Hypertrophic cardiomyopathy: quantification of late gadolinium enhancement with contrast-enhanced cardiovascular MR imaging.

PURPOSE: To determine the most reproducible semiautomated gray-scale thresholding technique for quantifying late gadolinium enhancement (LGE) in a large cohort of patients with hypertrophic cardiomyopathy (HCM).
MATERIALS AND METHODS: All study patients signed a statement approved by the internal review boards of the participating institutions, agreeing to the use of their medical information for research purposes. LGE cardiovascular magnetic resonance (MR) imaging was performed in 201 patients (71% male) with a mean age of 41.5 years ± 17.6 (standard deviation [SD]) by using standard techniques with administration of 0.2 mmol of gadopentetate dimeglumine per kilogram of body weight. The presence and quantity of LGE were determined first with visual assessment; then with gray-scale thresholds of 2 SDs, 4 SDs, and 6 SDs above the mean signal intensity for the normal remote myocardium; and then with 2 SDs above noise. The LGE quantifications were repeated 4 or more weeks apart to assess reproducibility. Bland-Altman analysis and correlation coefficients were used to compare the visual and various thresholding methods, with normally distributed variables expressed as means ± SDs.
RESULTS: LGE was identified in 103 (51%) subjects. The mean quantity of LGE at visual analysis was 13 g ± 20 compared with 12 g ± 17 at 6 SDs, 25 g ± 23 at 4 SDs, 55 g ± 31 at 2 SDs, and 64 g ± 69 at 2 SDs above noise. All gray-scale thresholds were significantly correlated with visual assessment. The 6-SD threshold had the strongest correlation (r = 0.913, P < .0001) compared with thresholds of 2 SDs (r = 0.81) and 4 SDs (r = 0.91) above the mean and 2 SDs above noise (r = 0.53) (P < .001 for all comparisons). In addition, compared with visual assessment, the 6-SD threshold yielded less intraobserver variability (difference, 0.6 g ± 8, κ = 0.66 [P < .0001] vs 1.4 g ± 9, κ = 0.49 [P < .0001]) and less interobserver variability (difference, 5.4 g ± 18, κ = 0.20 [P < .0001] vs -18.4 g ± 18, κ = 0.08 [P < .0001]).
CONCLUSION: Semiautomated LGE cardiovascular MR gray-scale thresholding with 6 or more SDs above the mean signal intensity for the visually normal remote myocardium yields the closest approximation of the extent of LGE identified with visual assessment and is highly reproducible. This objective method should be considered for quantifying LGE in patients with HCM. © RSNA, 2010