Study on identification of expiratory phase X-ray image / 生物医学工程学杂志

ABSTRACT

To divide the lung from the original X-ray image, the maximum classes square error is employed in finding the threshold. Then the thorax is ascertained according to the edge-detection and tracking-bug method, and the lung is divided from the thorax by the thresholding segmentation method. The gray mean in the lung is regarded as the eigenvalue to identify the image, but it doesn not work well. In view of the dosage of the X-ray exposure and the difference between persons, a new method that adopts the ratio of the gray in lungs to the gray in bones as the eigenvalue is brought forward to identify the image. And the two methods are compared by using the distance between classes. The result shows that the second method is more effective.