A high performance spatio-temporal displacement smoothing method for myocardial strain imaging

Abstract

Accurately assessing local myocardial strain is important for diagnosing ischemic heart diseases because decreased myocardial motion often appears in the early stage. The abnormal contraction motion can be visualized by myocardial strain images, but the strain calculation is very sensitive to noise. In our previous research, we proposed an adaptive dynamic grid interpolation (ADGI) method for overcoming the limitation of the trade-off between spatial resolution and accuracy in traditional moving-average filters. Usually, when the scanning frame-rate is high the correlation coefficient, which is calculated from ECAM, will be high. But only using two consecutive frames' phase-shift data, the displacements' dynamic range is low. Therefore, the strain calculation will be affected by the noise. In this research, we extend the proposed method with the ability to process two or more frames' data for improving the SNR of myocardial strain imaging. From the simulation results, we can conclude that our method can provide more accurate myocardial strain images. In a model with the infarcted region located around 1 to 3 o'clock, the RMS error is decreased to 16.8% without degrading spatial resolution.