This study established LDRLs for trunk CT exams at two hospitals in the Gaza Strip, with key factors influencing radiation dose. The findings also offer a snapshot of current CT practice and a basis for dose optimisation and future DRL development. Trunk CT exams made up a significant part of CT activity in the surveyed hospitals, mainly for oncology follow-up and lymphoma staging. SMC conducted more exams than AMH, reflecting Gaza City’s concentrated oncology services and cancer trends across the Gaza Strip [23]. The mean BMI shows a mostly overweight to obese group; this high BMI distribution affects dose optimisation, as patient size influences exposure and radiation dose [24]. Although the LDRLs in this study were from a patient group with a mean BMI of around 30, the CTDIvol and DLP values were similar to or lower than those in several international studies. The estimated ED was 15.6 mSv at SMC and 12 mSv at AMH, reflecting the higher CTDIvol and DLP values observed at SMC, which were mainly due to variations in scan parameters such as beam collimation and tube current. This is a general indication of radiation risk and should be interpreted with caution, as it represents a population-based estimate rather than a patient-specific dose measurement [25].

This discrepancy may stem from fixed or semi-fixed scanning settings, including tube voltage and reference tube current, which may limit dose escalation in larger patients when automatic exposure control is conservatively configured or not fully optimised for patient size. Scanner type and manufacturer-specific dose modulation can also influence dose efficiency. Additionally, tighter control of scan length and coverage can substantially reduce DLP, even in patients with higher BMI, thereby partially offsetting the dose increases associated with larger body habits. Conversely, reliance on standardised protocols not fully tailored to patient physique may result in similar dose indices across diverse patients. These findings illustrate the complex relationship between patient factors, protocol design, and scanner technology in CT radiation dosing.

The parameters that remained fixed within each hospital were not evaluated as sources of within-hospital variation in CTDIvol or DLP. The rotation time, number of rows, nominal slice thickness, beam collimation, reconstructed slice thickness and no tilting were recorded in all exams. Finally, the mean ED values are SMC 15.6 mSv and AMH 12mSv. The calculated LDRLs for trunk CT exams, CTDIvol (13 mGy); DLP (1010.4 mGy.cm), reflect current practice across the two hospitals and fall within the range reported internationally. These values were compared with DRLs from regional and international DRL reports, including the UK, Canada, the USA, Libya, Saudi Arabia, Singapore, Australia, and Japan [26, 27]. The CTDIvol LDRL is comparable to the UK value (13 mGy) and lower than that reported in the USA (15 mGy), indicating broadly similar dose optimisation practices for tube current modulation and image quality targets. In contrast, the DLP LDRL (1010.4 mGy·cm) is slightly higher than values reported in the UK (1003 mGy·cm) and the USA (947 mGy·cm), but remains lower than those reported in several other countries. Notably, both CTDIvol and DLP in this study are higher than those reported in Singapore (13 mGy and 1010.4 mGy·cm vs. 12 mGy and 823 mGy·cm). These intercountry differences likely reflect variations in patient body habitus, protocol philosophy, scanner generation, and scan length selection. In particular, longer scan coverage and less stringent control of anatomical scan extent may contribute to higher DLP values despite comparable CTDIvol, as DLP is directly influenced by scan length as well as tube output. The agreement at SMC shows console-reported CTDIvol and DLP are reliable for dose monitoring. Greater variability in AMH suggests that relying solely on these metrics may misestimate patient dose, thereby affecting protocol optimisation and benchmarking. Regular cross-validation is important, especially with older scanners or less standardised protocols.

On the other hand, significant differences in CTDIvol and DLP appeared between SMC and AMH, with SMC showing higher dose indices despite AMH using higher tube current. Variations likely stem from differences in scanner type, operator protocols and scanning parameters. This underscores the multifactorial nature of CT radiation dose, influenced by scanner characteristics, beam collimation, pitch, scan length and protocols. These findings align with prior studies highlighting hospital variability for the same exams, yet emphasising the need for local DRLs over international standards [28, 29].

Our correlation and multivariable analyses confirm that tube current and voltage mainly affect CTDIvol, especially at SMC, where protocols vary in kVp. The strong positive associations observed are consistent with prior CT research, demonstrating that automated kVp selection reduces CTDIvol compared with fixed kVp protocols while maintaining diagnostic image quality in oncology CT examinations [30, 31]. In addition, tube current is directly linearly related to radiation dose [32]. At AMH, with fixed kVp and pitch, tube current remains the main predictor, indicating limited protocol customisation.

Our findings demonstrate that scan length was a key predictor of DLP in both hospitals, highlighting its role in total dose. While CTDIvol had a stronger link, excessive or unclear scan ranges can raise patient exposure without additional benefit. Longer scan lengths in surveyed hospitals likely explain the higher DLP compared to international DRLs, such as Singapore [33]. Furthermore, pitch showed an inverse relationship with dose at SMC; however, this effect disappeared in multivariable modelling. This aligns with modern multislice CT systems, where automatic tube current adjustment compensates for pitch changes, resulting in minimal dose reduction [34]. BMI was significantly associated with dose at SMC but not at AMH, suggesting patient size was not well integrated into protocols at AMH. This underscores the need for BMI adapted protocols to ensure consistent image quality and prevent unnecessary dose increases.

In the present study, beam collimation showed significant differences in CTDIvol and DLP between hospitals, but its impact on dose variation was minor. While beam collimation was fixed within each hospital, it differed between hospitals. The higher dose indices at SMC are more likely due to differences in tube voltage, scan length, and scanner design rather than collimation alone. This aligns with prior studies indicating that in modern multi-detector CTs, radiation dose depends more on kVp, tube current, pitch, and scan length than on collimation [34, 35]. TCM did not show a significant dose reduction in either hospital, unlike the established literature [36]. However, TCM activation wasn’t systematically recorded during exams. Hence, the lack of observed effect may result from inconsistent use, setup, patient centering, or operator experience, not an inherent lack of dose-saving potential [37,38,39,40].

This study has, however, some limitations. It was conducted at only two hospitals and included only adult oncology patients, which could limit the applicability of the findings to other clinical settings and populations. CT services in the Gaza Strip are delivered through a limited number of governmental hospitals, with high-volume and oncology-related CT examinations commonly referred to central facilities. The two hospitals included in this study are among the principal public providers of CT imaging and receive referrals from across the Gaza Strip. Although inclusion of additional centres would enhance generalisability, the practices evaluated are likely representative of routine governmental CT practice in this setting. Furthermore, variations in CT scanner models and fixed protocol parameters restricted the evaluation of certain dose determinants. In addition, image quality was not systematically assessed; therefore, dose optimisation was evaluated based on radiation dose metrics alone without direct consideration of diagnostic image quality. Also, the cross-sectional design prevents assessment of long-term dose optimisation and removing extreme values may have excluded some high-dose examinations.