| 摘要 |
An automated method for evaluating image data taken over a sequence of image frames to determine a contour of a left ventricle (LV), using a gray scale value for each pixel in each image frame. Also provided are training data determined from manually-drawn contours of hearts of other individuals. End diastole (ED) and end systole (ES) aortic valve points and an apex point determined by viewing the sequence of images for the heart being analyzed are entered, and an initial ED/ES region classifier is developed for the image data using a probability lookup table. An ED/ES pixel class image and ES pixel class image are produced, yielding 100-point classifier ED and ES boundaries. Regression coefficients derived from the training data are applied to determine a 100-point regressed ED boundary and a 100-point regressed ES boundary, which can be used to calculate an ejection fraction for the heart. |
