Introduction: Computed tomography (CT) is a diagnostic imaging technique in which a beam of X-ray light is passed through a selected anatomical region while the X-ray tube and the detectors rotate around the patient. The X-ray tube and detectors must be rotated at least 180 degrees around the patient to obtain sufficient attenuation data from which the computer can reconstruct an image of acceptable quality using an algorithm. CT image reconstruction is basically divided into two major categories: analytical and iterative reconstruction. There are several different types of analytical image reconstruction, the most widely used of which is filtered backprojection. Rapid technological developments are enabling new techniques for image reconstruction, with the most attention being paid to the use of artificial intelligence. The main aim is to reduce the radiation esposure on patients while improving the quality of CT images. Purpose: The purpose of this diploma work is to present the automatic reconstruction of CT images and the role of artificial intelligence in CT imaging. The presentation covers the use of AI in reconstruction, a description of the advantages and disadvantages, where the limits of AI in diagnostic imaging lie, and a comparison of reconstruction made by AI and by radiological engineer. Methods: In my diploma work I used a descriptive method with a literature review, which I searched in Slovene and mainly in English. I used the syngo.via software at the Faculty of Medicine in Ljubljana to perform the measurements and reconstruct the CT images. Results: In the results, I have presented functions that the syngo.via software can perform. The transverze head reconstruction, automatic rib and vertebral labeling, coronary artery calcification assessment, stenosis, length and diameter measurement of thoracic aorta, full course view of the lower limb arteries, automatic lung segmentation and automatic detection of pulmonary nodules are shown. Discussion and conclusion: My diploma work has introduced me to the theoretical part of algorithms for CT image reconstruction and postprocessing. Based on the CT image processing I have performed with syngo. via, I have found that artificial intelligence makes the radiology engineer's job easier, while reducing the time needed to reconstruct and process the images. The artificial intelligence performed all of the procedures to the expected standards and can be compared as equaly good as procedures performed by a radiological engineer. Artificial intelligence is already routinely used in some areas of radiology, but it will take some time before artificial intelligence will able to independently interpret images and establish a diagnosis.
|