2022/12 - : Article: “Infarct quantification with cardiovascular magnetic resonance using 'standard deviation from remote' is unreliable: validation in multi-centre multi-vendor data”[Journal Club]-Journal Club Webinar

Video Transcription

Video Summary

The JCMR Journal Club discussion focused on a study testing whether the common “N-standard deviations from remote” method is reliable for quantifying infarct size on late gadolinium enhancement (LGE) cardiovascular MRI. Dr. Heiberg explained that LGE is the gold standard for scar imaging, but the SD-from-remote approach is widely used without strong validation. The study examined 196 multicenter, multivendor datasets from ischemic cardioprotection trials, using manual delineation as the reference standard and evaluating how results changed with remote ROI placement, ROI size, number of SDs used, spurious pixel handling, and image noise.<br /><br />The key finding was that infarct size estimates were highly unstable, with about 50% variability and substantial bias. Measured scar size changed markedly depending on where the “normal” ROI was placed, how large it was, and how many SDs were chosen. Worse image quality and higher noise generally led to smaller estimated infarct sizes. The method also produced spurious “scar” in healthy myocardium and was inconsistent across sites and vendors.<br /><br />The panel concluded that the N-standard deviations method should not be used for quantitative infarct assessment, especially in multicenter or multivendor settings. Instead, validated alternatives such as full-width half-maximum, FACT, and EVA were recommended. The discussion also highlighted the need for explicit reporting of quantification methods, better validation against animal/ex vivo ground truth, and careful use of AI/ML tools that still require robust training and validation.

Keywords

late gadolinium enhancement

cardiovascular MRI

infarct size

N-standard deviations method

scar quantification

manual delineation

multicenter study

multivendor datasets

image noise

full-width half-maximum