| Abstract [eng] |
Exposure of children to ionizing radiation during CT examinations is problematic due to the risk of developing cancer. The main purpose of the dissertation is to develop a CT dose optimization algorithm based on the generation of ultra-low dose CT images from retrospective images using modification with simulated noises corresponding to the reduced X-ray tube current and time product (mAs) values. Tasks: 1) to analyze and compare the levels of exposure of children during CT examinations with the recommended levels of medical exposure in Lithuania and Europe, 2) to create noise bases corresponding to different values of X-ray tube current strength and time product (mAs), 3) to evaluate the compliance of the generated noise bases with the actually recorded noise and to compare the quality and diagnostic suitability of the original and simulated CT images, 4) to prepare practical recommendations for the evaluation of specific bone pathology of the optimized scanning protocol. Comparing the DLP values of head CT scans with the approved DRLs of Lithuania and Europe and calculating the effective dose, it was found that the greatest need is to optimize the CT scans of the head of small children. An experiment was performed using model systems - phantoms, creating different noise bases corresponding to mAs, generating images of different quality. The images generated using the noise simulation algorithm correspond to real small mAs scanned images, so the developed algorithm is suitable for obtaining reduced mAs images without additional patient scans. Images using the head CT protocol with parameters of 120 kV and 13 mAs can be used to diagnose nonsyndromic craniosynostosis with statistically the same accuracy as using the standard protocol of 195 mAs. |
