An automated method to detect single-leg separation in radiographic images of Bjork-Shiley 60 and 70/spl deg/ convexoconcave heart valves

An accurate technique for the extraction of region of interest (ROI) is an integral process in the objective quantification of single-leg separation (SLS) In the 60/spl deg/ and 70/spl deg/ Bjork-Shiley Convexo-Concave heart valves. The authors have developed a fully automated technique to detect the region of interest (ROI) where SLS would most likely occur from sequential valve images. The technique employs a modified version of Kundu's (1990) method for segmentation and principal aids transformation for the alignment of the images. These methods were chosen for ease of implementation, speed, and accuracy. The first step of the technique is to extract the inner contour of the outlet strut. The second step involves a distance analysis where a set of points characterizing the ROI is extracted. This technique has been successfully applied to 480 radiographic images from 32 subjects.