Feature tracking versus manual methods of assessment of left atrial mechanics in acute myocardial infarction: a pilot study

Background: LA function is conventionally assessed by bi-plane method to compute LA end diastolic volume (LAEDV), end systolic volume (LAESV), stroke volume (LA SV), ejection fraction (LA EF). Voxel feature tracking (FT) is a novel technique for the assessment of LA function. We aimed to investigate if voxel FT derived longitudinal or radial strains are superior to traditionally derived parameters of LA function (LA SV and LA EF) in predicting LV function. Methods: Twelve-patients underwent CMR at 3T (Achieva CV, Philips Healthcare, Best, The Netherlands) within 3 days following AMI. CMR protocol included: cines and late gadolinium enhancement (LGE) imaging (0.1 mmol/kg gadolinium DTPA). Indexed LAEDV (LAEDVi), end-systolic volume (LAESVi) and ejection fraction were computed by bi-plane method. Voxel FT for the LA in the long axis 4-chamber cines was analysed offline using commercially available software (cvi42 v5.1, Circle Cardiovascular Imaging Inc., Calgary, Canada). Results: Demographics and basic CMR parameters are mentioned in Table 1. Analysis time was longer for manual contouring and computation of LA EF and SV compared to 4-CH strain analysis (4 plus or minus 2 min versus 2 plus or minus 1.5 min, p = 0.01). On univariate analysis, LV EF was correlated to peak longitudinal strain (PLS) (p = 0.01) and peak radial strain (PRS) (p = 0.02). On multivariate regression analysis, PLS of LA was most strongly correlated to LV EF (R=0.68; p = 0.015). All other parameters did not achieve statistical significance. Conclusion: Peak left atrial longitudinal strain (PLS) is the only parameter of LA function, which independently correlates to LV ejection fraction. PLS of the LA is easily assessed on 4-chamber cines alone and takes less time to compute than the manually generated parameters of LA function.