Disclosures: Several study authors report they have authorship rights for patents related to this study. The editorial authors report no relevant financial disclosures.
September 14, 2021
3 min read
Save

AI-enhanced cardiac MRI shows promise in hypertrophic cardiomyopathy

Disclosures: Several study authors report they have authorship rights for patents related to this study. The editorial authors report no relevant financial disclosures.
You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

Artificial intelligence-assisted enhancement of cardiac MRIs eliminated the need for IV contrast agents and provided better image quality and faster image development compared with late gadolinium-enhanced images, researchers reported.

According to data published in Circulation, virtual native enhancement using AI also achieved high agreement with late gadolinium-enhanced images in detecting and quantifying myocardial lesions in patients with hypertrophic cardiomyopathy.

CT scan
Source: Adobe Stock

According to the study, late gadolinium enhancement cardiac MRI (CMR) is the standard for noninvasive myocardial tissue imaging; however, it requires use of IV contrast agents. Researchers hypothesized that a CMR virtual native enhancement technology, developed with the use of AI, could eliminate the need for IV contrast agents, shorten scan times and reduce costs.

AI enhanced cardiac MRI

“The deep learning model for generating virtual native enhancement uses multiple streams of convolutional neural networks to exploit and enhance the existing signals in native T1 maps (pixel-wise maps of tissue T1 relaxation times) and cine imaging of cardiac structure and function, presenting them as late gadolinium enhancement-equivalent images,” the researchers wrote. “The virtual native enhancement generator was trained using generative adversarial networks. This technology was first developed on CMR data sets from the multicenter Hypertrophic Cardiomyopathy Registry, using hypertrophic cardiomyopathy as an exemplar.

“The AI deep learning is effectively acting as a ‘virtual contrast agent’ that enhances the native CMR,” the researchers wrote. “In other words, it produces a ‘virtual late gadolinium enhancement’ image without the need for gadolinium.”

To evaluate effectiveness of virtual native enhancement compared with late gadolinium enhancement for CMR, researchers enrolled 1,348 patients with hypertrophic cardiomyopathy (HCM), who provided 4,093 triplets of matched T1 maps, cines and late gadolinium enhancement data sets.

According to the study, virtual native enhancement and late gadolinium enhancement were scored and contoured by human operators who assessed image quality, visuospatial agreement and myocardial lesion burden quantification.

Researchers reported that virtual native enhancement was associated with better image quality compared with late gadolinium enhancement, as assessed by four independent operators (P < .001 by Wilcoxon test).

After the exclusion of images with uninterpretable late gadolinium enhancement, researchers conducted lesion quantification in 326 short-axis pairs of virtual native-enhanced and late gadolinium-enhanced images from 121 patients. According to the study, virtual native-enhanced images were in high visuospatial agreement with late gadolinium-enhanced images for detecting and quantifying both hyperintensity myocardial lesions (correlation coefficient, 0.77-0.79; intraclass correlation coefficient, 0.77-0.87; P < .001) and intermediate-intensity lesions (correlation coefficient, 0.7-0.76; intraclass correlation coefficient, 0.82-0.85; P < .001).

In addition, the cine plus T1 map images needed for virtual native enhancement were acquired within 15 minutes and the virtually enhanced image was produced in less than 1 second.

“Currently, the majority of CMR scans for tissue characterization requires intravenous access, the use of gadolinium-based contrast agents, related consumables and patient preparation by trained staff,” the researchers wrote. “Virtual native enhancement is available immediately after native T1 map-ping acquisition with no additional cost. Replacing late gadolinium enhancement with virtual native enhancement can significantly shorten the scan time to within 15 minutes, allowing twice as many patients to benefit from CMR at the same infrastructure capacity. The clinical impact and potential cost savings of popularizing this new CMR technology could be substantial.”

Potential to reduce procedural time

In a related editorial, Charlotte H. Manisty, PhD, clinical senior lecturer in CMR and HF in the Institute of Cardiovascular Science at University College London, and colleagues discussed the potential for this technology.

“CMR using late gadolinium enhancement imaging has been transformative for noninvasive myocardial tissue characterization, thereby helping diagnose and risk-stratify cardiomyopathies including HCM,” the editorial authors wrote. “Zhang et al have provided valuable evidence that applying deep learning automated analysis techniques to standard noncontrast CMR sequences generates synthetic late gadolinium-enhanced images that correlate strongly with conventional late gadolinium-enhanced images, demonstrating their clinical utility in HCM. The virtual native enhancement technique has the potential to improve feasibility and reduce overall CMR procedural time, which should further increase demand of this currently underutilized modality. The first major test for virtual native enhancement will be its performance as a prognostic marker within the Hypertrophic Cardiomyopathy Registry study itself. We await the outcome with much anticipation.”

Reference: