Ferencik M. Abstract #216.
The
5th Annual Scientific Meeting of the Society of Cardiovascular Computed
Tomography
Semi-automated software improved interobserver agreement
for coronary CTA of both stenosis and plaque volume, according to study
results.
Researchers analyzed CTA data from participants (n=31)
of the ROMICAT trial with at least one coronary stenosis >50%. They analyzed
stenoses (n=39) and associated coronary plaques via semi-automated software
(Vitrea Enterprise Suite; Vital Images) by two independent readers. The
evaluated degrees of stenosis and plaque volumes were defined as low (<90
Hounsfield units), intermediate (91-150 Hounsfield units) and high (>150
Hounsfield units).
Researchers found that manually measured stenosis
correlated between readers (mean difference, 4 ± 9%; R²=0.71).
Semi-automated stenosis measurements led to an improved correlation (mean
difference, 1 ± 8%; R²=0.85), although there was only a moderate
correlation between manual and semi-automated measurements (mean difference, 8
± 15%; R²=0.52). Plaque volume analysis showed notable correlation
between readers (mean difference, 19 ± 44 mm³; R²=.83), as
well as improvement with semi-automated software (mean difference, –11
± 14 mm³; R²=.86).
Maros Ferencik, PhD, MD, of Massachusetts General
Hospital, Boston, and researcher on the study, said the study limitations that
he noted in his presentation were the small sample size, the inclusion of
plaques only associated with >50% stenosis, the excellent
image quality required for the plaque evaluation and the lack of a comparison
to a gold standard (eg, IVUS).
“Semi-automated software improved interobserver
agreement for
coronary CTA evaluation of stenosis and plaque volume.
However, there was difference between manual and semi-automated
measurements,” Ferencik said. “Further improvements of semi-automated
software will be necessary prior to the implementation in clinical
trials.”
This idea has been around for several years but needed a boost in
technology, like this one, to make it fast, automated and reproducible to make
this feasible for clinical trials. Once validated and widely available, there
should be a multicenter evaluation of outcomes and treatments in the groups
with different plaque densities.
– Kim Allan Williams, MD
Cardiology
Today Editorial Board