Intraoperative CBCT imaging in endovascular abdomen aneurysm repair – Optimization of exposure parameters using a stent phantom

Cone beam computed tomography (CBCT) may provide essential additional image guidance to endovascular abdominal aneurysm repair (EVAR) operations but also significant radiation exposure to patients if scans are not carefully optimized. The purpose of our study was to define the image quality requirements for intraoperative EVAR CBCT imaging and to optimize the CBCT exposure parameters accordingly. A Multi-Energy CT phantom simulating a large patient was used by replacing the central phantom cylinder with a custom water-filled insert including an EVAR stent. Different exposure parameters covering a range of radiation qualities and dose levels were used to define the optimal image quality level regarding stent graft evaluation (compressed, bent, or collapsed). The radiation dose was measured with a calibrated air kerma-area product (KAP) meter and organ doses were calculated based on Monte Carlo simulations and a mathematical patient model. Based on the results, updated exposure parameters with the highest mean energy and lowest dose level available were recommended. With the updated protocol, the radiation exposure could be significantly decreased. The KAP value decreased from 9720 μ Gy ⋅ m 2 to 440 μ Gy ⋅ m 2 and reference point air kerma from 351 mGy to 16 mGy (a reduction of 96%) and organ doses of the organs in the irradiated region decreased on an average 91%. The new protocol resulted in acceptable clinical image quality based on testing with clinical cases.


Introduction
New technologies enable intraoperative volumetric imaging during interventional procedures such as endovascular abdominal aneurysm repair (EVAR).C-arm angiography systems have commonly been used to guide the operation and perform intra-operative clinical quality control at the end of the procedure.However, some complications may be missed when conventional single-projection views are used.Multi-slice computed tomography (MSCT) is routinely used for the postoperative follow-up.Modern C-arm systems offer an additional possibility to acquire 3D images with cone beam computed tomography (CBCT) scan mode, which can identify some clinically significant anomalies.The essential strength is the availability of 3D scan when the patient is still in the operating room and potential complications needing reinterventions can be corrected immediately.
Based on the EVAR-1 study, the risk of complications and reinterventions are 3-4 times higher in endovascular operations compared with open surgery [1].In the second trial, EVAR-2 it was concluded that every second patient had a complication related to an endovascularly installed endograft and every third had a reintervention within the first six years [2].Some of these complications could be avoided if proper clinical quality control could be carried out directly at the end of the first operation.Biasi et al. studied the use of CBCT in EVAR operations, and based on their study, the reinterventions decreased >6% in the following 30 days when CBCT was used for the post-operative quality control with contrast media [3].Törnqvist et al. concluded that intraoperative technical problems such as kinks and stent graft compression are better detected with CBCT than with conventional angiography [4].Tenorio et al concluded that CBCT reliably detected positive findings prompting immediate revisions in nearly one of five patients [5].
In our facility, in average around 140 EVAR operations are performed per year in two hybrid operating rooms with floor-mounted Carm angiography systems.MSCT is routinely performed postoperatively after 1 -3 months and one year.After that, the follow-up is continued with yearly ultrasound imaging.During the follow-up, there is a need for further action approximately in 20% of cases.Both C-arm systems have an option to perform CBCT imaging, but this had not been commonly used for EVAR operations due to the high radiation doses.Based on a local dose study of 71 EVAR operations, the average total kerma-area product (KAP) in the operation was 3300 µGy⋅m 2 without the CBCT [personal communication by T.Kaasalainen].Moreover, based on a few clinical cases where CBCT had been used, the CBCT related KAP ranged from around 4000 µGy⋅m 2 to 7000 µGy⋅m 2 .Thus, reduction and optimization of such a dominant contribution from CBCT to lower levels of radiation exposure is clearly motivated.
There are several studies concluding similar high dose levels for this type of CBCT examination.Bruschi et al. concluded that the median KAP for CBCT performed at the operating theatre was 6700 (variation from 4900 up to 7300) µGy⋅m 2 , which was comparable to the values of the operation excluding the CBCT (6800 µGy⋅m 2 ) [6].Törnqvist et al reported mean total KAP for CBCT of 7065 (from 3491 even up to 12 645) µGy⋅m 2 [4].Eide et al compared DynaCT (specific name for CBCT of Siemens systems) with multidetector CT and reported an average KAP of 3027 (variation from 2308 up to 3668) µGy⋅m2 [7].In another study, they used DynaCT for pre-treatment evaluation of aortic aneurysm and reported a mean DAP of 7549 µGy⋅m 2 (min: 2120, max: 9135, SD: 1625) [8].Tenorio et al reported average DAP of 5300 µGy⋅m 2 [5].
Generally, the patient radiation exposure in CBCT is much higher than in projection imaging, and consequently, the clinical need should be justified.A key benefit of CBCT image data in EVAR application is the axial view, which allows to check that the stent graft is not compressed or kinked.Since the stent is made of metal alloy and there are not strong clinical requirements for soft tissue contrast, there is no need for contrast media and the image quality aim is very different from other clinical cases where the focus is more on the tissue structures.Thus, this welldefined clinical need enables indication specific optimization, which should be the aim, but is not common in interventional radiology.Accordingly, the aim of this study was to define the image quality requirements for intraoperative (post-operational) EVAR CBCT imaging and to optimize the CBCT exposure parameters to lower the radiation exposure to patients.

X-ray system
A Siemens Axiom Artis Pheno C-arm (Siemens Healthineers, Erlangen, Germany) used for the study is mounted in a hybrid operation room with a robotic floor stand.It contains a Gigalix X-ray tube (125/30/40/ 90-5) and a zen40HDR flat panel detector with crystalline silicon and CsI-scintillator.The system allows the reconstruction of 3D images based on a rotational CBCT acquisition.The main technical parameters of the system are given in Table 1.KAP meter, KermaX plus (IBA Dosimetry, Schwarzenbruck, Germany) is 0.2 mmAl equivalent at 70 kV and it is a fixed part of X-ray tube housing, which in total corresponds to a half-value layer (HVL) of around 2.5 mmAl.In addition, the imaging setup contains a carbon fiber patient table (corresponding to less than 1.4 mmAl attenuation at 100 kV and HVL of 3.6 mmAl) with an 8 cm thick mattress (corresponding to less than 1 mmAl additional attenuation).
In CBCT (DynaCT) mode, the focus-detector distance (FDD) is fixed to 130 cm and isocenter to 78.5 cm (Fig. 1).The 3D acquisition field of view is in landscape orientation with 23.5 cm diameter and 17.5 cm height.The rotational exposure starts from 10 • angle from the AP direction and stops at 210 • angle covering total 200 • .

Exposure parameters
For this type of abdominal EVAR examination, there were two protocols available for the CBCT study: 5 s DCT Body and 4 s DCT Body Care both originally without copper filtration.Other related exposure parameters of these protocols are given in Table 2.The CBCT option was not routinely used in EVAR operations, but there were a few clinical cases and, in those cases, both protocols had been used.However, the 4 s DCT Body Care with lower dose level was considered more appropriate and therefore, this was selected for further testing with different radiation qualities (tube voltages and filtrations) and varying dose levels.In addition, both original protocols were tested with all available additional copper filter thicknesses varying from zero to 0.9 mmCu.The tested parameter range is given in Table 2.
The half-value layer (HVL) values were used as a radiation quality specifier and calculated based on the exposure parameters using a computer programme SpektriPaja v3.0 [9], which is based on the theory of Birch and Marshall [10].The actual anode angle of 11.5 • was used and the basic filtration was set to 3 mmAl to simulate the fixed filtration and other items (KAP meter, couch, mattress) in the beam.Since the main interest is in the copper-filtered beams, the actual amount of aluminium has a minor effect on the spectra.

Image quality phantoms
A water equivalent Multi-Energy CT phantom (Sun Nuclear, Melbourne, Australia) with diameters of 30 cm (anterior-posterior) and 40 cm (lateral) was modified for the study.The cylindrical insert in the center of the phantom was replaced by a tailored insert with an EVAR stent (GORE® Excluder C3®, W.L. Gore & Associates, Inc., Flagstaff, AZ, USA) inside of a condom filled with water and closed tightly (EVARphantom, Fig. 2).The phantom positioning was performed so that the stent was in the center of the CBCT scan field-of-view (FOV).In addition, a CIRS Atom 702-D Adult female anthropomorphic phantom (CIRS, Norfolk, VA, USA) was used separately to validate the image quality of anatomical structures with the updated exposure parameters.

Dosimetry equipment and calibration
The X-ray system contains a physical KAP meter with an ionization chamber which is mounted to the X-ray tube housing.The KAP meter integrates the air kerma values of complete radiation field and examination.The KAP meter was tested as a part of regular maintenance and the results were acceptable.However, since some KAP meters have a strong energy dependence, a proper calibration is especially important when the radiation quality is adjusted and optimized as in this study [11].Therefore, the KAP meter was calibrated using a beam area method as described in previous work [12] following the international Code of Practice [13].A reference dose meter Radcal Accu-Gold + 10x6-6 ionization chamber (Radcal, Monrovia, USA) calibrated at a national radiation metrology laboratory (STUK -Radiation and Nuclear Safety Authority, Finland) and traceable to PTB (Physikalisch-Technische Bundesanstalt, Germany) was used to cover different radiation qualities.The fixed field size was used, and it was measured using the known size of the dose meter and using the acquired image to calculate the field size at the dose meter location.Based on the results, for the over-couch position, the calibration coefficients variated between 1.05 and 1.09 in the range of studied exposure parameters.For maximal accuracy, radiation quality specific calibration coefficients were used to correct the displayed KAP and reference air kerma values.The attenuation of the couch impacts on the calibration coefficients but it was not considered in the further calculations since the irradiation geometry was always the same, so the impact on relative values was considered small and it was included in the uncertainty budget.

Radiation doses
The KAP and reference point air kerma values are part of the exposure information provided by the X-ray system.The displayed values were corrected based on the calibration data.However, the KAP value gives only an indication of the intensity of the radiation beam incident to the patient surface.The actual patient exposure and related organ doses are radiation quality dependent and therefore, specific conversion coefficients from incident air kerma were calculated based on simulations performed with a Monte Carlo (MC) based programme PCXMC Rotation Dose Calculations, Version 2.0.1.5,STUK [14].The main organs of interest were colon, gall bladder, kidneys, pancreas, small intestine, spleen and stomach.The effective dose was also calculated based on the organ doses.In this study, we use the term radiation dose as a generic term for all different dose quantities.
The PCXMC programme uses mathematical hermaphrodite phantom models, and for this study the standard size (height 178.6 cm and weight 73.2 kg) was used.With this selection, the PCXMC phantom width is 40 cm being the same as with the EVAR phantom.The thickness of the PCXMC phantom is 20 cm, which is smaller than the 30 cm thickness of the EVAR phantom.Therefore, the calculated absolute values do not fully represent the values for EVAR-phantom -sized patients.However, the conversion factors can be used for the optimization since the exposure geometry is the same, and the comparisons are done using relative values between different radiation qualities.The focus-reference point distance is used in PCXMC to indicate where the input dose is given and that was set at the isocenter 78.5 cm and the parameters for the field positions X ref and Y ref in the PCXMC programme were set to zero to be in the center of the phantom.The cranio-caudal location Z ref of 26 cm was selected to represent the correct anatomical region of a clinical EVARstudy.The beam width was 24 cm and height 18 cm.20 000 photons were used for the simulations.The spectral simulation and dose simulation used equivalent exposure parameters.The PCXMC with a rotation function was used to calculate the conversion factors from different angles.The simulations were performed in steps of 10 • and an average was used for the examination.The impact of the patient couch was not considered.

Table 2
The CBCT exposure parameters provided by the vendor pre-defined protocols.P. Toroi et al.

Uncertainties
All the estimated uncertainties are reported as expanded total relative uncertainties, obtained by multiplying the combined relative standard uncertainty with the coverage factor k = 2 corresponding to confidence level of 95% if not otherwise stated [15].The total uncertainty of absolute KAP values for the radiation beam incident to patient surface is large > 20% mainly due to the uncertainties related to variating couch attenuation.However, in this study the KAP values are compared between different exposure parameters and the focus is on relative KAP values.Thus, many uncertainty components are correlated, and the uncertainty of their ratio is much lower.The largest uncorrelated standard uncertainty components are the energy dependence of the response of reference dosimeter (1.2%) and the impact on couch attenuation changes with different radiation qualities (4%).In addition, the statistical uncertainties related to readings of reference meter and KAP meter (0.8%) are relevant components of the uncertainty budget.Based on this calculation, the uncertainty of 9% (k = 2) can be used for comparisons of measured KAP values in this fixed CBCT exposure setup.The same applies to the organ dose calculations for the specific phantom model.Here, an additional standard uncertainty component related to the impact of couch on conversion factors is estimated to be 3% and total expanded uncertainty of 11% can be used for comparative purpose.

Image quality assessment and selection of optimized exposure parameters
The image quality of EVAR CBCT phantom scan (axial images) was evaluated subjectively based on geometrical correctness and visibility of the stent by an endovascular surgeon with over 20 years of experience.The optimized exposure parameters were selected based on this evaluation.
In addition, an objective quantitative analysis was made based on a set of regions of interest (ROI) displayed in Fig. 3, in which mean values and standard deviation (SD) of voxel values have been measured.The ROIs were calculated from 20 slices (20⋅0.4765mm = 9.5 mm) where there were no artefacts.The ROI for the stent was manually segmented from the acquisition with the best image quality using the image processing platform 3D Slicer 5.0.2 [16].Small displacement between scans was corrected by rigidly registering the highest dose reference volume with each subsequent image volumes using BRAINSFit tool [17].Then by applying the obtained transformations the same ROIs were used for all CBCT volumes.The background and noise were defined as the average and standard deviation of the voxel values from the background ROI.The contrast to noise ratio (CNR) was calculated as the ratio between the calculated contrast (mean stent voxel value minus the background mean voxel value) and noise.For comparison, the CNR was calculated in the same way for insert A (Calcium 100 mg/ml) and B (Calcium 300 mg/ml).

Clinical validation
A new protocol with optimized EVAR CBCT exposure parameters was created and added to the list of clinical procedures on the equipment.Optimized protocol was thereafter used for several clinical cases and the image quality was critically assessed by an experienced vascular surgeon to verify the acceptable stent visualization in actual intraoperative (post-operative) quality control scans.

Radiation doses
Measured KAP values for different exposure parameters are given in Fig. 4. The vendor pre-defined protocols use a target tube voltage of 90 kV, but the actual tube voltage for our phantom was increased by automatic exposure control and varied from 104 kV for zero copper with 4 s DCT Body Care protocol to 122 kV for 0.9 mmCu with 5 s DCT Body protocol.The preselected dose to the detector level was 0.36 μGy/frame.
It was clear that this was too much for our indication and that is why 0.08 μGy/frame dose level was selected for further studies.

Image quality assessment and selection of optimized exposure parameters
Based on the ROI analysis, the standard deviation, representing noise, is mainly dependent on the dose to detector level.However, the noise increases slightly as a function of energy, which was observed from the images acquired with a fixed dose to the detector.The contrast decreased as a function of increasing X-ray energy as expected, and the CNR decreased together with the contrast (Fig. 5).Based on the Rose criterion, the CNR level of 5 is generally considered acceptable for reliable detection even though it is known that even lower values might still provide acceptable image quality for some tasks [18].In our case, the CNR level of the stent remains over 5 for all dose levels and radiation qualities while for inserts with calcium, the CNR is lower.
Based on the subjective visual image analysis, even for the lowest possible dose level with the lowest dose to detector 0.08 μGy/frame, highest tube voltage 125 kV, and thickest filtration 0.9 mmCu, the contrast of the stent was still acceptable, and the resolution was enough for visualization of the stent geometry (Fig. 6).Therefore, these exposure parameters were selected for the optimized protocol and proposed for further clinical study.

Updated exposure parameters
The spectral difference between the factory and optimized protocol is shown in Fig. 7. Based on the updated protocol, patient doses were calculated and listed in Table 3.

Clinical validation
The new protocol with optimized exposure parameters was used for CIRS adult female anthropomorphic phantom with some smaller stents, and they were clearly visible.After this confirmation, the new protocol was used for clinical cases.Based on the critical assessment of the experienced vascular surgeon, the image quality was considered acceptable.An example case where the new CBCT protocol was used beneficially for a patient is shown in Fig. 8.A separate study with a focus on clinical benefit is warranted.

Discussion
The main purpose of this study was to define the required image quality and optimize the radiation exposure to patients in EVAR CBCT scans while acknowledging the clinical potential of 3D image data in these generally demanding procedures.All the aims of this study were well achieved and even exceeded.The CBCT exposure parameters for EVAR operation were optimized and the radiation exposure to patients was significantly decreased, while still achieving acceptable image quality for the specific indication selected, thus whether the stent graft is open, compressed, bent, or collapsed.This has a major impact on clinical practice since now the CBCT option is much more feasible for routine use and may help prevent re-EVAR operations.Prior to the optimization process, the dose from CBCT would have multiplied the total radiation dose of the EVAR operation by a factor of 2-3.However, after the protocol optimization, the radiation dose is comparable to one digital subtraction angiography imaging set and forms only a minor part of the total exposure (~10%).Based on our knowledge, there are no other studies where the dose level of CBCT in EVAR operation had been decreased so dramatically as in our study.Such a radical change in exposure parameters and dose decrease would be difficult to achieve with clinical optimization with patients, and therefore, a realistic phantom setup was crucial for this work.
The imaging task should be well defined for an optimization study.Basic image quality parameters such as contrast, noise, and resolution are generally used, but especially the contrast is a task-specific parameter.Therefore, the setup for the study should represent a realistic clinical condition, which is to be optimized.In our case, the clinical task was to get information on the condition of EVAR stent graft and the CBCT imaging was performed without contrast media.However, contrast media may be needed for other clinical imaging tasks such as detection of endoleaks or graft thrombosis [3].In these cases, the required dose level and optimal radiation quality might be very different than in our case, and task-specific optimization is needed.The EVAR stent used in this study was made of nitinol, and the behaviour of contrast as a function of energy might be different compared to tissues for example.There are already some tools to improve indication specific optimization where a specific material of interest is selected, and the exposure parameters are adjusted automatically to provide a constant signal for the target across all patient thicknesses [19].This could provide an improved solution also for CBCT imaging.
Since there was no phantom available with an EVAR stent as an image quality target, we had to create one.The Multi-Energy CT phantom has holes with a reasonable size for a stent insert.The initial idea was to put the stent in a syringe allowing to use it several times and to change the position, etc.However, the size of the standard syringe was just a little bit too big for the hole.The condom solution is     disposable, but on the other hand, it provides a perfect insert without any noticeable border with the phantom.The Multi-Energy CT phantom is quite large compared to an average patient, and it does not represent real anatomical structures.The other inserts of the Multi-Energy CT phantom could be selected to better represent different anatomical regions.For example, in our study, the insert with a high calcium was used in the location of spine.The phantom worked for our needs but there is space for further development.
There are some limitations related to dosimetry.The KAP meter was calibrated with the X-ray tube in an over couch position.However, during the CBCT rotation, the radiation beam passes through the couch at different angles.The couch attenuates the beam and the calibration coefficient of the KAP meter is smaller than 1 and it depends on the angle and radiation quality.However, since the rotation geometry will be the same for all CBCT exposures, the impact of radiation quality on relative values is considered small and included as an additional component of the uncertainty.The same applies to the Monte Carlo calculations.The couch would also have a small impact on conversion factors and organ doses, but this impact is probably even smaller in this case when the spectrum is already heavily filtered.Thus, the filter impact of the couch is not significant.The anatomical model used for the simulations do not represent all patients, and thus, the absolute values should not be considered as reparative for real patients.However, the model can be used for comparative calculations, and it is needed to consider the impact of spectra on patient organ doses.
Additional copper filtration is generally used for optimization and lowering the dose.Since the CBCT was offered without any copper filtration, it was clear that there was a good potential for optimization, especially for this very specified clinical indication.The modification was started carefully with small addition of copper filtration and slightly lowering the dose to the detector parameter.However, quite soon, it was clear that the lowest possible detector dose level could be selected, and the spectra could be adjusted so that the mean energy of the photons was as high as possible achieved with the highest tube voltage and the thickest filtration.Even with this selection, we could not reach an unacceptable image quality level.The optimal spectrum to obtain the best CNR for nitinol is probably achieved with lower energies [20], but the highest tube voltage was selected to achieve the lowest possible radiation exposure.This along with the fact that CNR > 5 was maintained, allows some leeway for more attenuating patients.It would be interesting to extend the study even further to find the level, where it is impossible to decrease the radiation dose anymore.However, the selection of radiation dose to the detector should be extended by the company to do this.On the other hand, the copper filtration available by the system could possibly be increased.Also, reducing the dose even further may lead in artefacts dominating over the normal increase in image noise.Therefore, justification in extrapolating from the current results to the ultra-low-dose region is limited.
If the study had been performed with patients, probably such extreme parameters would not have been selected or at least there would have been several optimization steps with small adjustments.Now, by using the patient mimicking phantoms, the exposure parameters could be optimized without a need to involve any patients in the process and then only the final results were validated by clinical cases.The comprehensive study was performed only with one device, and therefore, the results could have been system specific.Therefore, we also repeated the optimization process in a short format with another clinical systems (Artis Zeego, Siemens Healthineers, Erlangen, Germany).The same phantom with the same configuration was used to confirm that the lowest radiation dose to the detector and highest mean photon energy provided acceptable image quality with the phantom, and this introduced a 91% decrease in the KAP value.Based on few clinical cases, also in this case, the optimized protocol was accepted for clinical practice.Therefore, we can just recommend checking if similar optimization could be implemented elsewhere.A similar approach could also be used for other clinical indications for CBCT.

Conclusions
A specific phantom was used for optimization of CBCT imaging in EVAR operations.Based on our findings, appropriate image quality can be achieved even with the lowest possible patient exposure by using the highest mean energy of the spectra and the lowest possible radiation dose to the detector.The KAP value, as well as organ doses and effective dose, decreased about 90% for our patient model.The results were validated by clinical cases.The contribution of CBCT on the total radiation dose of an EVAR operation was significantly changed from a dominant 2-3-fold contribution to minor 10% of the total exposure.A simplified optimization procedure with the same phantom was repeated with another clinical device with similar conclusion.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .
Fig. 2. The modified Multi-Energy phantom with an EVAR stent in the center (EVAR phantom; left) and the corresponding axial CT image (right).

Fig. 3 .
Fig. 3. Regions of interest (ROI) used for the quantitative image evaluation.The stent ROI is shown using light green color.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4 .
Fig. 4. Two upper curves show KAP values for the two factory protocols 5 s DCT Body and 4 s DCT Body Care with a fixed dose to the detector level of 0.36 μGy/ frame and with increasing amount of added copper filtration.The lowest curve is for fixed dose to the detector level of 0.08 μGy/frame using 4 s DCT Body Care rotation and different radiation qualities.

Fig. 5 .
Fig. 5. Two upper curves show contrast to noise ratios (CNR) for the stent using two factory CBCT protocols 5 s DCT Body and 4 s DCT Body Care with fixed dose to the detector level of 0.36 μGy/frame and with increasing amount of added copper filtration.The lowest three CNR curves are for a fixed dose to the detector level of 0.08 μGy/frame using 4 s DCT Body Care rotation and different radiation qualities.They are given for the stent and two other regions of interest (ROIs).A polynomial fit trendline was used to better visualize the connection of points.

Fig. 6 .
Fig. 6.A CBCT image of the modified Multi-Energy phantom with the EVAR stent in the center using the factory protocol 4 s DCT Body Care (left) and the optimized version of the CBCT protocol (right).

Fig. 7 .
Fig. 7. Spectra of the two protocols compared: Original 4 s DCT Body Care protocol (dashed blue line, 104 kV, no copper filtration) and optimized protocol (solid red line, 125 kV, 0.9 mmCu).(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 8 .
Fig. 8.An example patient case (male, height 186 cm, weight 110 kg): Partly thrombosed iliac limbs in stenosed aortic bifurcation A) postoperative CT angiography image (arterial phase) and B) postoperative CT angiography image (venous phase) C) On-table CBCT after high-pressure kissing-balloon angioplasty using the new optimized CBCT protocol.

Table 1
Exposure parameters of Siemens ARTIS Pheno.

Table 3
The radiation doses for different CBCT protocols.