Regresar

Medical imaging preprocessing techniques: Foundations and scope

Abstract:

In 3-D cardiac imaging, Multi-Slice Computed Tomography (MSCT) stands out for its clinical utility to generate information relative to the morphology of the heart. Typically, MSCT produces cardiac images with sub-millimetric spatial resolutions which exhibit certain problems related to noise, artifacts and low contrast that affect both the adequate identification of the various anatomical structures of interest to cardiologists and the evaluation of the cardiac function. Because the analysis of cardiac function may be affected by the aforementioned problems, several preprocessing techniques have been proposed to address them. In this sense, to reduce the effect of noise and artifacts, filtering techniques are used; While to overcome the problem of low contrast are considered techniques that define Regions Of Interest (ROI). Most of these techniques are aimed at reducing the negative impact on the quality of such images, noise and artifacts produced by movements (voluntary and involuntary) of the patient and those generated by objects located in his body. However, in spite of the efforts of software developers and the manufacturers of scanners, in order to correct such an impact, certain factors still exist that enable their generation and, consequently, their notorious appearance in the images of MSCT affecting the present information in them. Therefore, in the present work, we analyze both the problems that affect the MSCT images of the heart and the theoretical foundations of the preprocessing (filtering + definition of ROI) techniques most used in the medical area to deal with the mentioned problems.