Feasibility of a pulsed multiphase contrast media injection protocol in head and neck computed tomography angiography: a systematic retrospective study

Research object

We retrieved data on 522 patients who underwent head and neck CTA at our hospital from March 1, 2022, to March 31, 2024, and excluded 174 patients who did not meet the established criteria. All enrolled patients underwent only head and neck CTA examinations. All data were uniformly retrieved and recorded between April 10, 2024, and May 31, 2024. Inclusion criteria: (1) Age >18 and ≤80 years; (2) patients without a history of carotid artery and its branch surgeries, as well as without a history of head and neck cancers or benign tumors. Exclusion criteria: (1) Patients with severe cardiac, hepatic, or renal insufficiency; (2) pregnant or lactating women; (3) patients with a history of iodine allergy; (4) patients with metal implants, where severe metal artifacts are present in the images. Following examination and approval, the institution designated the standardized operation plan on March 25, 2023 and decided to implement the pulsed-multiphase contrast agent injection protocol for all subsequent head and neck CTA cases, with patient grouping being time-limited. The patients at our hospital from March 1, 2022, to March 25, 2023, were assigned to the conventional group, all having utilized the two-phase contrast agent injection protocol. Patients at our hospital from March 26, 2023, to March 31, 2024, who all received the pulsed-multiphase contrast agent injection protocol, were assigned to the Pulse Group. Before March 25, 2023, patients undergoing head and neck CTA examinations at our hospital were administered a conventional two-phase contrast agent injection protocol. After the policy was established on March 25, 2023, all patients were switched to a pulsed multi-phase contrast agent injection protocol. Patients were grouped based on the time periods they visited our hospital for examinations, as different injection protocols were applied during different time periods.

The study received ethical approval (Lenko Review (2023-01) No. 1) from the Ethics Committee of the Second Affiliated Hospital of Qiqihar Medical University (No. 37, West Zhonghua Road, Jianhua District, Qiqihar, China) on March 9, 2023. This study adhered to the Declaration of Helsinki. Since this was a retrospective analysis, the Ethics Committee permitted the waiver of informed consent requirement; nonetheless, it was emphasized that the confidentiality of patients’ personal information should be maintained. The specific study flow chart is shown in Fig. 1.

Image acquisition

All patients were scanned using the Revolution™ computed tomography (CT) scanner (GE Healthcare Life Sciences, Waukesha, WI, USA). The scanner is calibrated every morning before it is used. Recalibration is performed in the event of special circumstances, such as system exit. And all imaging data was retained for at least five years. Head and neck CTA imaging protocols: The patient was placed in the supine position. The temporal subtraction method was employed. The initial scan range is as follows: Scanning extends from the aortic arch (AA) to the cranial vertex. Initiating the scan requires a 5-second delay prior to commencement. The parameters for the CTA scanning are as follows: The tube voltage was set to 120 kV for the conventional group and 100 kV for the pulse group in CTA scanning. The tube currents for both groups were configured for automatic adjustment, with values ranging from 150 to 400 mA. The collimator width was 64 × 0.6 mm, with a pitch of 10.992:1 and a tube rotation speed of 0.5 s/rot. All other scanning parameters and contrast media were consistent across both groups. Both patient groups underwent head and neck CTA scans after applying the test-bolus technique. The drug administered was Iopromide (Ultravist 370, 370 mg I/mL; Bayer Schering Pharma) at a volume of 15 mL and an injection speed of four mL/s. A 15 mL saline flush was subsequently administered. A 15-mL saline flush was also used. Scanning of the same layer commenced after a 10-second delay. Monitoring at the bifurcation of the common carotid artery, which is situated at the level of the C4/C5 intervertebral disc, was performed using a region of interest (ROI), with scans every 2 s. Scanning was halted after the peak contrast concentration of the layer was reached, and a time-density curve was then plotted. ROI refers to the area of the target vessel delineated by a circle for treatment or CT value measurement purposes. The peak value of the time-density curve corresponds to the CT value of the peak iodine concentration within the ROI. Additionally, the time-axis value at the peak corresponds to the time point (t_p) when peak iodine concentration occurs. The arterial-phase scanning delay time (T) is calculated using the formula: T = t_p + 2 − plain scanning time. Contrast injection was performed concurrently with the scan. A plain scan without iodine-containing contrast from the AA to the cranial vertex was performed first (scan duration: 1.5−2.5 s), followed by an enhancement scan (scan duration: 1.5−2.5 s), with a delay interval of T seconds between the two scans. All scans were performed without breath-holding.

The contrast agent was injected using a high-pressure double-barreled injector (Urich INJECT CT Motion™; Ulrich Medical, Ulm, Germany). Before the injection, a 24-gauge indwelling needle was inserted into the superficial vein or the median cubital vein of the patient’s right forearm. Injecting 20 mL of saline in advance assessed that the catheter was unobstructed. The infusion rate was set at five mL/s. In the contrast injection program, the conventional group was administered the two-phase injection protocol. The first phase involved the injection of 60 mL of contrast, followed by a second phase that injected 40 mL of saline. The flow rates for the contrast agent and the normal saline were both five ml/s. The total injection time was 20 s. Conversely, the pulse group was administered a pulsed multi-phase contrast injection protocol. In this group, the first phase injected 35 mL of contrast, followed by the second phase with five mL of saline. From the third to the fifth phase, the contrast agent and saline were alternately injected to achieve mixed injection of them. The injection volume of the contrast agent or saline was five mL per phase. Inject 40 mL of saline solution at the final stage. The flow rate was five mL/s, and the total injection duration was 19 s. The contrast agent or saline was injected continuously at each stage. In the first phase, a pure contrast agent was injected to ensure that the target blood vessels are clearly visible. The total injection time in the first stage was 7 s. From phase two to phase five, a mixture of contrast agent and saline was injected to maintain vascular contrast and clear residual contrast agent from the brachiocephalic veins. The injection time at this stage is 4 s. Finally, 35 ml of saline was injected to eliminate any remaining contrast agent more thoroughly. The injection time for the final stage was 8 s. Before conducting this study, we validated the feasibility of the low-kilovoltage, low-contrast-dose scanning protocol using animal models. New Zealand rabbits were randomly divided into three groups: Group A was scanned at 120 kV with a contrast agent dose of 2.5 ml/kg, Group B at 100 kV with a contrast agent dose of 2.5 ml/kg, and Group C at 100 kV with an 80% reduced contrast agent dose (2.5 ml/kg × 80%). The results of the experiment demonstrated that the low contrast agent injection protocol effectively reduced the CT values of LSVC under low-kilovoltage scanning conditions, while producing images of comparable quality to those obtained in Group A, which used higher kV and contrast agent doses (Li et al., 2022). Before modifying the injection protocol, we used a two-phase bolus injection technique. This often resulted in excessive beam-hardening artefacts in the subclavian vein or superior vena cava of patients, which adversely affected image quality. Based on preliminary experimental studies conducted on animals, we redesigned a pulsed, multi-phase contrast agent injection protocol aimed at reducing the occurrence of beam-hardening artefacts. Comparisons of the injection regimens and the parameters of the patients in conventional group and pulse group are detailed in Table 1 and Fig. 2.

Image post-processing and data analysis

CTA data was reconstructed using the GE Healthcare’s Advantage Workstation. The three-dimensional volume data was reconstructed using techniques such as curved planar reformation (CPR), maximum intensity projection (MIP), and volume rendering (VR). The reconstruction parameters were as follows: the slice thickness for all images was set to 0.625 mm, with a slice interval of 0.625 mm. The window width (WW) was set to 600 HU, and the window level (WL) was set to 100 HU.

Using the double-blind method, two experienced radiologists independently measured the CT values of the AA, the bilateral common carotid arteries (CCA) at the level of the bifurcation of the CCA, the bilateral vertebral arteries (VA) also at the level of the bifurcation of the CCA, the C2 segment of the internal carotid artery (C2), the M1 segment of the middle cerebral artery (M1), the subclavian vein (SCV), and the superior vena cava (SVC). The aforementioned vessels’ CT values were measured independently on the Picture Archiving and Communication System workstation. Use the circular measurement tool in the workstation to measure the CT values of the aforementioned blood vessels, one by one. When measuring CT values, the vascular walls and small calcifications should be excluded. The CT values of the target vessels were recorded, and the average value was used as the mean CT value of the vessel. If one side of an artery was occluded and could not be measured, the value from the contralateral side was used as a substitute. The mean vessel CT values were determined based on the CT values of the AA, CCA, VA, M1 and C2. The SNR was equal to the ratio of the mean CT value of blood vessels between background noise. The CNR was equal to the ratio of mean CT value of blood vessels—background CT value between background noise. The standard deviation of the CT values measured in the trachea at the level of the first cervical vertebra was regarded as background noise, whereas the CT value of the soft tissue positioned behind the trachea at that same level was considered the background CT value. Subjective scoring was conducted by two radiologists (physician A with 15 years of experience and physician B with 35 years of experience in head and neck CTA). Two physicians scored the imaging data in a double-blind manner, and the image sequences of all patients were randomly presented to the two physicians. Physicians were instructed to score the images based on their personal preferences. Referring to the scoring criteria in the study by Tawfik et al. (2012), we have re-established a similar scoring system consisting of three components. These are residual contrast in the brachiocephalic vein ((5) No residual contrast in the vein and the surrounding tissues are clearly displayed; (4) minimal residual contrast in the vein that does not increase vascular contrast; (3) moderate residual contrast in the vein that is only visible within the vessel; (2) residual contrast in the vein that affects the surrounding tissues, but does not impair diagnostic imaging; and (1) residual contrast in the vein that severely affects the surrounding tissues and bone tissues are not properly displayed) and vascular sharpness ((5) Perfect vascular contour boundaries; (4) clear structural contour boundaries; (3) moderate structural contour boundaries; (2) blurred contours; (1) significantly blurred structural contours; and subjective image noise or graininess ((5) no perceptible image noise; (4) Below-average image noise; (3) average image noise and slight graininess; (2) Above-average image noise and moderate graininess; (1) significantly unacceptable noise level). The final score is presented according to Schimeller L’s five-point system.The scoring criteria were based on the 5-point scale utilized in the study by Schimmöller et al. (2013), as presented in Table 2.

The evaluation of the CT radiation dose level was based on two metrics: the dose length product (DLP, measured in mGy cm) and the effective dose (ED, measured in mSv). The DLP is the overall radiation dose associated with a medical examination and was calculated using the formula DLP = scan length× volume CT dose index (CTDIvol). After the examination, the DLP can be directly obtained from the CT control panel. The ED is determined by the equation ED = DLP× k, with k representing the anatomic region-specific conversion coefficient. In the case of CTA scanning for the head and neck, k is set at 0.0023 mSv mGy⁻¹ cm⁻¹ (Schmid, Uder & Lell, 2017).

Statistical analysis

We analyzed the data using SPSS 26.0 (IBM Corp., Armonk, NY, USA). Normality tests and chi-square tests were conducted on general patient data, measured CT values, SNR, CNR, and radiation doses. For data exhibiting a normal distribution, the results were expressed as the mean ± standard deviation (SD). In contrast, data that did not conform to a normal distribution were reported differently; specifically, these results were presented in terms of the median (interquartile range (IQR)). This method of presentation is often represented as (M (P25, P75)). Categorical data were expressed as frequencies and proportions. When comparing the measurement data of two groups of samples, the independent two-sample t-test was employed if the data met the criteria for normal distribution and homogeneity of variance. Conversely, when the data did not conform to a normal distribution, it became necessary to employ the Mann–Whitney U test as a non-parametric alternative. Categorical data was tested with the Pearson chi-square (χ²) test. We set the significance level at α = 0.05. A statistically significant difference was declared when the calculated P-value fell below this threshold. An analysis of the disparity in subjective image quality scores between the conventional group and the pulse group was conducted using the Mann–Whitney U test. To assess the agreement of the two physicians’ evaluations of image quality, we used Kappa coefficient. A Kappa value below 0.4 denotes poor agreement, values ranging from 0.4 to 0.75 reflect good agreement, while a value exceeding 0.75 signifies very good agreement.

The general information of patients

A total of 348 participants were included in this study, with 179 participants assigned to the conventional group and 169 to the pulse group. No statistically significant differences were found in concerning gender, age, height, weight, and BMI when we compared the two groups. The chi-square test was employed in the analysis of gender differences (χ² = 0.063, P = 0.802). The Mann–Whitney U test was employed for analyzing differences in age, height, weight, and BMI (Z = −0.481, Z= −0.808, Z= −0.682, Z = −0.428; P = 0.631, P = 0.419, P = 0.495, P = 0.669). General patient information is summarized in Table 3 and Fig. 3.

CT Value

Collect all CT values of target vessels measured by the circular measuring tool in the above workstation. The results showed that the CT values were all above 300 HU. The comparison results of the target vessel CT values between conventional group and pulse group are shown in Table 4, Figs. 4, and 5. The CT attenuation values between the AA, M1, C2, VA, and CCA for the two groups showed no significant differences between groups (P > 0.05). The differences in CT values between the SCV and the SVC across both groups were statistically significant (Z = −7.951, P < 0.001; Z = −5.539, P < 0.001). The pulse group exhibited lower CT values for the SCV and SVC compared to the conventional group, as shown in Figs. 6 and 7.

SNR and CNR

The CNR and SNR of the AA, M1, C2, VA, and CCA were compared between the conventional group and the pulse group. The differences were not statistically significant (P > 0.05), as shown in Table 5 and Fig. 8.

Scoring criteria of subjective evaluation of imaging reconstruction

Two physicians evaluated the image quality of both patient groups, assigning scores ≥ 3 points. Physician A’s subjective ratings for the conventional group and the pulse group were 3.34 ± 0.54 and 3.76 ± 0.78, respectively. Physician B’s subjective ratings for the conventional and pulse groups were 3.23 ± 0.55 and 3.91 ± 0.79, respectively. Physician A’s subjective evaluation results indicate that in the conventional group, 124 cases (69.2%) were rated as three points, 49 cases (27.4%) as four points, and six cases (3.4%) as five points. In the pulse group, 76 cases (45.0%) received a rating of three points, 57 cases (33.7%) as four points, and 36 cases (21.3%) as five points. Physician B’s subjective evaluation results indicate that in the conventional group, 149 cases (83.3%) were rated as three points, 19 cases (10.6%) as four points, and 11 cases (6.1%) as five points. In the pulse group, 61 cases (36.1%) received a rating of three points, 62 cases (36.7%) received four points, and 46 cases (27.2%) received five points. The image quality of the pulse group was higher than that of the conventional group, with statistically significant differences (Physician A: P < 0.001; Physician B: P < 0.001). The two physicians demonstrated a high level of consistency in their ratings of image quality for the Conventional group and the Pulse group, with a kappa value of 0.573 for the Conventional group and 0.684 for the Pulse group, as shown in Table 6 and Fig. 9.

Comparison of radiation dose

The differences in DLP and ED between the conventional group and the pulse group were statistically significant (Z = −11.966, P < 0.001; Z = −11.966, P < 0.001). Compared with the conventional group, the DLP and ED in the pulse group decreased by approximately 9%. Refer to Table 7 and Fig. 10 for further details.