Fractal Feature Analysis and Classification in Medical Imaging

Following Mandelbrot's fractal theory, it was found that the fractal dimension could be obtained in medical images by the concept of fractional Brownian motion. An estimation concept for determination of the fractal dimension based upon the concept of fractional Brownian motion was discussed. Two applications were found: 1) classification; 2) edge enhancement and detection. For the purpose of classification, a normalized fractional Brownian motion feature vector was defined from this estimation concept. It represented the normalized average absolute intensity difference of pixel pairs on a surface at different scales. The feature vector used relatively few data items to represent the statistical characteristics of the medical image surface and was invariant to linear intensity transformation. Finally, by calculating normalized fractional Brownian motion feature vectors in five different ultrasonic image surfaces, it was found that the classification of normal and abnormal ultrasonic liver images could be obtained from the differences between their feature vectors. For edge enhancement and detection application, a transformed image was obtained by calculating the fractal dimension of each pixel over the whole medical image. The fractal dimension value of each pixel was obtained by calculating the fractal dimension of a 7 x 7 pixel block centered on this pixel. Preliminary results using projection radiographs suggest that the fractal based image transformation appears to hold promise as an edge enhancement and preprocessing algorithm that does not increase noise in the way that gradient operators do. © 1989 IEEE

296 citations (Crossref) [2023-10-31]