Masateru Kawakubo RT, Presenter: Nothing to Disclose

Michinobu Nagao MD, Abstract Co-Author: Research Grant, Bayer AG Research Grant, Koninklijke Philips NV

Seiji Kumazawa PhD, Abstract Co-Author: Nothing to Disclose

Masato Yonezawa, Abstract Co-Author: Nothing to Disclose

Yuzo Yamasaki MD, Abstract Co-Author: Nothing to Disclose

Hiroshi Honda MD, Abstract Co-Author: Nothing to Disclose

Akiko Suyama Chishaki MD, Abstract Co-Author: Nothing to Disclose

Yasuhiko Nakamura RT, Abstract Co-Author: Nothing to Disclose

Junji Morishita PhD, Abstract Co-Author: Nothing to Disclose

Assessment of ventricular function with cardiac magnetic resonance (MR) imaging requires ventricular volumetry for a cardiac cycle. To reduce the time and effort, we developed a semi-automatic method that can detect the biventricular margin for four-chamber (4CH) cine MR imaging, and we performed longitudinal strain (εL) analysis to predict left ventricular (LV) dysfunction.

In 20 patients with heart failure [mean age, 55 years; mean LV ejection fraction (LVEF), 39%], 4CH cine images were obtained using a 3-Tesla MR system. The εL was defined as the percentage of the longitudinal length at end-diastole with respect to the difference between the longitudinal length at each phase and end-diastole on 4CH cine images. The LV and right ventricular (RV) εL values were calculated semi-automatically (εL_auto) and manually (εL_manu) for a cardiac cycle. The correlation between εL_auto and εL_manu, between minimum εL_auto and LVEF were analyzed using Pearson correlation coefficients. The εL_auto values were compared by the Wilcoxon rank-sum test between patients with LVEF > 40% (n = 10) and LVEF ≤ 40% (n = 10). Receiver operating characteristic (ROC) analysis was performed to determine the optimal cutoff of the minimum εL_auto for detection of patients with LVEF ≤ 40%.          

Excellent correlations were observed between εL_auto and εL_manu (Pearson r = 0.85, 0.92 for LV, RV; p < 0.0001). Significant negative correlations between the minimum εL_auto and LVEF were observed (Pearson r = -0.75, -0.89 for LV, RV; p < 0.0001). Significant differences were found in the εL_auto between two patient groups (LV: -15.8 ± 3.0 vs. -7.2 ± 3.2%, RV: -31.9 ± 5.5 vs. -15.0 ± 3.8%; p < 0.0001). ROC analysis revealed the optimal cutoff for identifying patients with LVEF ≤ 40% (LV εL_auto = -12.0%: area under the curve, 0.96; sensitivity, 100%; specificity, 90%; RV εL_auto = -24.0%: area under the curve, 1.00; sensitivity and specificity, 100%).

The εL from our semi-automated method showed excellent agreement with that from the manual tracing and significantly correlated with LVEF. Our method predicted LV dysfunction with simple and easy measurements.

Our proposed semi-automatic method is easy and accurate for longitudinal strain analysis, and it enables the prediction of LV dysfunction by using only one slice of 4CH cine MR imaging.

Kawakubo, M, Nagao, M, Kumazawa, S, Yonezawa, M, Yamasaki, Y, Honda, H, Chishaki, A, Nakamura, Y, Morishita, J, Semi-automatic Cardiac Longitudinal Strain Analysis Using Four-chamber Cine MR Imaging: Correlation with Left Ventricular Dysfunction.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14012823.html