Enhancement of edge detection: Qualitative and quantitative effects of B-color
The use of color in B-mode echocardiography, or B-color, refers to the display of images in previously defined 'pseudocolor' scales as compared to the standard gray scale display. The purpose of our study was to determine, both in vitro and in vivo, whether B-color transforms the image so that pixel intensity is altered when measured quantitatively on an off-line videodensitometric system. In addition, we examined the qualitative effects of B-color by visually interpreting cardiac ultrasound images and evaluating the qualitative differences in detection of endocardial borders using gray as well as colorized scales. Using a frequency independent tissue phantom and the left ventricle of a volunteer, images were analyzed off-line using videodensitometric analysis. Small differences in videointensity were found, approximately 5%-10% (20 gray levels, NS) change across the border zones in the tissue phantom and the left ventricle. The difference in videointensity across the border zone varied by up to 35% (89 gray levels) when different post-processing settings were utilized. Qualitative analysis showed that indeed a difference exists when interpreting cardiac ultrasound images with gray and color scales. A grading scale was utilized to evaluate the clarity of the endocardial border of a series of 50 echocardiographic images by 10 blinded observers (0 = poor, 1 = fair, 2 = good, and 3 = excellent). The average scores were as follows: gray = 0.77 ± 0.54, temp = 2.1 ± 0.68, magenta = 2.3 ± 0.75, rainbow = 2.4 ± 0.67, and sepia = 2.0 ± 0.75. There was significant difference (p < 0.05) between the gray scales and the colorized scales, demonstrating that the human eye can perceive small differences utilizing shades of color rather than gray scales. We conclude that although color encoding video images enhances the qualitative assessment of echocardiographic images, it is not due primarily to a change in absolute pixel intensity. Furthermore, we conclude that the use of video post-processing with or without color enhancement of data will result in quantitative changes in pixel intensity.
