Indication-Based Diagnostic reference levels for computed tomography examinations of paediatric head and neck tumours at Uganda Cancer Institute

Abstract

Introduction Paediatric head and neck cancer is intensively imaged with Computed Tomography which contributes significantly high radiation dose. The paediatric population is most sensitive to the stochastic effects of ionising radiation hence the need for stringent dose optimisation. Indication-based typical DRLs are superior to anatomic DRLs in ensuring that children receive adequate radiation dose that is not only appropriate for age group but also for clinical indication. Objectives This study set out to establish typical indication-based DRLs for head and neck CT examinations of paediatric Head and Neck tumours at Uganda Cancer Institute and to compare them with National, Regional and International DRLs. Methods This was a retrospective cross-sectional study where 253 study participants with dose data for children between January 2020 and December 2022 were consecutively recruited. These were categorised into two strata namely clinical indication (diagnosis, staging, and follow-up) and age-grouping (0-<1, 1-<5, 5-<10, 10-<15, 15-<18). Relationships between CT acquisition parameters and the dose quantities were evaluated. For each indication and each age-group, the 50th and 75th percentile of the CTDIvol and DLP were derived. These DRLs were compared with National, Regional and International DRL values published in literature. Results A total of 253 participants were included in the study with an average age of 9.6 years of which 58.5% were male. The majority of patients underwent two scan phases with nearly all using 120 kVp (99.6%) and 3 mm slice thickness (84.2%). Diagnostic CT scans had the highest mean mAs of 369 and lowest average scan length of 30.3cm. Staging CT had the highest average scan length of 37.8cm. The 1-<5 years age-group had the highest average mAs of 361 and lowest scan length of 26.9cm. There was a strong positive significant relationship for both mAs (rs = 0.9471, p-value = 0.000) and scan length (rs = 0.8566, p-value = 0.000) with the CTDIvol estimates. For DLP estimates, there was a strong positive significant relationship with mAs and a weak negative significant relationship with scan length. There was significant relationship between the DRL estimates and slice thickness and phantom size with no significant relationship with kVp and number of phases. The median CTDIvol and DLP values for pre-contrast and post-contrast phases were comparable for the different indications and age-groups. The 15-<18 group had the highest median CTDIvol and DLP of 52.9 mGy and 2784.6 mGy.cm respectively. The Diagnostic CT scans had higher doses compared to other indications across all age-groups. Significantly, the 1-<5 group showed higher median doses across the different clinical indications compared to other age-groups. Generally, the 75th centile CTDIvol values were comparable to a Ugandan National DRL study but higher than a Nigerian and UK study. The most notable difference was seen in the 1-<5 group with this study registering a value of 42.45 mGy that was comparable with the Ugandan study (43.1 mGy) but significantly higher than the Nigerian (37 mGy) and UK study (24 mGy). Conclusion Selection of tube voltage and number of scan phases was largely consistent unlike slice thickness, scan length, mAs and phantom size which showed a wide variation across patients. The derived DRLs did not have a clear linear distribution across patient age or clinical distribution. The 1-<5 age group notably received high radiation doses when compared with Regional and International DRL values. This study re-emphasised the need for stringent radiation dose optimisation techniques. Recommendation Standardisation and adoption of indication-specific protocols for head and neck CT scans as well as regular training of imaging technologists and 3-yearly dose surveys will improve dose optimisation etiquette. Upscaling this study to involve more patients and different CT scan characteristics will improve the outcome of the radiation dose survey. These DRL values should be adopted as an initial optimisation benchmark.