Reproducibility of manual and semi-automated late enhancement quantification in patients with Fabry disease

Abstract

Background

Late enhancement (LE) imaging is increasingly used for diagnosis of non-ischemic cardiomyopathy. However, the mostly patchy appearance of LE in this context may reduce the reproducibility of LE measurement.

Purpose

To report intra- and inter-observer variabilities of LE measurements in Fabry disease using manual and semi-automated quantification.

Material and Methods

Twenty MRI data-sets of male patients aged 44 ± 7 years were analyzed twice (interval 12 months) by one observer and additionally once by a second observer. Left ventricular (LV) parameters were determined using cine MRI. Gradient-echo LE images were analyzed by manual planimetry and by a semi-automatic prototype software. Variabilities were determined by Bland-Altman analyses and additionally intra-class correlation coefficient (ICC) values were calculated to survey intra- and inter-observer reproducibility.

Results

The amount of LE was 5.2 ± 5.1 mL or 2.8 ± 2.6 % of LV mass (observer 2). LE was detected predominantly intramurally in a patchy pattern. All patients had LE restricted to the basal infero-lateral parts of the LV. The extent of LE correlated to LV mass (207 ± 70 g, P < 0.05, r = 0.6). The intra- and inter-observer variabilities were −0.6 to 1.0 mL and −0.7 to 1.6 mL, respectively (95% confidence intervals). ICC values were 0.981–0.999. The semi-automatic software allowed quantification of LE areas in all patients. The comparison of LE amount determined by semi-automatic software versus manual planimetry yielded an intra-observer variability ranging from −1.9 to 2.3 mL.

Conclusion

Semi-automatic planimetry of patchy LE in patients with Fabry disease is feasible. The determined intra- and inter-observer variabilities for manual and semi-automatic planimetry were in the range of 20–40% of LE amount with high ICC values.