ANALYSIS AND INTERPOLATION OF ANGIOGRAPHIC IMAGES BY USE OF FRACTALS.

Conventional image magnification of digital coronary angiograms (DCA) is most often accomplished by a bandlimited interpolation filter, resulting in blurred images. In order to improve upon this situation, images are interpolated by applying the concept of fractals, a class of mathematical functions which have found use in classification of natural shapes and textures. It is shown that digital coronary angiograms can be described by a fractal model by computing the fractal dimension D and its corresponding correlation coefficient, r, for a set of 9 DCAs from a variety of patients and views. The D values range from 2. 267 to 2. 527 with correlation coefficients ranging from 0. 98 to 0. 99. An algorithm is developed that preserves the fractal dimension as an image is magnified. The results indicate that the fractal dimension of an angiogram may be a preferred method upon which to base image magnification. 5 refs