Vendor-agnostic versus vendor-specific boost methods in low-concentration contrast-enhanced computed tomography of the abdomen

Introduction

Contrast-enhanced abdominal computed tomography (CT) is a crucial imaging modality widely utilized in the evaluation of various abdominal pathologies, including postoperative complications, inflammatory conditions, oncologic assessment, and undifferentiated nonspecific abdominal pains (1). Contrast media-related adverse events pose serious risks, not only due to allergic reactions but also for patients with renal failure, older age, heart failure, and diabetes mellitus, primarily owing to the high injection rate or volume of the contrast medium (2,3). To reduce this risk, reduced doses of the contrast medium are recommended in clinical practice (4,5). Lowering the concentration of contrast media can help alleviate patient discomfort compared with the use of higher concentrations (6). During the shortage of iodinated contrast media in 2022, several studies investigated the optimization of image quality by administering a reduced contrast media dose (7).

Various efforts have been made to maximize vascular contrast enhancement (CE) using low-concentration contrast media (8,9). Lower tube voltages can achieve higher vascular CE while reducing the radiation dose. However, this approach tends to increase the image noise, which adversely affects the detectability of low-contrast lesions on abdominal CT (10). Commercially available vendor-specific deep learning reconstruction [e.g., Advanced Intelligent Clear-IQ Engine (AiCE), Canon Medical Systems; TrueFidelity, GE Healthcare; Precise Image, Philips] provides higher image quality with lower image noise and reduced artifacts than conventional hybrid iterative reconstruction, utilizing model-based iterative reconstruction and filtered-back projection images as ground truth references (11-14).

Advances in artificial intelligence have resulted in reduced image noise and enhanced vascular contrast attenuation without increasing the flow rate or concentration of contrast media (15,16). Vendor-agnostic deep-learning-based contrast-boosting algorithms (DL-CBs) (ClariCT.ACE, ClariPi) extract the iodine component images using a U-net based architecture (17). Conversely, vendor-specific boost technique (CE-boost, Canon Medical Systems Corporation) significantly increase CT attenuation through non-rigid registration using a subtraction method (18). Several studies have demonstrated significantly higher CT attenuation using both vendor-agnostic and vendor-specific boost methods on aortic, abdominal, and chest CT angiography (15,17,19-21). Moreover, the application of vendor-agnostic and vendor-specific boost methods has been shown to increase lesion detectability in patients with chronic liver disease and hepatocellular carcinoma (21,22), even when using low-concentration iodinated contrast media (270 mg I/mL) (23). However, to our knowledge, no study has comprehensively analyzed both quantitative and qualitative variables comparing vendor-agnostic and vendor-specific boost methods for contrast-enhanced abdominal CT imaging. Thus, the present study aimed to compare quantitative and qualitative image characteristics as well as lesion conspicuity in abdominal vessels and liver lesions between vendor-agnostic and vendor-specific boost methods in low-concentration contrast medium settings against standard-concentration contrast medium settings. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0096/rc).

Methods

Patients

This single-center study was approved by the Institutional Review Board of Wonju Severance Christian Hospital (IRB No. CR325017), with a waiver of informed consent due to the retrospective nature, and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments or comparable ethical standards. We retrospectively reviewed the radiology database to identify patients who underwent contrast-enhanced abdominal CT using low-concentration contrast media (270 mg I/mL) between October 2024 and December 2024, and those who underwent CT with standard-concentration (350 mg I/mL) between August 2022 and August 2024. Patients were excluded if they exhibited severe motion artifacts, age less than 18 years old, double energy acquisition, lacked raw data necessary for image reconstruction, had more than ten lesions, or absence of portal venous phase acquisition.

CT image acquisition

All CT examinations were performed using a 320-detector row CT scanner (Aquilion ONE PRISM; Canon Medical Systems Corporation). The scanning parameters were as follows: tube voltage, 100 kVp; tube current, 100–300 mA; gantry rotation time, 0.275 s; detector collimation, 130 mm × 0.5 mm in the craniocaudal direction; field of view, 320 mm; matrix, 512×512; slice thickness, 0.5 mm, and pitch, 0.813. Two different concentrations of iodinated contrast media were used in this study; the standard-concentration and low-concentration groups received a contrast medium at a concentration of 350 and 270 mg I/mL, respectively. In both study groups, 85 mL of contrast medium was administered at a rate of 3 mL/s via the antecubital vein for patients weighing less than 60 kg, and 95 mL for those weighing more than 60 kg, followed by a 30 mL saline flush at the same flow rate. The injections were performed using a dual-head power injector (Dual Shot Alpha 7; Nemoto Kyorindo Co., Ltd.). Scanning was initiated using an automatic bolus-tracking program (^SUREStart, Canon Medical Systems Corporation) within the region of interest (ROI) placed in the ascending aorta and a trigger threshold of 100 Hounsfield units (HU). Arterial and portal venous phase images were acquired at 35 and 80 s, respectively, after the administration of the contrast medium. All images were reconstructed using the deep learning reconstruction body sharp standard option (AiCE; Canon Medical Systems Corporation).

Vendor-specific and vendor-agnostic boost methods

For the vendor-specific boost method, CE-boost images (Canon Medical Systems Corporation) were generated by combining contrast-enhanced images with iodinated images, which were created by subtracting contrast-enhanced images from non-contrast images using postprocessing (18). As a vendor-agnostic boost method, a DL-CB (ClariCT.ACE, ClariPi) derives iodine-specific features, scales them by a fixed factor, and recombine them with the original image using a U-Net-based architecture (17). A denoising model for the body standard with a noise blending factor of 0.2 and an enhancing model for the body with an enhancing blending factor of 0.8 were used.

Quantitative image analysis

In each patient, objective image analysis was performed by a radiologist with 5 years of experience (C.O.). ROIs, each measuring 10 mm², were placed in subcutaneous fat tissue to determine image noise, which was calculated as the standard deviation of the CT number. The largest possible ROIs for vessels, including the abdominal aorta, celiac trunk, common hepatic artery, superior mesenteric artery, main portal vein, right portal vein, and left portal vein, were drawn at the center of the vessels, avoiding the vessel wall, stent, and calcification. Another ROI of 10 mm² was placed in the erector spinae muscle. The ROIs of liver lesions were as large as possible while avoiding areas of necrosis, calcification, and hemorrhage. ROI sizes and locations were kept consistent across vendor-agnostic CE-boost, vendor-specific CE-boost, and standard-concentration images. The liver lesions were subsequently reviewed and confirmed by a radiologist with 15 years of experience (J.H.A.). The signal-to-noise ratio (SNR) of vessels was calculated as the CT value of each vessel ROI divided by its standard deviation. The contrast-to-noise ratio (CNR) of vessels was calculated as the difference between CT values of each vessel and muscle divided by image noise. The CNR of liver lesions was calculated as the difference in HU between liver lesions and normal liver parenchyma divided by image noise. In normal liver parenchyma, ROIs were positioned in homogeneous regions, avoiding vessels, focal lesions, and artifacts. For lesion characterization, the arterial phase was used for the hemangiomas, hepatocellular carcinoma, and nodular arterioportal shunts, whereas the portal venous phase was applied for cysts, metastases, bilomas.

The noise power spectrum (NPS) was analyzed to evaluate noise texture and magnitude using the open-source software package imQuest (https://deckard.duhs.duke.edu/~samei/tg233.html). The noise amplitudes were compared using the noise magnitude and NPS peak, and the noise texture was assessed using the NPS average spatial frequency (24). For each patient, the NPS curve was generated by positioning ROIs of 5 mm² in the abdominal aorta at the level of the celiac artery, carefully avoiding areas of calcification.

Image sharpness was assessed using a non-reference quantitative blur metric test (25,26). The results were expressed as numeric values ranging from 0 to 1, with lower values indicating sharper images.

Qualitative image analysis

Qualitative image characteristics of the arterial and portal phase images were independently analyzed by two radiologists with 15 and 28 years of experience in diagnostic radiology, respectively. Both radiologists were blinded to the vendor-agnostic and vendor-specific boost methods. Image quality was graded using a 5-point Likert scale: 5 points (excellent image quality, minimal image noise, excellent vessel enhancement, excellent lesion conspicuity, minimum artifacts, minimal artificial sensation), 4 points (good image quality, some image noise, above-average vessel enhancement, above-average lesion conspicuity, less-than-average artifacts, some artificial sensation), 3 points (moderate image quality, moderate image noise, acceptable vessel enhancement, acceptable lesion conspicuity, average artifacts, moderate artificial sensation), 2 points (poor image quality, severe image noise, suboptimal vessel enhancement, suboptimal lesion conspicuity, above-average artifacts, severe artificial sensation), and 1 point (very poor image quality, very severe image noise, very poor vessel enhancement, very poor lesion conspicuity, unacceptable artifacts, and very severe artificial sensation) (15,26). Lesion conspicuity was assessed in the arterial phase for hemangiomas, hepatocellular carcinoma, and nodular arterioportal shunts, and in the portal venous phase for cysts, metastases, bilomas. The window levels and settings were freely adjusted during the evaluation.

Statistical analysis

Continuous variables are presented as mean ± standard deviation. Data normality was assessed using the Kolmogorov-Smirnov and Shapiro-Wilk tests. Body mass index (BMI) was compared between two groups using the unpaired t-test. CT attenuation, image noise, SNR, CNR, NPS magnitude, NPS peak, NPS average frequency, blur metrics, and image quality among the standard-concentration, vendor-agnostic CE-boost, and vendor-specific CE-boost groups were compared using one-way analysis of variance (ANOVA). Post hoc pairwise comparisons were performed using Tukey’s, which accounts for multiple comparisons. A post hoc power analysis was performed using G*power (version 3.1; Heinrich Heine University Dusseldorf, Germany). The analysis demonstrated a statistical power of 0.93 with a sample size of 50 patients per group, assuming an effect size of 0.5 and an alpha level of 0.05. Interobserver agreement was evaluated using the intraclass correlation coefficient. Statistical significance was defined as P<0.05. All statistical analyses were conducted using the SPSS statistical software version 25.0 (IBM).

Results

Table 1 summarizes patient demographics. This study enrolled 80 patients (mean age, 63±10 years; 57 men) who received standard-concentration contrast media and 80 patients (mean age, 64±10 years; 61 men) who received low-concentration contrast media for contrast-enhanced abdominal CT. There were no significant differences in age (P=0.75), sex (P=0.99), and BMI (P=0.98) between the two groups. The mean size of the identified liver lesions was 21.3±19.42 mm in the standard-concentration contrast media and 17.33±14.58 mm in the low-concentration contrast media group.

Standard-concentration vs. low-concentration groups

The two groups did not significantly differ in image noise (standard-concentration 8.29±1.03 vs. low-concentration 8.56±1.53, P=0.55). Table 2 summarizes the CT attenuation results. The low-concentration group had significantly (all P<0.001) lower CT attenuation in the abdominal aorta, celiac artery, superior mesenteric artery, common hepatic artery, main portal vein, right portal vein, and left portal vein than the standard-concentration group. Moreover, the low-concentration group showed significantly lower SNR and CNR values in the vessels than the standard-concentration group (P<0.001). The CNR of the liver lesions did not significantly differ between the low-concentration and standard-concentration groups (Figure 1).

Figure 1 SNR and CNR values of the standard-concentration and low-concentration groups with and without boost. Both vendor-agnostic and vendor-specific boost methods significantly improved the SNR and CNR values in the low-concentration group relative to the standard-concentration group. No significant differences were found between the vendor-specific and vendor-agnostic boost methods regarding SNR and CNR. *, P<0.03; **, P<0.002; ***, P<0.001; ns, no significance. CNR, contrast-to-noise ratio; SNR, signal-to-noise ratio.

The low-concentration and standard-concentration groups had not only equivalent NPS noise magnitude (P=0.49), average frequency (P=0.25), and blur metric values (P>0.99; Table 3), but also equivalent subjective image quality, including overall image quality, image noise, vessel enhancement, artifacts, and artificial sensation (Table 4).

Standard-concentration group vs. vendor-agnostic and vendor-specific CE-boost groups

Both vendor-agnostic and vendor-specific boost methods significantly reduced noise in images of the low-concentration group compared to those of the standard-concentration group (P<0.001). While the boost methods resulted in higher CT attenuation in the vessels of the low-concentration group, both SNR and CNR values were also significantly improved compared to the standard-concentration group (all P<0.001). Although both boost methods resulted in significantly higher CNR values of liver lesions in the low-concentration group than in the standard-concentration group, the vendor-specific method showed a slightly higher CNR than the vendor-agnostic method.

Both vendor-agnostic (7.46±1.52 HU) and vendor-specific (7.73±1.44 HU) boost methods noticeably improved image noise reduction and clarity, as indicated by significantly lower NPS noise magnitudes compared to that of the standard-concentration group (9.99±1.01 HU; P<0.001). The vendor-specific boost method (0.19±0.03 mm⁻¹) resulted in a greater reduction in average frequency, indicating superior suppression of high-frequency noise compared to the standard-concentration group (0.31±0.03 mm⁻¹; Figure 2). Conversely, image sharpness improved with both vendor-specific (0.31±0.02) and vendor-agnostic (0.29±0.01) boost methods as evaluated using the no-reference blur metric test. Both boost methods had slightly lower scores for overall image quality, image noise, and artifact assessment than the standard-concentration group. Notably, vessel enhancement was rated excellent for both boost methods and equivocal for the standard-concentration group (Figure 3). However, both vendor-specific and vendor-agnostic boost methods had significantly lower artificial sensation scores than the standard-concentration group (both P<0.001; Table 4).

Figure 2 NPS of each study group. The vendor-specific boost method had better noise suppression, with a lower NPS average frequency and minimal visible noise in the texture, particularly for tasks requiring low noise at the dominant spatial frequency, compared to the vendor-agnostic boost method. HU, Hounsfield unit; NPS, noise power spectrum.

Figure 3 Representative axial images demonstrating the outcomes of vendor-specific and vendor-agnostic boost methods. The hepatocellular carcinoma (red arrows) is clearly visualized with both boost methods in the low-concentration contrast medium group. Vessel enhancement (white arrows) is significantly improved by both vendor-specific and vendor-agnostic boost methods.

Vendor-agnostic vs. vendor-specific boost methods in the low-concentration group

In the low-concentration group, the vendor-specific method (6.06±1.54) significantly reduced the image noise compared to the vendor-agnostic method (6.41±1.17; P<0.001). Although the vendor-agnostic method resulted in significantly higher CT attenuation of vessels than the vendor-specific method (Table 2), no significant SNR and CNR differences were observed between the two methods. Likewise, the CNR values of liver lesions did not significantly differ between the vendor-agnostic and vendor-specific boost methods (Figure 1).

Additionally, the vendor-specific boost method (0.19±0.03 mm⁻¹) led to a more substantial reduction in average frequency, suggesting it more effective at suppressing high-frequency noise than the vendor-agnostic boost method (0.26±0.04 mm⁻¹; Figure 2). Conversely, the vendor-agnostic method (0.29±0.01) resulted in sharper images than the vendor-specific method (0.31±0.02), as assessed by the no-reference blur metric test (Table 3). Both observers rated the vendor-agnostic and vendor-specific boost methods as equivalent regarding image noise and vessel enhancement; however, the vendor-specific boost method received slightly higher ratings for overall image quality, artifacts, and artificial sensations (Figure 4).

Figure 4 Contrast-enhanced abdominal CT images using vendor-specific and vendor-agnostic boost methods. Vascular enhancement is significantly improved with both vendor-agnostic (arterial phase: 4.98±0.10; portal phase: 4.99±0.08) and vendor-specific (arterial phase: 4.98±0.11; portal phase: 4.99±0.07) boost methods compared to the non-boost group (arterial phase: 4.91±0.28; portal phase: 4.93±0.24) in the low-concentration contrast medium group. Although both boost methods had lower artificial sensation scores than the standard-concentration group, the vendor-specific method resulted in significantly higher scores from both observers than the vendor-agnostic method (arterial phase: 4.17±0.54 vs. 3.40±0.55; portal phase: 4.34±0.59 vs. 3.51±0.50), suggesting a more natural image appearance. CT, computed tomography.

Lesion conspicuity

Lesion conspicuity significantly differed among the low-concentration group without boosting (4.43±0.62), vendor-agnostic (4.82±0.40), and vendor-specific (4.82±0.40) boost methods (P<0.001). Although both boost methods significantly improved lesion conspicuity compared to low-concentration images without boosting, no significant difference was observed between the two boost methods (P=0.99).

Discussion

Both vendor-agnostic and vendor-specific boost methods resulted in higher image quality regarding image noise, SNR, and CNR in liver lesions and vessels; noise texture; image sharpness; and vessel enhancement compared to standard-concentration contrast medium settings. The vendor-specific method received slightly higher ratings for overall image quality, artifact reduction, and artificial sensation than did the vendor-agnostic method.

Vascular CE increases correlate directly with both the concentration and flow rate of the contrast medium. Several studies have demonstrated significant differences in vascular enhancement with higher iodine contrast concentrations compared to low-concentration contrast media (27,28). In contrast, the incidence of adverse effects including heat sensation, local pain, abdominal pain, nausea, and vomiting was markedly lower with the administration of low and moderate concentrations than with high concentrations of iodine contrast media (29,30). Therefore, achieving higher vascular CE while utilizing a lower iodine concentration in the contrast media is crucial for minimizing the risk of contrast-induced adverse effects, especially in patients requiring multiple contrast-enhanced CT examinations for follow-up evaluation, treatment response, and staging (31,32). Although the present study demonstrated significantly lower CT attenuation and reduced SNR and CNR in the low-concentration group than in the standard-concentration group, the blur metric value, noise magnitude, and subjective image analysis, including image noise and vessel enhancement, showed no significant differences between the two groups. Therefore, we believe that low-concentration contrast media could not only help reduce patient discomfort (29,30) but also maintain subjective image quality, even in the absence of vendor-agnostic and vendor-specific boost methods.

Previous studies with vendor-agnostic and vendor-specific boost methods demonstrated equivalent vascular attenuation for low and standard concentrations of iodine contrast media, effectively increasing CT attenuation (15,17,19,33,34). Hou et al. investigated the image quality of the portal vein using a vendor-specific boost method with two contrast medium concentrations [320 mg I/mL at 3 mL/s and 370 mg I/mL at 4.5 mL/s (19)] and concluded that the combination of this boost method with a lower iodine concentration achieved significantly higher CT attenuation and CNR than did a higher contrast medium concentration. Shin et al. evaluated the image quality of a vendor-agnostic boost method using a low-dose protocol in comparison with the standard-dose protocol (350 mg I/mL) in children (15). Their findings highlighted the efficacy of this boost method in maintaining the image quality while utilizing a reduced iodine concentration. Our findings align with previous results, demonstrating that both vendor-agnostic and vendor-specific boost methods significantly improve CT attenuation, even with the administration of low-concentration contrast media. However, to our knowledge, no study has been conducted to compare quantitative and qualitative image analyses between vendor-specific and vendor-agnostic boost techniques following the administration of low-concentration contrast media. The mean CT attenuation improved by approximately 61.0% and 44.5% with the vendor-agnostic and vendor-specific boost methods, respectively. Moreover, the vendor-agnostic method yielded significantly higher CT attenuation than the vendor-specific technique. In contrast, the SNR and CNR values did not significantly differ between both methods.

In our comprehensive quantitative analysis, including an assessment of image noise texture and sharpness, NPS noise magnitude did not significantly differ between vendor-agnostic and vendor-specific boost methods. Notably, the vendor-specific method led to a more substantial reduction in the average frequency, suggesting that it more effectively suppressed high-frequency noise than the vendor-agnostic method. By contrast, the vendor-agnostic method resulted in sharper images than the vendor-specific method, as evaluated using a non-reference blur metric test. Moreover, the vendor-agnostic method was rated as having a moderate artificial sensation rather than a natural appearance, whereas the vendor-specific boost method was rated as having some artificial sensation, with a significant difference between the two methods. Therefore, the vendor-agnostic boost method, which relies on a deep-learning-based contrast boosting algorithm, remains challenging regarding artificial sensation, as perceived by human readers, compared with the vendor-specific method. These differences may be attributed to the distinct underlying approaches used by each method to enhance CT attenuation. The vendor-agnostic method enhances CT attenuation exclusively based on postcontrast images without considering precontrast information, potentially leading to overemphasis of nonvascular structures and an increase in artificial enhancement or artifacts. In contrast, the vendor-specific boost employs a subtraction approach, utilizing both pre- and postcontrast images to generate an iodine map, which is then incorporated into the postcontrast image to selectively enhance CT attenuation while minimizing structured noise.

Our study has certain limitations. First, the sample size was relatively small, which may limit the generalizability of our findings. Second, this study focused on a single-center population and used only one specific CT scanner model. Third, the study did not directly compare clinical outcomes or diagnostic accuracy because the analysis was limited to image quality parameters. Currently, only a single vendor offers commercially available CE-boost method. Finally, we used fixed default values for the enhancing and noise blending factors in the vendor-agnostic CE-boost method. Further multicenter studies with larger sample sizes and direct clinical comparisons would be beneficial for validating these findings.

Conclusions

The low-concentration group demonstrated non-inferior performance to the standard-concentration group regarding the CNRs of liver lesions, blur metric, noise texture, and subjective image quality. Although vendor-agnostic and vendor-specific CE-boost methods effectively enhanced image quality and CT attenuation in low-concentration contrast medium settings, subjective image analysis revealed lower ratings due to increased artificial appearance. The vendor-agnostic method improved image sharpness, whereas the vendor-specific method excelled in reducing structured noise, offering alternatives to reduce the contrast dose and minimize the risk of contrast-related adverse effects.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0096/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0096/dss

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0096/coif). C.O. and H.S. are employees of Canon Medical Systems Korea, Seoul, Korea (the subsidiary in Korea of Canon Medical Systems Corporation, Otawara-si, Japan). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This single-center study was approved by the Institutional Review Board of Wonju Severance Christian Hospital (IRB No. CR325017), with a waiver of informed consent due to the retrospective nature, and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments or comparable ethical standards.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Pickhardt PJ. Positive Oral Contrast Material for Abdominal CT: Current Clinical Indications and Areas of Controversy. AJR Am J Roentgenol 2020;215:69-78. [PubMed]
2. Junck L, Marshall WH. Neurotoxicity of radiological contrast agents. Ann Neurol 1983;13:469-84. [PubMed]
3. Kariyanna PT, Aurora L, Jayarangaiah A, Das S, Gonzalez JC, Hegde S, McFarlane IM. Neurotoxicity Associated with Radiological Contrast Agents Used during Coronary Angiography: A Systematic Review. Am J Med Case Rep 2020;8:60-6.
4. Barrett BJ, Parfrey PS. Clinical practice. Preventing nephropathy induced by contrast medium. N Engl J Med 2006;354:379-86. [PubMed]
5. Karout L, Kalra MK. Survey of CT radiation doses and iodinated contrast medium administration: an international multicentric study. Eur Radiol 2025;35:1915-32. [PubMed]
6. Sun YX, Shang J, Cui Y, Zhang KJ, Li XT, Li DN, Gu XL, Tian XY, Sun N, Zhang L, You MN, Zhang XY, Sun YS. Impact of different concentration iodinated contrast media on pain and comfort in abdominal computed tomography. Eur J Radiol 2024;179:111664. [PubMed]
7. Koeppel DR, Boehm IB. Shortage of iodinated contrast media: Status and possible chances - A systematic review. Eur J Radiol 2023;164:110853. [PubMed]
8. Lell MM, Jost G, Korporaal JG, Mahnken AH, Flohr TG, Uder M, Pietsch H. Optimizing contrast media injection protocols in state-of-the art computed tomographic angiography. Invest Radiol 2015;50:161-7. [PubMed]
9. Watanabe H, Kanematsu M, Miyoshi T, Goshima S, Kondo H, Moriyama N, Bae KT. Improvement of image quality of low radiation dose abdominal CT by increasing contrast enhancement. AJR Am J Roentgenol 2010;195:986-92. [PubMed]
10. Kalra MK, Maher MM, Toth TL, Hamberg LM, Blake MA, Shepard JA, Saini S. Strategies for CT radiation dose optimization. Radiology 2004;230:619-28. [PubMed]
11. Koetzier LR, Mastrodicasa D, Szczykutowicz TP, van der Werf NR, Wang AS, Sandfort V, van der Molen AJ, Fleischmann D, Willemink MJ. Deep Learning Image Reconstruction for CT: Technical Principles and Clinical Prospects. Radiology 2023;306:e221257. [PubMed]
12. Otgonbaatar C, Jeon PH, Ryu JK, Shim H, Jeon SH, Ko SM, Kim H. Coronary artery calcium quantification: comparison between filtered-back projection, hybrid iterative reconstruction, and deep learning reconstruction techniques. Acta Radiol 2023;64:2393-400. [PubMed]
13. Szczykutowicz TP, Toia GV, Dhanantwari A, Nett B. A Review of Deep Learning CT Reconstruction: Concepts, Limitations, and Promise in Clinical Practice. Curr Radiol Rep 2022;10:101-15.
14. Mileto A, Yu L, Revels JW, Kamel S, Shehata MA, Ibarra-Rovira JJ, Wong VK, Roman-Colon AM, Lee JM, Elsayes KM, Jensen CT. State-of-the-Art Deep Learning CT Reconstruction Algorithms in Abdominal Imaging. Radiographics 2024;44:e240095. [PubMed]
15. Shin DJ, Choi YH, Lee SB, Cho YJ, Lee S, Cheon JE. Low-iodine-dose computed tomography coupled with an artificial intelligence-based contrast-boosting technique in children: a retrospective study on comparison with conventional-iodine-dose computed tomography. Pediatr Radiol 2024;54:1315-24. [PubMed]
16. Noschang J, Tachibana A, da Silva Viviani LB, Silva MMA, de Almeida Mendes C, Rocha MA, Ishikawa WY, Wolosker N, Szarf G. New virtual contrast enhancement boost technique can reduce the dose of intravenous contrast medium used in computer tomography angiography of the abdominal aorta. Abdom Radiol (NY) 2025;50:2770-7. [PubMed]
17. Kang HJ, Lee JM, Ahn C, Bae JS, Han S, Kim SW, Yoon JH, Han JK. Low dose of contrast agent and low radiation liver computed tomography with deep-learning-based contrast boosting model in participants at high-risk for hepatocellular carcinoma: prospective, randomized, double-blind study. Eur Radiol 2023;33:3660-70. [PubMed]
18. Iizuka H, Yokota Y, Kidoh M, Oda S, Ikeda O, Tamura Y, Funama Y, Sakabe D, Nakaura T, Yamashita Y, Utsunomiya D. Contrast Enhancement Boost Technique at Aortic Computed Tomography Angiography: Added Value for the Evaluation of Type II Endoleaks After Endovascular Aortic Aneurysm Repair. Acad Radiol 2019;26:1435-40. [PubMed]
19. Hou J, Zhang Y, Yan J, Zhang T, Xia W, Zhu Y, Liu W. Clinical application of the contrast-enhancement boost technique in computed tomography angiography of the portal vein. Abdom Radiol (NY) 2023;48:806-15. [PubMed]
20. Otgonbaatar C, Ryu JK, Shim H, Jeon PH, Jeon SH, Kim JW, Ko SM, Kim H. A Novel Computed Tomography Image Reconstruction for Improving Visualization of Pulmonary Vasculature: Comparison Between Preprocessing and Postprocessing Images Using a Contrast Enhancement Boost Technique. J Comput Assist Tomogr 2022;46:729-34. [PubMed]
21. Yoon G, Ahn JH, Jeon Bs SH. Improving Image Quality and Visualization of Hepatocellular Carcinoma in Arterial Phase Imaging Using Contrast Enhancement-Boost Technique. J Comput Assist Tomogr 2025;49:348-57. [PubMed]
22. Yabe S, Sofue K, Hori M, Maebayashi T, Nishigaki M, Tsujita Y, Yamaguchi T, Ueshima E, Ueno Y, Murakami T. Added value of contrast enhancement boost images in routine multiphasic contrast-enhanced CT for the diagnosis of small (<20 mm) hypervascular hepatocellular carcinoma. Eur J Radiol 2023;160:110696. [PubMed]
23. Lim Y, Kim JS, Lee HJ, Lee JK, Lee HA, Park C. Image Quality and Lesion Detectability of Low-Concentration Iodine Contrast and Low Radiation Hepatic Multiphase CT Using a Deep-Learning-Based Contrast-Boosting Model in Chronic Liver Disease Patients. Diagnostics (Basel) 2024;14:2308. [PubMed]
24. Son W, Kim M, Hwang JY, Kim YW, Park C, Choo KS, Kim TU, Jang JY. Comparison of a Deep Learning-Based Reconstruction Algorithm with Filtered Back Projection and Iterative Reconstruction Algorithms for Pediatric Abdominopelvic CT. Korean J Radiol 2022;23:752-62. [PubMed]
25. Lee YJ, Hwang JY, Ryu H, Kim TU, Kim YW, Park JH, Choo KS, Nam KJ, Roh J. Image quality and diagnostic accuracy of reduced-dose computed tomography enterography with model-based iterative reconstruction in pediatric Crohn's disease patients. Sci Rep 2022;12:2147. [PubMed]
26. Zhang D, Mu C, Zhang X, Yan J, Xu M, Wang Y, Wang Y, Xue H, Chen Y, Jin Z. Image quality comparison of lower extremity CTA between CT routine reconstruction algorithms and deep learning reconstruction. BMC Med Imaging 2023;23:33. [PubMed]
27. Cademartiri F, Mollet NR, Runza G, Bruining N, Hamers R, Somers P, Knaapen M, Verheye S, Midiri M, Krestin GP, de Feyter PJ. Influence of intracoronary attenuation on coronary plaque measurements using multislice computed tomography: observations in an ex vivo model of coronary computed tomography angiography. Eur Radiol 2005;15:1426-31. [PubMed]
28. Becker CR, Vanzulli A, Fink C, de Faveri D, Fedeli S, Dore R, Biondetti P, Kuettner A, Krix M, Ascenti G. Multicenter comparison of high concentration contrast agent iomeprol-400 with iso-osmolar iodixanol-320: contrast enhancement and heart rate variation in coronary dual-source computed tomographic angiography. Invest Radiol 2011;46:457-64. [PubMed]
29. Nagamoto M, Gomi T, Terada H, Terada S, Kohda E. Evaluation of the acute adverse reaction of contrast medium with high and moderate iodine concentration in patients undergoing computed tomography. Radiat Med 2006;24:669-74. [PubMed]
30. Masui T, Katayama M, Kobayashi S, Sakahara H. Intravenous injection of high and medium concentrations of computed tomography contrast media and related heat sensation, local pain, and adverse reactions. J Comput Assist Tomogr 2005;29:704-8. [PubMed]
31. Cicin I, Erdogan B, Gulsen E, Uzunoglu S, Sut N, Turkmen E, Kodaz H, Ustundag S. Incidence of contrast-induced nephropathy in hospitalised patients with cancer. Eur Radiol 2014;24:184-90. [PubMed]
32. Ng CS, Kalva SP, Gunnarsson C, Ryan MP, Baker ER, Mehta RL. Risk of renal events following intravenous iodinated contrast material administration among inpatients admitted with cancer a retrospective hospital claims analysis. Cancer Imaging 2018;18:30. [PubMed]
33. Otgonbaatar C, Ryu JK, Jung WB, Jang SW, Hwang S, Kim T, Shim H, Seo JW. Utilization of double contrast-enhancement boost for lower-extremity CT angiography. J Med Imaging (Bellingham) 2024;11:067001. [PubMed]
34. Otgonbaatar C, Jeon SH, Cha SJ, Shim H, Kim JW, Ahn JH. Image quality and radiation dose of reduced-dose abdominopelvic computed tomography (CT) with silver filter and deep learning reconstruction. Sci Rep 2025;15:25757. [PubMed]