论文部分内容阅读
按照Ｍａｎｄｅｌｂｒｏｔ的分形理论，医学图象及许多自然图象的灰度表面的形成均符合分形布朗运动规律，而且可以用分形的维数来表征图象灰度表面的精细与粗糙程度。文中正是基于这种思想，采用图象的分形维数作为一个特征参量，对人体的肌肉组织进行超声定征。对６０ 多个样本三类病变图象提取分形维数，并采用基于Ｂａｙｅｓ法则的分类器分类，实验表明：用分形维数对组织进行定征，正确率达８８．３３％ 。这为医学的临床辅助诊断提供了一种新的参考量，对提高病变诊断的正确率有重大的意义
According to Mandelbrot’s fractal theory, the formation of the grayscale surface of medical images and many natural images conforms to the law of fractal Brownian motion, and the fractal dimensions can be used to characterize the fineness and roughness of the image’s grayscale surface. In this paper, based on this idea, the fractal dimension of the image is used as a characteristic parameter to perform ultrasonic identification on the muscle tissue of the human body. Fractal dimension was extracted from the images of more than 60 samples of the three types of lesions, and classifiers based on Bayes rule were used to classify them. Experiments showed that the accuracy of the organization was 88.33%. This provides a new reference for the clinical auxiliary diagnosis of medicine and is of great significance in improving the accuracy of the diagnosis of lesions.