GE "Advantages of Deep Learning in Clinical CMR Practice"-Webinar

Video Transcription

Video Summary

The webinar highlighted how deep learning is improving cardiac MRI practice in both adult and pediatric patients. Dr. Alexander Hirsch discussed myocardial tissue characterization, focusing on late gadolinium enhancement (LGE) and parametric mapping. He explained how T1, T2, and extracellular volume mapping help detect diffuse disease such as fibrosis, amyloidosis, Fabry disease, myocarditis, and iron overload. He then presented GE’s Air Recon DL, which reduces noise and ringing artifacts, sharpens images, and can shorten scans. However, he cautioned that LGE quantification may change with noise reduction, especially when using standard deviation-based thresholds.<br /><br />Dr. Shreyas Vasanawala described rapid, automated pediatric cardiac MRI approaches. A 4D flow volumetric scan can capture anatomy, flow, and ventricular function in a single acquisition, with automated reformatting, vessel segmentation, and flow correction. He also presented a deep learning–based respiratory-triggered cine method that reduces the need for breath holds, improves image quality, and speeds up exams while preserving quantitative accuracy.<br /><br />In the panel discussion, both speakers emphasized that these tools improve efficiency and image quality, but careful attention is needed when using them for quantitative clinical decisions.

Keywords

deep learning

cardiac MRI

myocardial tissue characterization

late gadolinium enhancement

parametric mapping

T1 T2 and extracellular volume mapping

Air Recon DL

pediatric cardiac MRI

4D flow volumetric scan