This study prospectively included 60 patients with suspected coronary artery stenosis from March 2024 to September 2024. The Ethics Committee of our hospital approved this prospective study (2024-H-003), and all participants provided written informed consent prior to undergoing a CT scan. Inclusion criteria were as follows: (a) no history of contrast agent allergy; (b) normal renal function (glomerular filtration rate $\ge$80 mL/min); (c) no prior history of cardiac surgery. Exclusion criteria were as follows: (a) incomplete clinical data; (b) respiratory and motion artifacts resulting in insufficient images for reconstruction; (c) scan interruption due to procedural discomfort. The participant flowchart is depicted in Fig. 1. Clinical characteristics were recorded, including patient age, gender, heart rate during examination, weight, height, and calculated BMI. Of 60 enrolled patients, 14 were excluded: 8 due to motion artifacts, 4 with incomplete clinical data, and 2 for scan interruptions. The final analysis included 46 patients (Group A: 25, Group B/C: 21).

Fig. 1.

Patient enrollment workflow. CT, computed tomography.

All CT angiography examinations were performed on a 320-row CT scanner (Aquilion ONE Genesis, Canon Medical System, Otawara-shi, Japan). The CT scanner underwent daily quality assurance tests using air calibration. Tube current modulation and detector sensitivity were recalibrated semi-annually using a water phantom. Water phantom calibration ensured the measurement of CT number accuracy, noise, and uniformity. Participants were positioned supine with their heads oriented forward, with their hands crossed to support their heads, and participated in breathing training. Scans were performed after inhaling and holding their breath as instructed. The scanning range extended from the bifurcation of the trachea to the diaphragmatic surface of the heart, using cardiac volume scanning with a spiral time of 0.275 seconds and a collimator width of 160 mm. All patients were divided into Group A and Group B according to different acquisition protocols.

Group A: Tube voltage: 120 kV, automatic milliampere-second technology. The contrast agent employed was a concentration of 370 mgI/ mL (iodoprolol injection, Bayer Healthcare, Berlin, Germany), injected via a high-pressure injector using the Bayer MEDRAD Stellant CT Injection System (Stellant D-CE, Pittsburgh, PA, USA) through the median cubital vein. Contrast agent was delivered at 6 mL/s, with a dosage of 0.9 mL/kg [17, 18]. After the contrast agent injection was completed, 30 mL of physiological saline was flushed at the same rate. The CCTA scan was triggered once the detection threshold of the descending aorta reached 150 Hounsfield Unit (HU).

Group B: Tube voltage: 100 kV. Contrast agent was delivered at 2 mL/s, with a dosage of 0.3 mL/kg. The rest of the scanning parameters were consistent with those of Group A.

Images from both Group A and Group B were reconstructed using hybrid iterative reconstruction (AIDR 3D, FC43 kernel, Canon Medical Systems, Otawara-shi, Japan) for all transverse CT images, with a slice thickness of 0.5 mm and intervals of 0.25 mm. The original data for Group C images were derived from Group B. Group C images were generated from the raw data of Group B: the raw data were reconstructed using deep learning reconstruction (AiCE, Cardiac Standard, Canon Medical Systems, Otawara-shi, Japan) to obtain the CT images, with a slice thickness of 0.5 mm and intervals of 0.25 mm. Further post-processing was performed using CE-Boost (SURESubtraction Iodine Mapping, Canon Medical Systems, Otawara-shi, Japan) to generate the final Group C images.

To estimate radiation dose exposure, we recorded the dose length product (DLP) and volume CT dose index (CTDI_vol) for all participants. According to international recommendations, the effective dose (ED) was calculated by multiplying the DLP by an adult chest-specific conversion factor (0.014 mSv Gy⁻¹$\cdot$cm⁻¹) [19].

The image quality of the three sets of images was evaluated by two radiologists with 8 and 6 years (CS and WH) of experience in coronary artery computed tomography angiography (CTA) diagnosis, respectively. Both radiologists were blinded to the patients’ clinical information and image processing methods. They used a 5-point scoring system to evaluate the clarity, contrast, fine structure, and image noise of the original and post-processed images. A visual reference for the scoring criteria is provided in Supplementary Fig. 1, The scoring criteria were as follows [15]:

5 points: Excellent image contrast, no obvious noise, clear display of fine structures with sharp boundaries, and clear visualization of the entire process and terminal segments of vascular reconstruction.

4 points: Good image contrast, increased noise, clear display of fine structures with well-defined boundaries, and fewer peripheral branches throughout vascular reconstruction.

3 points: Fair image contrast, moderate noise, clear vascular reconstruction, and slightly unclear display of small anatomical structures.

2 points: Poor image contrast, significant noise, unclear display of small anatomical structures, with difficulty in identification, and unclear visualization of vascular reconstruction.

1 point: Very poor image contrast, excessive noise, inability to distinguish small anatomical structures, and unclear display of vascular reconstruction.

An image score of $\ge$3 was considered to meet clinical diagnostic requirements, while a score of $\le$2 was considered not to meet these requirements.

Two experienced radiologists (MC and YW), with 15 and 9 years of experience in coronary artery CTA diagnosis, respectively, evaluated the image quality of three types of images without knowing the patients’ clinical information or image processing methods.

Regions of interest (ROI) were delineated in the left main coronary artery (LM), left anterior descending coronary artery (LAD), left circumflex coronary artery (LCX), and the tertiary branches of the right coronary artery (proximal right coronary artery (RCA-1), middle right coronary artery (RCA-2), and distal right coronary artery (RCA-3)). The ROI was placed in the center of the blood vessel, avoiding the vessel wall, plaque, calcification, and stents. The ROI area for RCA-1 and LM was about 15 mm², RCA-2 and LAD about 10 mm², and RCA-3 and LCX about 5 mm². An ROI with an area of 50 mm² was selected in the right erector spinae muscle (RES), and the resulting standard deviation (SD) was considered the background SD.

CT values and noise SD of these blood vessels were recorded, and the signal-to-noise ratio (SNR) (1) and contrast-to-noise ratio (CNR) (2) were calculated [20]. The calculation formulas were as follows:

(1)$$SNR=\frac{H{U}_{CA}}{SD}$$

(2)$$CNR=\frac{H{U}_{CA}-H{U}_{RES}}{\sqrt{\frac{S{D}_{CA}}{S{D}_{RES}}}}$$

In the above formulas, HU_C⁢A represents the signal intensity of the coronary artery, HU_R⁢E⁢S represents the signal intensity of the erector spinae muscle, SD_C⁢A represents the standard deviation of the coronary artery, SD represents local noise, and SD_R⁢E⁢S represents the standard deviation of the erector spinae muscle.

We used SPSS 26.0 (IBM Corporation, Armonk, NY, USA) and R 4.2.1 (The R Foundation for Statistical Computing, Vienna, Austria) for data analysis and processing and performed a normality test on the obtained data. Data conforming to a normal distribution were reported as mean $\pm$ SD ($\overline{x}$ $\pm$ s). Independent samples t-tests were used for comparisons between two groups, while one-way ANOVA was used for comparisons among multiple groups. Data not following a normal distribution were represented by the median and interquartile range. The Mann-Whitney U test was used for comparisons between two groups, and the Bonferroni correction was used for comparisons among multiple groups. Categorical variables were compared using the chi-square test. A p value of 0.05 was considered statistically significant. The consistency of CT values, noise measurements, and subjective scores was analyzed using intraclass correlation coefficients (ICCs), which were classified as poor (ICC 0.50), moderate (ICC 0.50–0.74), good (ICC 0.75–0.89), or excellent (ICC $>$0.90). A priori power analysis (G*Power 3.1, Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany) indicated 17 patients per group were required ($\alpha$ = 0.05, power = 0.8, effect size d = 1.2). Our study included 21 participants in Group B/C, exceeding this threshold. Intergroup comparisons were performed using one-way ANOVA with Bonferroni correction for multiple comparisons. All statistical tests reported adjusted p-values (p_adj), mean differences with 95% confidence intervals, and effect sizes calculated via Cohen’s d method.

Additionally, we performed age and BMI adjusted statistical analyses to account for potential confounders. The results of these analyses, including the adjusted p-values for the comparisons between Groups A and B and between Groups A and C. For the comparison between Groups B and C, no adjustment was necessary as they belong to the same patient cohort.

This study included a total of 46 participants (29 males and 17 females). Group A consisted of 25 participants (19 males and 6 females) with an average age of 57.76 $\pm$ 10.62 years. Group B consisted of 21 participants (10 males and 11 females) with an average age of 62.05 $\pm$ 11.52 years.

We compared the clinical data and radiation exposure of the patients and found that the contrast agent dosage in Group A was significantly higher than those in Group B. However, there was no statistically significant difference between Group A and Group B in terms of age, gender, heart rate during examination, height, weight, BMI, history of smoking, hyperlipidemia, diabetes, family history of coronary heart disease and hypertension (all p $>$ 0.05). Detailed data can be found in Table 1, Fig. 2A–C.

Fig. 2.

Comparison of patient demographics, contrast agent dosage, and radiation dose metrics between Group A (conventional protocol) and Group B (triple-low protocol). (A,B) Comparison of general patient data statistics and dosage of contrast agents. (C) Comparison of patient radiation dose metrics. BMI, body mass index; CTDI_vol, CT dose index; DLP, dose length product; ED, effective dose.