise the technology further, particularly in terms of image contrast and artefact reduction. If these steps are successful, AI-assisted MRI could play a significant role in daily clinical practice for cardiac imaging.
AI-driven cardiac MRI
Last year, Philips and the Mayo Clinic launched a collaboration to improve cardiac MRI technology using AI. In doing so, the two parties are focusing on shortening complex MRI scans and optimising workflows for radiologists. Shorter scans can increase the accessibility of MRI, reduce waiting times and alleviate the workload in hospitals.
Cardiac MRI offers significant advantages in the diagnosis of heart disease, but is often less accessible due to higher costs and limited availability of scanners. By combining AI technologies from both parties, the researchers hope to significantly reduce scan times and make the examination less burdensome for patients. In addition, AI can support less experienced radiologists in complex examinations. Ultimately, this should lead to more efficient care and better diagnostic outcomes for patients with heart disease.
