Optimizing computed tomography : quality assurance, radiation dose and contrast media

Abstract

Computed tomography (CT) is an important modality in radiology; it enables imaging of the inside of patients without superimposed anatomy. The radiation dose and quality of a CT image are highly dependent on the CT scanner, the scan settings and, if applicable, the timing and dosage of the intravenous contrast media (CM). The aim of this Thesis was to develop tools and insights that help maximize the value of examinations for patients undergoing CT and to reduce its cost in terms of radiation and CM dose. The Thesis consists of five studies.

The first paper was on quality control (QC) of CT, which is the foundation for a radiology clinic: it provides trust that the equipment functions as expected. A new method of performing routine QCs was proposed where the concept of key performance indicators (KPI) was introduced, together with a semi-automatic process allowing for daily QCs. During the time of the study, multiple deviations were discovered that would have been difficult to detect using traditional QCs. Performing QCs more frequently facilitates more extensive trend analysis. The second paper was on automatic tube current modulation (ATCM). A phantom and a method for the characterization of ATCM were developed. These allowed for a characterization of CT scanners from the four main CT vendors in Sweden, summarized in four extensive tables showing how the ATCM responds to changes in scan parameters. More specifically, the tables present how changes in scan settings of the localizer radiograph (LR), scan settings of the acquisition, reconstruction parameters and patient miscentering affect the ATCM. The third paper was on radiation dose estimation uncertainties coupled to the patient table. In most commercial radiation dose estimation software packages for CT, the patient table is not included. That effect was previously unknown but could be shown using Monte Carlo (MC) calculations of CT scans performed with and without the patient table. It was shown that by not including the effect from the patient table in radiation dose estimations, the radiation doses are overestimated by 5% to 23%, depending on the scan mode.

The fourth paper evaluated whether the standard LR can be replaced by a low-dose spiral scan, a so-called synthetic LR (SLR). Such an SLR can potentially improve ATCM, CM dosage and CT planning. Radiation doses were estimated using MC, the image quality was compared in a prospective study of ten patients and the impact of miscentering was investigated with a phantom measurement of water equivalent diameters. It was shown that the radiation doses and the image quality of SLR and LRs were similar. Estimated water equivalent diameters were more consistent when calculated from the low-dose spiral scan compared to the LRs. It was concluded that it is feasible to replace the traditional LR with an SLR for CT scan planning. The fifth paper was a continued investigation of the low-dose spiral scan, but with focus on intravenous CM dosage planning. Altogether, 238 patients who had undergone PET/CT and ii for whom body metrics (height and weight) had been acquired were retrospectively analyzed, the CT number enhancement of the liver was measured, and body volumes of muscle and fat were calculated using the attenuation correction CT (low-dose spiral scan). Multiple linear regressions showed that for CM dose planning, the body volumes of muscle and fat are better to use than body weight. However, the adjusted R2 values of all the investigated models were low, indicating that responses to CM dosage are complex and require more research.

In this PhD Thesis, tools and insights were developed to improve the imaging stability of the CT scan by developing semi-automatic QC protocols and techniques to better estimate patient size and shape potentially reducing variation in image quality, radiation dose and CM enhancement among patients.