Evaluation of change detection techniques in simulated mammography

摘要

Due to the multiple applications of image processing techniques, their use and need for appropriate treatment are currently an area of great interest. Among these techniques, special importance has the detection of changes whose main characteristic lies in the interest to identify differences between two images of the same place but recorded at different times. This is very useful in medicine, for example in a postoperative evaluation, comparing the images before and after the operation. For a good comparison, both images must coincide in space and illumination conditions. In this work, it was first addressed the problem of mismatching by means of pre-processing techniques, as calibration and automatically indentifying and segmenting the region of interest. Then, change detection methods were applied on mammograms obtained by simulation, for breast microcalcification screening. Thresholding methods, implemented in the difference image, were evaluated and compared. Thus, this work contributes to increase and interpret the information that can be extracted from medical images.