The purpose of this study was to establish typical DRLs for common CT indications among adult patients in Uganda. In this study, IB-DRLs for common indications of CT examinations among adults were developed as typical DRLs set at the 50th percentile of the pooled distribution of dose data from 12% (3/25) of the CT scanners in Uganda, a low-income country. The common CT scan indications were similar to those in a study within Ghana and in Europe (10, 21).
The overall median weight for all indications excluding head trauma and stroke was 74(63.6-79.7) kg with pulmonary embolism having the most heavy patients with 78(71.5-85) kg and interstitial lung diseases having the least heavy patients with 69.5(60-77) kg. This overall median weight was within that of a standard adult population defined by ICRP with 70+/-20kg. These findings were also similar to those of other studies in which heavier participants suffered more recurrent episodes of venous thromboembolism and pulmonary embolism (22) and lower weight was associated with disease progression in interstitial lung diseases (23).
CT scanning parameters:
X-ray tube voltage (measured in peak kilovoltage, kVp): The tube voltage of 130kVp that was used for all indications besides pulmonary embolism was higher than values used in other studies for example higher than 118 (±8.3) to 121.8(±7.4) in Ghana (10), 120 kVp in Egypt (11) and (100 to 120) kVp in France (20). The tube voltage of 110 (110- 114.25) kVp for pulmonary embolism was similar to trends in some studies for example (100-120)kVp in France (20) while this kVp for PE was lower than values in other studies for example 117.8(±4.0) in Ghana. There is potential to reduce CT radiation doses by lowering kVp for patients of smaller body weight, especially for ILD/HRCT by setting a specific kVp for a particular weight category on the CT machines at the hospitals as kVp in the recent CT machine models is fixed for an examination and is not modulated automatically during an examination to suit body size like it happens for x-ray tube current (24). There is hope though that newer CT scanner models may bring further improvements in the range of available tube voltages and more advanced automatic tube voltage selection tools which can automatically alter kVp according to patient size to achieve ALARA doses (25).
X-ray tube current-time product (measured in milliampere per second, mAs): Indications that did not use contrast material like ILD/HRCT generally had a lower effective mAs and a lower total mAs and compared to those that used contrast material like ABDPL, because of the lower attenuation in the absence of iodinated contrast material in the noncontrasted examinations. Indications within the chest generally had a lower total mAs than those in the abdomen and head because of the lower density/attenuation of the chest tissues for which tube current modulation software automatically reduces the mAs during the scan to minimize radiation dose (24). This trend was similar to the trend in the mAs used in a study within Ghana (10).
Scan length: The scan length for head trauma was longer than that for acute stroke due to the inclusion of a longer part of the neck to rule out concomitant cervical spine injury for early management to mitigate complications similar to a studies in Ghana and Uganda (10, 15, 16, 19).
The scan length for PE was shorter than that for ILD/HRCT due to the exclusion of the most peripheral chest parts in which emboli are not generally seen within the terminal pulmonary arterioles. This trend was similar to that in a study within Ghana (10). This trend of scan length was largely within the recommended limits of the anatomical extent for PE examinations which extends from the aortic arch to the base of the hear (26).
The scan length for UC was shorter than that for ABDPL probably due the focus on calculi within the upper urinary system as per the scan request form with less inclusion of the most distal parts of the pelvis after the urinary bladder This was dissimilar to findings in a study within Ghana (10) in which both ABDPL and kidney stones had similar scan lengths. This was also dissimilar to the recommendation by ESR that advises the scan length for UC to extend longer than that of ABDPL (appendicitis) i.e., from inferior margin of T10 to lower edge urinary bladder (approximately at lower edge of pubis symphysis) for UC and from inferior margin of T10 to superior border of pubis symphysis (26).
Scan length depends on the height of the participant and the extent of the anatomy that has to be demonstrated therefore it is important to keep it within limits that answer the clinical question for the CT scan.
Number of scan sequences: Only a precontrast scan sequence was used to examine head trauma, acute stroke and ILD/HRCT as the tissues provide adequate natural contrast to allow visualization of the questioned pathology in these indications similar to a study within Ghana (10).
In the chest, the scan sequences for PE were more than for ILD/HRCT partly due to the higher image quality requirements with contrast use and partly due to the inclusion extra sequences at some hospitals for example a precontrast HRCT phase to rule COVID-19 pneumonia during the pandemic and a delayed phase to rule emboli in pulmonary veins. These findings differed from other studies in Ghana and France that used less sequences (1-2) (10, 20).
In the abdomen, a higher-than-expected number of scan sequences were used for UC contrary to a single noncontrast sequence which is usually adequate as calculi provide high contrast to visualize them easily (4, 27). Many postcontrast scan sequences were used in UC and ABPL mainly due to absence of standardized scan protocols optimized for these indications and these findings differed from other studies that used one noncontrast scan sequence for UC and one postcontrast scan sequence for ABDPL in Ghana, France and Europe (10, 20, 21).
There is potential to reduce CT doses by lowering the number of scan sequences for PE, ABDPL and UC through adherence to the developed indication-based examination protocols where they exist or development of the optimized indication-based scan protocols where they are absent.
Slice thickness: The acquisition slice thickness was similar for all indications across all body regions even for pulmonary embolism and HRCT/ILD which require thinner slice acquisition due to the need for higher image quality. The slice thickness for indications in the head(head trauma and acute stroke) and for abdominopelvic indications (ABPL and UC) was expectedly comparable as the indications in both regions do not necessarily require very thin slice thickness to have diagnostic images similar to findings in a study within Ghana (10). The slice thickness finding for PE differed from findings of a study within Ghana in which the slice thickness for PE was lower than that of other indications in the chest region (10). It is important to note that image quality in the current study was assessed subjectively and was adequate for PE similar to the high image quality in the cited study (10) that assessed the quality objectively. There is need to assess image quality objectively in future studies to better assess if a slice thickness greater than 1mm provides good enough image quality for PE assessment.
The typical DRLs
The CTDIvol DRLs were observed to be comparable for different indications within the same body regions similar to a study in Ghana (10). The CTDIvol DRLs in the head region (acute stroke and head trauma) and abdomen region (abdominopelvic lesion and urinary calculi) were similar due to use of similar CT scan parameters (effective mAs). The CTDIvol DRLs in the indications within the chest region (ILD and PE) ended up being comparable due to use of a combination of CT scan parameters that raise and reduce CTDIvol i.e., in ILD, the lower pitch raises the CTDIvol while the lower effective mAs reduces the CTDIvol while in PE, the higher effective mAs raises the CTDIvol while the higher pitch reduces the CTDIvol.
The DLP DRLs were observed to differ among indications with the same body region due to the differences in image quality needs and use of different scan parameters. The DLP DRL for head trauma being higher than for acute stroke can be explained by a longer scan length due to the need to rule out cervical spine injury similar to a study within Ghana (10). The DLP DRL for PE being higher than for ILD/HRCT can be explained by the higher image quality need that requires contrast use and therefore a higher total mAs, plus the use of a higher-than-expected number of scan sequences. This trend of CT doses within the chest findings were similar those in a study within France (20). The DLP DRL for UC being unexpectedly higher to that for ABDPL can be explained by the absence or less frequent utilization of an optimized scan protocol for UC which allowed room for use of postcontrast sequences which were high in number than those for ABDPL. This finding was different from that in other studies in which the DLP DRL for UC was lower than for ABDPL due to use of a single noncontrast scan sequence for UC in an optimized protocol (10, 11, 20).
Comparison of the IB-DRLs to anatomical based national DRLs (AB-NDRLs) in Uganda.
The CTDIvol DRLs of all indications being lower than corresponding anatomical based values was probably due to use of a lower total mAs. The IB-DLP DRLs for head trauma, acute stroke and ILD/HRCT being lower than AB-DLP DRLs for head CT and chest CT scans respectively can be explained mainly using a lower total mAs. The need for lower image quality requirements without need for contrast media may have further contributed to the DLP DRLs of acute stroke, head trauma and ILD/HRCT being lower. Some of the current study’s indication based DLP DRLs were lower than DRLs of an anatomical region by 36.4% on average similar to findings in a studies within Finland (28) and Switzerland (29) that found IB-DRLs lower by 20% and by 21-32% respectively.
However, the DLP DRL for PE was higher than for chest CT scans due to use of thinner acquisition slice thickness and generally a higher need for higher image quality similar to another study in Switzerland (29).
The DLP DRL for ABDPL in the current study ended up being comparable to the DLP value for abdomen CT scans (Erem et al., 2022) due to a combination of CT scan parameters that raise the DLP including a longer scan length of 49.12 (46.29-53.09)cm and a thinner slice thickness of 3 (1-5) mm in ABDPL examinations compared to a scan length of 41.25 (32.0-63.3) cm and to a slice thickness of 3.75mm slice thickness that were used in abdomen CT scans (Erem et al., 2022), and due to use of a lower total mAs of 4783 (3496-7088) mAs which reduces the DLP compared to 6115.5 (3619-9869) mAs of abdomen CT scans (Erem et al., 2022). The finding of the DLP DRL of an ABDPL being comparable to the DLP of abdomen CT scans (17) differed from findings in a study within Switzerland(29) that found the DLP DRL for appendicitis to be lower than for abdomen CT scans because this cited study in Switzerland used optimized protocols for appendicitis with probably fewer scan sequences.
The DLP DRL for UC in the current study ended up being unexpectedly comparable to the high DLP value for abdomen CT scans (17) due to a combination of CT scan parameters that raise the DLP including many scan sequences that included many postcontrast phases (0-6), a longer scan length of 46.88 (42.83-49.13) cm and a thinner slice thickness of 3 (1-5) mm in UC examinations compared to 41.25 (32.0-63.3) cm scan length and to 3.75mm slice thickness that were used in abdomen CT scans (17), and due to use of a lower total mAs of 4888 (3324-6221) mAs which reduces the DLP compared to 6115.5 (3619-9869) mAs of abdomen CT scans (17). The finding of the DLP DRL of UC being comparable to the DLP of abdomen CT scans (17) differed from findings in a study within Switzerland(29) that found the DLP DRL for kidney stones to be lower than for abdomen CT scans because this cited study in Switzerland used optimized protocols for kidney stones with probably fewer scan sequences.
Comparison of the overall IB- DRLs at 75th percentile to some of the published national IB-DRLs at 75th percentile
Acute Stroke
The CTDIvol and DLP DRLs of acute stroke being much lower than values in Ghana (10)was mainly due to use of a lower effective mAs of 169-(160-203) compared to the higher tube loading of 238.0 (± 80) mAs in Ghana(10). The reasons for the CTDIvol and DLP DRLs for acute stroke being lower than France’s values (20) were not ascertained because very few CT scan parameters were mentioned in the cited study within France which limited more comparative analysis.
Head Trauma
The CTDIvol and DLP DRLs of head trauma being much lower than values in Ghana (10) was mainly due to use of a lower effective mAs of 181(168-206) compared to the mAs used in Ghana (229.4 ± 73.5 mAs) (10). The reasons for the CTDIvol DRL for head trauma being lower than, and the DLP DRL for head trauma being comparable to, the corresponding values in France(20) were not ascertained as very few scan parameters were mentioned in the cited study in France for more comparative analysis.
ILD/HRCT
The CTDIvol and DLP DRLs of ILD/HRCT being much lower than the values in Egypt (11) was probably due to use of a lower effective mAs of 61.5 compared to the mAs used in Egypt within a range of (100 minimum mAs to 430 maximum mAs)(11).
The CTDIvol and DLP DRLs for ILD/HRCT being higher than values in France (20) was due to use of a higher kilovoltage of 130 compared to 100 kVp used in most examinations (61 %) the cited study within France(20).
PE
The CTDIvol and DLP DRLs of PE being lower than values in Ghana (10) was probably due to use of a wider slice thickness of 3 (1-5) mm compared to Ghana’s (2.20 ± 1.7)mm, a lower peak tube kilovoltage of 110 (110- 114.25) kVp compared to Ghana’s (117.8 ± 4.0)kVp, and a lower effective mAs of 98 compared to Ghana’s tube loading of (167.5 ± 92.9) mAs (10). Even though the current study’s DRLs for PE were lower than Ghana’s, the quality of PE examination images was of acceptable quality to make a diagnosis as assessed by radiologists.
The reason for the CTDIvol DRL of PE being lower than the value in France (20) was unknown as the cited study did not mention its scan parameters which limited comparative analysis.
The corresponding DLP DRL for PE was instead higher than France’s value (20) due to use of a higher tube peak kilovoltage of 110 (110- 114.25) compared to 100kVp used in most examinations (55%) in the French study (20), probably due to frequent use of unoptimized examination protocols with a higher number of scan sequences (1-4) and probably due to a longer scan length compared to the usually optimized PE protocols in European countries(26).
ABDPL
The CTDIvol DRL being lower than Ghana’s value (10) was due to using a lower effective mAs of 98(81.33) compared to Ghana’s tube loading mAs of (137.0 ± 91.4)(10).
In a comparison to the cited study in Ghana(10), it was observed that most of the scan parameters used in the current study to examine an ABDPL were higher than those in Ghana for example, the current study used a higher number of postcontrast scan sequences(1-3), longer scan length 49.12 (46.29-53.09)cm, higher kilovoltage(130kVp), and thinner slice thickness 3 (1-5)mm compared to Ghana’s single postcontrast scan sequence(1), 45.99(±4.3)cm scan length, 118.7(±7.5)kVp and 5.40(±2.6)mm slice thickness respectively (10). For this reason, the DLP DRL of ABDPL in the current study was expected to be much higher than Ghana’s value but was instead comparable due to use of a lower effective mAs of 98 compared to Ghana’s tube loading mAs of (137.0 ± 91.4)(10).
The reason for the CTDIvol DRL for ABDPL being comparable to France’s value (20) was not ascertained as the cited study did not mention much about its scan parameters which limited comparative analysis .
The corresponding DLP DRL for ABDPL being higher than France’s value (20) was due to use of a higher tube peak kilovoltage of 130 compared to 100/120 kVp used for most examinations (59%) in the French study (20) and probably due to frequent use of unoptimized examination protocols with a higher number of scan sequences (2-4) compared to the usually optimized protocols for ABDPL in European countries(26).
UC
The CTDIvol DRL of UC being lower than Ghana’s value (10) was due to use of a lower effective mAs of 105 compared to Ghana’s (138.4 ± 98.4) mAs (10).
The DLP DRL for UC being higher than Ghana’s value was probably due to use of a higher kVp (130), thinner slice thickness 3 (1-5) mm and a higher total number of (1-7) scan sequences compared to Ghana’s 118(±8.3) kVp, 5.3(±2.5) mm and only one noncontrast scan sequence in the examination respectively (10).
The reason for the CTDIvol DRL of UC being comparable to France’s value (20) was not ascertained because the French study did not mention much about their CT scan parameters which limited comparative analysis.
The DLP DRL for UC being higher than France’s value (20) was due to use of a higher tube peak kilovoltage of 130 compared to 100 kVp used in most examinations (69%) in this cited French study(20) and probably due to frequent use of unoptimized examination protocols with a higher number of scan sequences (1-7) which included (0-6) postcontrast sequences compared to the optimized UC protocol with only one noncontrasted scan sequence in the French study(20).
Like in other published studies, the overall DRLs of some indications at 75th P varied significantly from national IB-DRLs in other countries mainly due to the difference in the scan parameters chosen for use in the examinations including kilovoltage, acquisition slice thickness, effective mAs and number of scan sequences plus the use of protocols that are less optimized for indications (10, 11, 21, 30).
Study strengths and limitations
The strengths of this study include having performed QC tests on the CT scanners, use of actual CT scanner output radiation doses to develop the IB-DRLs following corrections of the doses and the development of DRLs using the required minimum of 20 participants per CT scanner room for most indications as recommended by ICRP.
The main study limitation was having selected three (3) out of 13 (23%) CT facilities in Kampala and central region of Uganda, the developed typical IB-DRLs may be less representative of the doses and practices used during examinations of the selected CT indications country-wide. However, findings from the study still give an indication that locally determining typical DRLs can optimize the use of CT equipment. Further studies should be conducted using more CT facilities to develop IB-DRLs for the common CT indications that are more representative of other settings.