Effective Dose Measurements of the Latest-Generation Angiographic System in Patients with Acute Stroke: A Comparison with the Newest Multidetector CT Generation

DISCUSSION

Our study has the following main findings: 1) Collimation has a powerful impact on the estimated effective dose to the patient because it can reduce the dose by almost 50%, 2) the estimated effective dose to the patient of the collimated FDCT-P and of the portrait FDCT-A does not deviate substantially from analogous MDCT protocols, and 3) the eye lens dose appears to be similar on FDCT and MDCT.

Two prior publications estimating the effective dose to a patient of FDCT-P on predecessor systems (Artis Q and Artis Axiom; Siemens Healthineers AG) reported higher doses with 5.9 and 5.1 mSv, respectively.^6,10 However, these measurements must be compared with our uncollimated dose results (4.52 mSv) because no use of collimation was reported in either publication. In this case, a 12% difference in the measurements of Struffert et al¹³ is well within the margin of error; however, the 30% difference compared with our prior publication⁶ is outside this margin. This difference could partly be explained by a different phantom and fewer measurement points. Furthermore, the directly and indirectly irradiated tissues were not analyzed separately in our previous publication, possibly leading to an overestimation of the organ doses, especially red bone marrow.⁸ From a clinical standpoint, the collimated dose of FDCT-P is more important because the z-coverage can easily be reduced to parallel the parameters of MDCT-P.

Even more relevant is our result that the effective dose estimated for the collimated FDCT-P is only slightly higher than that with MDCT-P (33%), despite the considerably larger z-axis coverage area of 15 cm compared with the MDCT-P coverage of 11.4 cm (Fig 2 for reference). Nevertheless, MDCT-P for the triage of patients with late-window stroke has been validated in 2 randomized controlled trials,^11,12 while the technical equivalence of FDCT perfusion has yet to be established. A recent pilot trial of 13 patients showed promising results with high correlation for both ischemic core volume measurements on FDCT-P and MDCT-P and for follow-up infarct volumes.⁴ We presuppose that our current results will contribute to the effort to reproduce such clinically meaningful results in larger patient collectives.

While it is widely accepted that FDCT offers higher spatial resolution for high-contrast structures (eg, vessels and bones), an essential shortcoming of FDCT-A is the limited z-coverage area.¹³ This problem is largely solved by the development of a portrait FDCT-A protocol, which has a z-coverage area large enough to simultaneously visualize the circle of Willis and the intra- and extracranial carotid arteries down to their origins (Fig 4). This information is vital for planning interventions because the aortic arch configuration can influence the optimal vascular access site (eg, radial versus femoral) and for determining which catheters should be used for navigation to the intracranial vessels.^14,15 Another difference between FDCT-A and MDCT-A is that the timing of the acquisition after the injection of the contrast media bolus is operator-dependent. In contrast to MDCT-A, in which an automated Hounsfield unit threshold trigger is typically used for the start of the scan (by placing an ROI in the ascending aorta), the scan start has to be executed manually in FDCT-A.¹⁶ To address this difference, we developed a “bolus-watching” protocol in which DSA is used to monitor the visible influx of contrast media into the common carotid arteries (following a 10-second delay after intravenous contrast injection) to manually initiate the 3D angiography.¹⁶ According to previous measurements, this protocol adds only a minor radiation dose.⁶

• Download figure
• Open in new tab
• Download powerpoint

FIG 4.

Comparison of the coverage area of portrait FDCT-A (A) and landscape FDCT-A (B) (based on reconstructed coronal MIPs).

The overall dose for perfusion and angiography on FDCT is, at 3.79 mSv, only slightly higher (8%) compared with MDCT. This difference is well below the accuracy threshold for the effective dose for patient measurements and, therefore, could be neglected. Both protocols (perfusion and angiography) are often used, even in the early time window of thrombolysis because it is more commonly recognized that perfusion plays an important role in the detection of medium and distal vessel occlusions,^17,18 ie, as potential targets for mechanical thrombectomy.¹⁹ Recent measurements have shown an effective dose of 2 mSv for noncontrast FDCT parenchymal imaging, which is comparable with that of MDCT.^20,21 Because the cumulative dose of commonly used protocols in one-stop management of acute ischemic stroke does not differ substantially from routine MDCT, we anticipate that our findings will mitigate dose considerations in the triage decisions of patients with AIS.

The lower dose to the lens of the eye from FDCT-P compared with MDCT-P is explained by the reduced range of rotation of the flat detector, which is between 200° and 220° for most protocols with the radiation source being, importantly, below the patient.²² However, the eye lens doses for the perfusion protocols have to be interpreted with caution because we were not able to incline the head of the phantom. Flexing the head toward the chin, as is performed for MDCT-P at our institution, might reduce the direct irradiation of the lens and, therefore, might reduce the dose substantially.²³ With regard to the standard MDCT-A protocol, the effect of inclining the head should not have a relevant proportional impact due to the large z-coverage area. However, there is an alternative protocol (HeadAngio_Xcare; Siemens Healthineers AG) using an organ-based tube-current modulation. which can reduce the eye lens dose. In this protocol, the direct x-ray exposure to the eye lens is reduced by lowering the x-ray tube current for a certain range of projection angles when the eye lenses are facing toward the x-ray tube. This protocol was not evaluated in this article. Overall, the certainty for the magnitude of the reduction of the lens dose is low because we were not able to incorporate all influencing factors.

One major strength of our study is that we used an anthropomorphic phantom with an identical measurement setup for both systems. This feature allows a reproducible comparison between different x-ray imaging modalities, acquisition protocols, and studies. This approach is superior to other measurement approaches, such as simulations or CT dose index–based approaches because the modern C-arm devices typically use a 210° rotation compared with MDCT, which has a 360° rotation. In the case of FDCT, this difference leads to a nonuniform dose distribution with the peak dose occurring in the central plane, on the side of the phantom closest to the radiation source.²⁴ Furthermore, the larger z-coverage of FDCT compared with MDCT renders the traditional, weighted CT dose index approaches impractical.²² Another strength of our study is that we measured analogous protocols on both systems, enabling us to directly compare the effective dose to the patient from modern stroke imaging protocols.

However, our study has some limitations as well. The ATOM phantom is constructed to represent a broad cohort of different patients. Therefore, the actual dose to a patient might differ from the dose measured with the ATOM phantom. Because the ICRP 103 does not define the distribution and number of measurement points within the phantom, the investigator typically chooses these parameters.⁹ This situation can lead to differences in the estimated effective dose, depending on the number and position of measurement points within an anthropomorphic phantom. In addition, Roser et al²⁵ showed that the organ-equivalent dose values calculated from discrete measurements might underestimate the simulated organ dose that was calculated on the basis of a continuous dose distribution by up to 50%. In the clinical routine, variance to our phantom study could occur not only with the collimation of the x-ray field but also the with the ROI, through normal practitioner and patient differences. However, because these parameters can be largely standardized in stroke protocols and because the coverage areas of the FDCT were at least as large as on the MDCT, our collimated results should be generalizable to clinical routine.