Utilisation of Objective and Subjective Image Quality Metrics to Guide Optimal Contrast Volumes for Abdominal CT Authors: Deborah Carrick, Mitchell Ashton, Joshua Bognar, Natalia Diaz, James Rogers, Daniel Serra, Daniel Schick Published: 2026 ABSTRACT ABSTRACTIntroductionDetermining the appropriate volume of iodinated contrast media (ICM) for portal venous phase abdominal computed tomography (CT) remains challenging. Evidence is limited regarding the volume required for consistent enhancement and whether this translates into optimal subjective image quality. To address this gap, we compared fixed‐volume and weight‐based dosing strategies, utilising objective enhancement and subjective image quality to identify approaches that may optimise diagnostic performance.MethodThe abdominal CT ICM dosing protocol was revised from fixed (100 mL) to weight‐based (1 mL/kg, 10 kg increments) volume dosing. Data from 90 fixed and 90 weight‐based ICM volume procedures were matched by age, gender, weight and kVp. Objective measures of ICM enhancement (Hounsfield unit, HU) were obtained in key vessels and organs. Subjective image quality scores were obtained from 24 cases per group, matched by weight and kVp, followed by visual grading characteristic (VGC) analysis.ResultsWeight‐based dosing reduced the dependence of key vessel and organ enhancement on patient size compared with fixed dosing. VGC analysis comparing patients < 90 kg and ≥ 90 kg suggests light patients require greater HU enhancement than heavy patients to achieve equal subjective image quality.ConclusionConsistent HU enhancement across all patient sizes can be achieved by using weight‐based ICM volume dosing in abdominal CT, although this does not necessarily translate to consistent subjective image quality. Our findings suggest that lighter patients may benefit from greater enhancement to achieve equivalent subjective image quality to larger patients, highlighting the need for further optimisation of ICM dose strategies beyond standard weight‐based approaches. Introduction Intravenous administration of iodinated contrast media (ICM) is routinely applied in abdominal computed tomography (CT) to delineate anatomy and detect pathology, with portal venous phase imaging serving as the standard approach for abdominal assessment. Determining the appropriate volume of ICM remains a persistent challenge. Historically, abdominal CT examinations have commonly used fixed ICM volumes of approximately 100–150 mL [1, 2, 3]. During the 2022 global ICM shortage, however, many imaging centres adopted weight‐based dosing strategies as a conservation measure [4], supported by studies previously demonstrating reductions in total ICM use, cost savings and decreased waste when transitioning from fixed to weight‐based ICM dosing [5, 6, 7]. The benefits from shifting from fixed to weight‐based ICM dosing are particularly valuable for ICM conservation when applied to abdominal studies, as CT abdomen or pelvis examinations have been reported to represent the largest contribution to ICM volume use (34%) from 2011 to 2024 in the US [8]. Several studies also suggest that weight‐based ICM dosing improves enhancement consistency across a broad range of patient sizes [1, 9, 10]. However, evidence is limited regarding the specific iodine dose required to achieve consistent enhancement in abdominal CT and it is unclear whether such consistency translates into optimal image quality and improved patient‐reported outcomes. Reported weight‐based dosing strategies vary widely, range from 266 to 667 mg iodine per kg body weight [1, 7, 11, 12, 13], and prior studies demonstrating more consistent enhancement with weight‐based ICM dosing have either not evaluated subjective image quality [1, 3] or have only compared subjective quality between groups without addressing the question of optimal enhancement [1, 10, 14]. To address these gaps, we compared fixed‐volume and weight‐based ICM dosing strategies, evaluating both objective enhancement metrics and subjective image quality to identify approaches that may optimise diagnostic performance. Methods The quality improvement project was submitted to the Chair of the Low Risk Review Panel of Gold Coast Hospital and Health Service (GCHHS) Human Research Ethics Committee (HREC) for advice regarding ethical and scientific review. The project was categorised as a Quality Activity and was assigned an HREC exemption reference number EX/2021/QGC/775002. In our CT practice at GCHHS, the ICM agent is 350 mg/mL iodine concentration and the fixed ICM volume protocol for portal venous abdominal CT procedures is 100 mL. The weight‐based ICM volume protocol was 1 mL/kg (350 mg iodine per kg) to be in line with published data, and broadly consistent with the fixed protocol volume. This weight‐based dosing was implemented in 10 kg increments to ensure the protocol was practical to use by simply asking the patient their approximate weight or radiographer estimation if questioning was not possible. Both methods of ‘asking the patient’ or ‘operator estimated weight’ have been previously validated against measured weight with a mean inaccuracy of 3 kg and 6 kg respectively, which was deemed acceptable for our purposes [3]. The volume was capped at a minimum of 60 mL (for < 60 kg patients) to minimise the risk of inadequate enhancement in the smallest patients. Larger patients usually have a higher percentage of body fat which is not as well perfused and as such does not uptake ICM to the same extent as visceral organs. Consequently, the volume was also capped at a maximum of 120 mL (for > 120 kg patients). Prior research has demonstrated the importance of injection duration when tailoring the iodine administration rate to patient size [1]. Consequently, flow rates were adjusted to maintain similar injection duration across body weights. The weight‐based dosing chart is shown in Table 1. This was implemented for CT CAP (chest abdomen pelvis) and CT AP (abdomen pelvis) studies utilising identical portal‐venous phase timing. TABLE 1Department weight‐based ICM volume dosing chart.Weight (kg)ICM volume (mL)Flow rate (mL/s)< 60601.860–69652.070–79752.280–89852.490–99952.6100–1091052.8110–1191153.0120+1203.2Abbreviation: ICM, iodinated contrast media. Department weight‐based ICM volume dosing chart. Abbreviation: ICM, iodinated contrast media. Image data was collected for 153 consecutive abdominal CT procedures for fixed‐volume ICM. The abdominal CT image was acquired as part of either a CT CAP or CT AP study on one of 2 Siemens Definition Flash CT scanners (Erlangen, Germany) in the department. Following implementation of the weight‐based ICM dose chart, image data was collected for a further 105 consecutive procedures over the same scanners. Both scanners were configured with identical acquisition parameters, portal‐venous phase timing and utilising automated kilovoltage selection which tends toward selection of relatively low tube kilovoltage for ICM enhanced procedures. Patient data for fixed volume and weight‐based ICM were matched by weight category, gender and age (to the nearest decade) resulting in 90 matched examinations in each group for analysis (see Table 2). CT acquisition was performed at 100 kVp for 89 patients in each group and 120 kVp for 1 patient in each group (weight category ≥ 120 kg). TABLE 2Patient demographics in the fixed ICM volume and weight‐based ICM volume groups for objective assessment.Weight categoryFixed ICM volumeWeight‐based ICM volume n Age (yrs) n Age (yrs)TotalF (M)MedianRangeTotalF (M)MedianRange< 601210 (2)7033–841210 (2)5936–9260–69179 (8)5921–83179 (8)6141–8670–79195 (14)6426–84195 (14)6027–8580–89207 (13)5830–86207 (13)6130–8690–99134 (9)6037–90134 (9)6637–84100–10941 (3)5637–6641 (3)5537–66110–11930 (3)5349–7630 (3)5752–60≥ 12020 (3)6049–7020 (3)7065–74TOTAL9036 (54)6021–909036 (54)6027–92Abbreviations: F, female; ICM, iodinated contrast media; M, male. Patient demographics in the fixed ICM volume and weight‐based ICM volume groups for objective assessment. Abbreviations: F, female; ICM, iodinated contrast media; M, male. Objective measurements of ICM enhancement (mean HU) were made in major abdominal organs, vessels and background muscle in both patient cohorts by placing a 100 mm2 circular region of interest (ROI) in the following locations—portal vein, abdominal aorta, liver parenchyma, splenic parenchyma, and psoas muscle. The portal vein and aorta are good measures of the venous and arterial enhancement respectively. The liver and splenic parenchyma are large homogenous organs which facilitate robust and accurate measurement and are good indicators of visceral enhancement. The psoas muscle demonstrates soft tissue muscular enhancement which can be correlated to muscular structures throughout the abdomen including abdominal wall, spinal and pelvic musculature. Lateral and anterior–posterior (AP) diameter were also measured at the transaxial slice of the ROIs to determine patient effective diameter (PED) (Equation 1) as a discrete variable for patient size. (1) Patient Effective Diameter=Lateral Diameter×APDiameter Multivariate linear regression, including an interaction between PED and ICM dosing group, was used to assess the effect of patient size and ICM dosing protocol on enhancement. This method was chosen as it directly tests the differences in slopes to determine whether the ICM dosing strategy alters the dependence of enhancement (mean HU) on patient size (PED). Subjective analysis was performed using visual grading of the reproduction of important anatomical structures, as the underlying theory of visual grading analysis (VGA) is that the ability to detect pathology correlates well with accurate anatomical presentation [15]. A sub‐set of the full dataset was created of 24 fixed and 24 weight‐based ICM volume dosing, matched by kVp and number in each weight category (2–4). Three blinded observers (one consultant radiologist, one senior radiology registrar and one junior radiology registrar) assigned a score to each of four key image characteristics: (1) portal vein opacification vs. aorta, (2) visualisation of liver parenchyma, (3) visualisation of splenic parenchyma, and (4) visualisation of kidney parenchyma. Absolute VGA was performed, meaning each image was assigned an independent image quality score of between 1 and 5 (1 awful; 2 poor; 3 sufficient; 4 good; 5 excellent) for each image characteristic. Patient size information (weight category) and acquisition kilovoltage (kV) were used to pair data in both cohorts to provide comparative datasets without major biases due to scanning method or patient cohort differences. A sub‐set of the full dataset was used due to limited radiologist availability for image scoring. Visual grading characteristic (VGC) analysis was applied to the absolute VGA image characteristic scores for images acquired using both fixed and weight‐based ICM volume dosing and presented similarly to receiver operating characteristic (ROC) analysis [16]. Suitable ROC software [17] was used to plot VGC‐curves and calculate the area under the curve (AUCVGC), including associated 95% confidence intervals (CI). AUCVGC is a measure of the difference in the image quality between the two ICM protocols (weight‐based vs. fixed), where a curve equal to the diagonal (AUCVGC = 0.5) indicates equality between the protocols [16]. Additional VGC analysis was performed between light (< 90 kg) and heavy (≥ 90 kg) patients within the same ICM volume dosing group and between light and heavy patients within opposite ICM volume dosing groups to interrogate subjective image quality as a function of both HU enhancement and patient size. Results Linear regression lines for mean HU values in the ROIs for both fixed and weight‐based ICM volume dosing are shown in Figure 1a–e, for each anatomical region. The slope of each multivariate regression line (HU/mm) and the interaction p‐value comparing the difference in slopes are shown in Table 3. All anatomical regions demonstrated a loss in enhancement with increasing PED for fixed ICM volumes, with psoas muscle demonstrating the weakest dependence. In contrast, the weight‐based ICM group demonstrated significantly less dependence of enhancement on PED in the portal vein, abdominal aorta, splenic parenchyma, and psoas muscle. Although using weight‐based ICM dosing reduced the size‐dependence of parenchymal enhancement in the liver, the improvement was not statistically significant. The liver retained the steepest regression slope of all anatomical regions, indicating a persistent mild dependence of enhancement on patient size. Scatter plot and linear regression lines for mean HU values for each anatomical region of interest (a—portal vein; b—abdominal aorta; c—splenic parenchyma; d—liver parenchyma; e—psoas muscle) plotted against patient effective diameter (PED) for fixed ICM volume and weight‐based ICM volume dosing. TABLE 3Slope of multivariate regression analysis, including an interaction term between PED and ICM dosing group, and interaction p‐value to assess the dependence on PED and ICM dosing group (fixed vs. weight‐based) in various anatomical regions.Anatomical regionSlope (HU/mm)Interaction p FixedWeight‐basedPortal Vein−0.381−0.011< 0.001Abdominal Aorta−0.328−0.0690.002Splenic Parenchyma−0.270−0.053< 0.001Liver Parenchyma−0.284−0.1680.065Psoas Muscle−0.125−0.0410.020Abbreviations: HU, Hounsfield Units; ICM, iodinated contrast media; PED, patient effective diameter. Slope of multivariate regression analysis, including an interaction term between PED and ICM dosing group, and interaction p‐value to assess the dependence on PED and ICM dosing group (fixed vs. weight‐based) in various anatomical regions. Abbreviations: HU, Hounsfield Units; ICM, iodinated contrast media; PED, patient effective diameter. The AUCVGC comparing fixed and weight‐based ICM dosing for all observers combined was 0.52 with 95% CI 0.47–0.57. The VGC analysis for ‘consultant radiologist’ scores gave a similar result to ‘all observers’ with an AUCVGC of 0.47 with 95% CI 0.39–0.55. This result indicates no ability to distinguish between the 2 protocols in terms of observer rated image quality across the full sub‐set of data. However, reviewing Figure 1 shows the impact of weight‐based ICM dosing was not consistent across all patient sizes. Enhancement increased for heavier patients and decreased for light patients when compared to the fixed ICM dose group. By assessing the difference between the regression lines in Figure 1 for each anatomical region, we estimated the point of crossover to be 320 mm which corresponded to approximately 90 kg in the study population. To more thoroughly interrogate the implication of this on subjective image quality, light (< 90 kg) and heavy (≥ 90 kg) patient groups were separated and VGCA performed for these subgroups both within the same volume ICM dose group (fixed light vs. fixed heavy; weight‐based light vs. weight‐based heavy) and between each volume ICM group (fixed light vs. weight‐based heavy; weight‐based light vs. fixed heavy). The corresponding AUCVGC and 95% CI are shown in Tables 4 and 5 respectively. TABLE 4AUCVGC ± 95% CI for VGC analysis comparing light (< 90 kg) vs. heavy (≥ 90 kg) patient groups within each ICM volume dosing group (fixed and weight‐based) for all observer scores and consultant scores only.Fixed light vs. Fixed heavyWeight‐based light vs. Weight‐based heavyAll observersConsultant onlyAll ObserversConsultant onlyAUCVGC 0.620.600.370.38Lower 95% CI0.560.480.310.27Upper 95% CI0.690.710.440.50Abbreviations: AUC, area under curve; CI, confidence interval; ICM, iodinated contrast media; VGC, visual grading characteristic. AUCVGC ± 95% CI for VGC analysis comparing light (< 90 kg) vs. heavy (≥ 90 kg) patient groups within each ICM volume dosing group (fixed and weight‐based) for all observer scores and consultant scores only. Abbreviations: AUC, area under curve; CI, confidence interval; ICM, iodinated contrast media; VGC, visual grading characteristic. TABLE 5AUCVGC ± 95% CI for VGC analysis comparing light (< 90 kg) vs. heavy (≥ 90 kg) patient groups between opposing ICM volume dosing group (fixed and weight‐based) for all observer scores and consultant scores only.Fixed light vs. Weight‐based heavyWeight‐based light vs. Fixed heavyAll observersConsultant onlyAll observersConsultant onlyAUCVGC 0.460.490.530.54Lower 95% CI0.390.370.460.42Upper 95% CI0.530.600.600.65Abbreviations: AUC, area under curve; CI, confidence interval; ICM, iodinated contrast media; VGC, visual grading characteristic. AUCVGC ± 95% CI for VGC analysis comparing light (< 90 kg) vs. heavy (≥ 90 kg) patient groups between opposing ICM volume dosing group (fixed and weight‐based) for all observer scores and consultant scores only. Abbreviations: AUC, area under curve; CI, confidence interval; ICM, iodinated contrast media; VGC, visual grading characteristic. Results in Table 4 (columns 2 and 3) show that subjective image quality was improved for fixed light compared to fixed heavy patients (AUCVGC and 95% CI > 0.5). This corresponds with increased vessel and organ enhancement in the light (< 90 kg) patient group (Figure 1). Results in Table 4 (columns 4 and 5) demonstrate that subjective image quality was improved for weight‐based heavy compared to weight‐based light patients (AUCVGC and 95% CI < 0.5) despite more consistent organ enhancement across all images (Figure 1). Results in Table 5 show that no difference in subjective image quality was demonstrated between the 2 higher subjective image quality groups from Table 4 (fixed light vs. weight‐based heavy) and the 2 lower subjective image quality groups (weight‐based light vs. fixed heavy) as the 95% CI for AUCVGC bounded 0.5 in each case. In Tables 4 and 5, the AUCVGC for all observers was generally consistent with the consultant only scores. However, the 95% CI upper and lower boundaries were wider for the consultant only scores due to the smaller number of data points. Discussion Our study has utilised objective (mean HU in key vessels and organs) and subjective (VGC analysis) image quality metrics to compare fixed and weight‐based ICM volume dosing for abdominal CT and identify approaches to optimise diagnostic performance. Multivariate linear regression, including an interaction between PED and ICM dosing group, was used to assess the patient size‐dependence of HU enhancement in each group. In the fixed ICM dose group, enhancement in the portal vein, abdominal aorta, splenic parenchyma, and liver parenchyma decreased by between 0.27–0.38 HU per mm PED, indicating substantial size‐dependent reduction in enhancement. Enhancement in the psoas muscle decreased by 0.13 HU per mm PED, demonstrating a milder reduction with size. In contrast, the weight‐based dose group showed a nearly flat relationship (reduction of 0.01–0.07 HU per mm PED) for portal vein, abdominal aorta, splenic parenchyma, and psoas muscle. The difference in slopes between groups was highly statistically significant (Table 3), confirming that weight‐based dosing improves enhancement consistency in these vessels and organs across body sizes. Although the liver parenchyma demonstrated a mild reduction in slope between the fixed and weight‐based ICM groups (−0.28 and −0.17 HU per mm, respectively), the difference is not statistically significant (p = 0.07). This is likely to be due to the dual blood supply of the liver from both the portal vein and hepatic artery, resulting in parenchymal enhancement being less dependent on ICM volume than other vascular organs. These results are consistent with previous studies that have compared fixed and weight (or LBM) based ICM protocols [1, 10, 14]. However, this finding alone does not necessarily translate to optimal subjective image quality across all patient sizes. VGC analysis between the fixed and weight‐based ICM volume dosing groups indicates no ability to distinguish between the groups in terms of observer rated image quality across the small sub‐set of data, as demonstrated by the AUCVGC of 0.52 (95% CI 0.47–0.57). This result is consistent with previous studies where subjective image quality assessment yielded overall comparable results between fixed and LBM‐based ICM doses [3, 4]. However, the extended VGC analysis performed in this study between light (< 90 kg) and heavy (≥ 90 kg) patients indicates that subjective image quality may have a dependence on weight for both fixed and weight‐based ICM volume groups (Table 3). This difference in results shows that comparing the mean enhancement or subjective image quality scores between the groups over all patient sizes may oversimplify the findings. The AUCVGC in Table 4 (columns 2 and 3) indicates superior subjective image quality in fixed light (< 90 kg) compared to fixed heavy patients (≥ 90 kg), which is expected due to the greater ICM organ enhancement (Figure 1) for lighter patients. However, the AUCVGC in Table 4 (columns 4 and 5) indicates that despite consistent ICM organ enhancement with patient size for weight‐based ICM volume dosing (Figure 1), subjective image quality may be improved for weight‐based heavy compared to weight‐based light patients. This finding suggests that greater enhancement is required in lighter patients to achieve comparable subjective image quality to heavy patients. The results in Table 5 also support this, as there is no demonstrated difference in observer‐rated image quality scores (95% CI for AUCVGA bounded 0.5) between fixed ICM dose in light patients and weight‐based ICM in heavy patients despite greater enhancement in the light patient group. In addition, there is also no demonstrated difference between weight‐based ICM in light patients and fixed ICM in heavy patients, despite greater enhancement in the light patient group. Published studies that utilise subjective image quality metrics have generally compared fixed and weight‐based ICM dosing across all patient sizes, which does not address the different requirements of light and heavy patients with respect to ICM enhancement [1, 10, 14]. Megibow et al. [18] compared contrast enhancement and subjective image quality (by reviewing total number of studies with unacceptable enhancement) across different weight groups. Consistent with the results in this study, the authors noted a larger proportion of unacceptable ICM enhancement in patients in the low weight group compared to the high weight group and suggested two potential reasons for this. Firstly, the study started all scans immediately after contrast injection was completed, meaning small patient (low ICM volume) scans started earlier than large patient (high ICM volume) scans and potentially too early to see peak enhancement. The authors stated that using bolus tracking could give different results, however the protocol in our study utilised bolus tracking and yielded similar outcomes, suggesting that this peak enhancement issue did not significantly influence the results. The second reason suggested by the authors was that establishing a fixed HU enhancement across all patient sizes may not be important or even clinically desirable since radiologists may be more critical of a scan with less intra‐abdominal fat and therefore be less willing to tolerate lower levels of enhancement in small patients. This conclusion is consistent with a widely held view that less noise is tolerable in lighter or smaller patients with reduced abdominal tissue contrast due to low fat content. Our study agrees with previously published work that the implementation of weight‐based ICM volume dosing achieves more consistent enhancement of major organs and vessels in abdominal CT imaging across all patient sizes, compared to fixed ICM volume dosing. However, additionally utilising VGC analysis of subjective image quality scores across different patient sizes (light vs. heavy) between the fixed and weight‐based ICM volume dosing groups suggests that consistent enhancement does not necessarily result in consistent subjective image quality across all patient sizes, and that increased enhancement may be required in lighter patients to achieve equivalent subjective image quality to larger patients. These conclusions suggest that different strategies may be required to optimise contrast enhancement in abdominal portal venous CT examinations across a range of patient sizes. This may be achieved by a combination of fixed (lighter patients) and weight‐based (heavier) ICM dosing protocols, or weight‐based ICM dosing protocols that utilise an increased mL/kg ICM in lighter patients compared to heavier patients. However, neither approach has been trialled in our institution. This study has limitations. For the objective image quality analysis, there were 90 scans in each group, matched by age, gender, weight category and kVp. However, due to the smaller number of patient images used for subjective image quality analysis (24 in each group) these were only matched by weight category and kVp, so any inherent differences in subjective image quality due to age and gender (e.g., differences in patient weight distribution) were not accounted for. Also, of the three radiologists scoring the images, only one was a consultant with two being registrars. However, the derived ‘all radiologist’ and ‘consultant only’ AUCVGC values aligned closely, but with wider 95% CI for the ‘consultant only’ scores due to the smaller number of data points. Further investigation into appropriate dosing for very large patients may also be warranted. Local scan protocols involve an increase in the kVp for the largest patients. This technique adjustment produces an X‐ray beam with sufficient penetration of the greater tissue volume to provide adequate CT image signal to noise ratio. However, a higher kVp beam will result in fewer photoelectric absorption interactions in the ICM, reducing enhancement. Therefore, there may be some benefit in raising the maximum ICM volume above 120 mL (maximum used in this study) for patients greater than 120 kg to boost the enhancement of key organs and vessels and maintain optimal image quality. There were limited patients greater than 120 kg included in this study (n = 6 and n = 8 for fixed and weight‐based ICM volume respectively) which compromised the review of the impact in this cohort. Conclusions Our study demonstrates that weight‐based ICM volume dosing improves HU enhancement of major vessels and vascular organs in abdominal CT imaging across all patient sizes, compared to fixed ICM volume dosing. However, the additional utilisation of subjective image quality scoring suggests that consistent enhancement does not necessarily translate to consistent subjective image quality. Our findings suggest that lighter patients may benefit from greater enhancement to achieve equivalent subjective image quality to larger patients, highlighting the need for further optimisation of ICM dose strategies beyond standard weight‐based approaches. Funding The authors have nothing to report. Ethics Statement The quality improvement project was submitted to the Chair of the Low Risk Review Panel of Gold Coast Hospital and Health Service Human Research Ethics Committee (HREC) for advice regarding ethical and scientific review. The project was categorised as a Quality Activity and was assigned an HREC reference number EX/2021/QGC/775002. Consent The authors have nothing to report. Conflicts of Interest The authors declare no conflicts of interest.