Numeric Expression of Clinical Image Characteristics by Converting X-ray Images into Binary Images.

在Macintosh个人计算机系统上建立了临床X线图像分析系统，建立了一种基于图像二值化和膨胀图像处理的病变区域分维D_s的测量方法。以不同条件下的高斯噪声模型图像为例，考察了该方法的精度。对38例涎腺肿瘤患者的CT片和28例正常下颌骨小梁的口腔X线片进行测量，并与28例纤维增生症患者进行比较。CT图像的分布在1.90~2.4时，正常牙体图像的分布在2.496~0.190之间。纤维异常表现为较大的斑点。一般来说，D越大，图像的像素值分布越均匀。

An image analysis system of clinical X-ray images was constructed by a Macintosh personal computer system, and a method to measure fractal dimensions, Ds of the pathological regions was established, which was based on the conversion of images into binary images and on dilation image treatment. The accuracy of the method was investigated by a model image of Gauss noise of various conditions. The Ds were measured on clinical images such as CT images of patients with salivary gland tumors, 38 cases and intraoral dental X-ray images for normal madibular trabeculae, 28 cases comparing with those of patients with fibrous displasia. It was found that the Ds were distributed in the range of 1.9 to 2.4 for the CT images and 2.496 + 0.190 for normal dental images. The fibrous displasia showed larger Ds. Generally speaking the larger D, the image was more uniform in its pixel value distribution.