Using fractal techniques to reduce haziness in medical imaging

In the last decades, medical exams became a regular act; thus, the amount of mammograms interpreted by a radiologist increases dramatically. A focused effort is under way to develop a Computer-Aided Diagnosis of Mammograms (CADM) able to store, process and interpret data related to a mammogram. An important component is dedicated to reduce the haziness of interpretation; this will be done by tracing Focus of Attention Regions (FARs), which can have tumors, abnormalities, so. The goal is to underline these FAR zones, instead of paying attention to the whole image, accomplishing two major objectives: on one side, reducing the volume of processed data, and on the other side, reducing the possibility of false detection. In this paper two way of detecting FARs will be investigated. Both are provided by fractal geometry, a field which has already proved its utility in medical imaging. Natural shapes can be characterized through fractal characteristics; therefore fractal analysis techniques represent an efficiently working instrument which allows automation of analysis procedures of medical images. Experiments were done on 20 cases of mammograms; the results show that fractal techniques may be used to reduce the haziness in medical imaging.