| Abstract: | Speckle is prominent on all cross sectional echocardiograms. In order to assess its effects on image quantification, frames from a sector scanner with a six bit grey scale were stored and processed off line to identify and smooth the speckle by means of an adaptive filter based on fully developed speckle. In 14 controls, 12 patients with hypertrophic cardiomyopathy, and 12 with secondary left ventricular hypertrophy, filtering significantly reduced the standard deviation of echo intensity, which was used as a measure of the scatter of pixel amplitude, in all three groups (by 52%, 46%, and 46% respectively). The mean value of back-scattered echo intensity itself, however, was reduced by only 7%, 5%, and 8% respectively, and median values were not affected at all. Mean (SD) left ventricular cavity areas on the apical four chamber view were significantly increased from 26 (15) to 30 (17) cm2. The valve dimensions in the parasternal minor axis in 10 patients with mitral stenosis were significantly increased by 11% laterally, but were unaffected anteroposteriorly. Subjective image quality was appreciably modified: endocardial boundaries in apical views were enhanced and the septal "ground glass" appearance was lost in hypertrophic cardiomyopathy. Speckle reduction therefore greatly reduced the scatter of pixel values, with little effect on the mean regional back scattered echo amplitude. It also modified the perceived image texture. Improved boundary definition consistently increased the area estimates, particularly when these depended on lateral rather than range resolution. |
