Computed Tomography (Ct) Scan Assisted Machine Learning in the Management of Artifacts Related to Paranasal Sinuses and Anterior Cranial Fossa

Abstract

Computed tomography (CT), through the use of ionizing radiation, allows us to assess the different parts of the body. It is made up of an X-ray tube that rotates rapidly around the patient generating the radiation beam. This is attenuated with the patient producing information, which is collected by the detectors that are opposite to the tube located in the gantry (part of the tomography equipment); finally, these collected data are sent to the computer that will reconstruct the information obtained and will represent it as an image on the monitor. In the practice of a study, artifices or artifacts may appear regardless of their origin, which limits the scan examination; this leads to stopping the examination and starting again, and added to this with the contrast media, they have to apply these drugs again. State-of-the-art scanners allow complete reconstructions to be performed with few projections, limiting radiation doses, by means of statistical algebraic reconstruction methods. The present work shows the simulation of artifacts in sinusitis diagnosis computed tomography images, the extraction of features from each image, and an automatic classification algorithm for the differentiation of artifacts. The results show that the algorithm is able to classify the simulated artifacts with a percentage of 90%.