Spatiotemporal Fourier filtration of acoustic quantification endocardial borders using Cartesian vs. Polar coordinate system

The analysis of endocardial borders by Acoustic Quantification (AQ) of 2D echocardiographic images is limited by the presence of random high frequency noise. Using 1D filtration in the Fourier domain we have shown previously that smoothing of AQ images improves signal-to-noise ratio. In this paper we have extended the study by applying 2D Fourier filtration (spatial and temporal) to the AQ borders, and study which of the two sampling methods, Polar or Cartesian sampling, yields a more accurate representation. Various image processing techniques including region labeling, thresholding, edge detection, and area fill algorithms along with 2D-DFT and Hanning filters were used to process the images. Using the original AQ as the gold standard, our results favor Cartesian sampling over Polar sampling