Dynamic computed tomography angiography (CTA) imaging provides additional diagnostic value in characterizing aortic endoleaks. This protocol describes a qualitative and quantitative approach using time-attenuation curve analysis to characterize endoleaks. The technique of integrating dynamic CTA imaging with fluoroscopy using 2D-3D image fusion is illustrated for better image guidance during treatment.
In the United States, more than 80% of all abdominal aortic aneurysms are treated by endovascular aortic aneurysm repair (EVAR). The endovascular approach warrants good early results, but adequate follow-up imaging after EVAR is imperative to maintain long-term positive outcomes. Potential graft-related complications are graft migration, infection, fraction, and endoleaks, with the last one being the most common. The most frequently used imaging after EVAR is computed tomography angiography (CTA) and duplex ultrasound. Dynamic, time-resolved computed tomography angiography (d-CTA) is a reasonably new technique to characterize the endoleaks. Multiple scans are done sequentially around the endograft during acquisition that grants good visualization of the contrast passage and graft-related complications. This high diagnostic accuracy of d-CTA can be implemented into therapy via image fusion and reduce additional radiation and contrast material exposure.
This protocol describes the technical aspects of this modality: patient selection, preliminary image review, d-CTA scan acquisition, image processing, qualitative and quantitative endoleak characterization. The steps of integrating dynamic CTA into intra-operative fluoroscopy using 2D-3D fusion-imaging to facilitate targeted embolization are also demonstrated. In conclusion, time-resolved, dynamic CTA is an ideal modality for endoleak characterization with additional quantitative analysis. It can reduce radiation and iodinated contrast material exposure during endoleak treatment by guiding interventions.
Endovascular aortic aneurysm repair (EVAR) has shown superior early mortality results than open aortic repair1. The approach is less invasive but may result in higher mid to long-term re-intervention rates due to endoleaks, graft migration, fracture2. Hence better EVAR surveillance is critical to achieving good mid to long-term results.
Current guidelines suggest the routine use of duplex ultrasound and triphasic CTA3. Dynamic, time-resolved computed tomography angiography (d-CTA) is a relatively new modality used for EVAR surveillance4. During d-CTA, multiple scans are acquired in different time points along the time attenuation curve after contrast injection, hence the term time-resolved imaging. This approach has shown better accuracy in characterizing endoleaks after EVAR than conventional CTA5. An advantage of time-resolved acquisition is the ability to quantitatively analyze the Hounsfield unit changes in a selected region of interest (ROI)6.
The additional benefit of accurately characterizing endoleaks with d-CTA is that the scan can be used for image fusion during interventions, potentially minimizing the need for further diagnostic angiography. Image fusion is a method when previously acquired images are overlaid onto real-time fluoroscopy images to guide endovascular procedures and subsequently reduce contrast agent consumption and radiation exposure7,8. Image fusion in the hybrid operating room (OR) using a 3D dynamic CTA scan can be achieved by two approaches: (1) 3D-3D image fusion: where 3D d-CTA is fused with intraoperatively acquired non-contrast cone-beam CT images, (2) 2D-3D image fusion, where 3D d-CTA is fused with biplanar (anteroposterior and lateral) fluoroscopic images. 2D-3D image fusion approach has been shown to significantly lower the radiation compared with 3D-3D technique9.
This protocol describes the technical and practical aspects of dynamic CTA imaging for endoleak characterization and introduces a 2D-3D image fusion approach with d-CTA for intra-operative image guidance.
This protocol follows the ethical standards of the national research committee and with the 1964 Helsinki declaration. This protocol is approved by Houston Methodist Research Institute.
1. Patient selection and prior image review
NOTE: Dynamic CTA imaging should be considered as a follow-up imaging modality in patients with increasing aneurysm size and endoleak after stent-graft implantation, persistent endoleak after interventions, or in patients with increasing aneurysms sac size without demonstrable endoleak. Like conventional CT imaging, this technique involves iodinated contrast injection that may be relatively contraindicated in patients with severe renal failure.
2. d-CTA Image acquisition
3. Dynamic-CTA image analysis
4. Intra-operative image fusion guidance
The dynamic imaging workflow in two patients is illustrated here.
Patient I
An 82-year-old male patient with chronic obstructive pulmonary disease and hypertension had a previous infrarenal EVAR (2016). In 2020 the patient was referred from an outside hospital for a possible type I or type II endoleak based on conventional CTA. and an adjunctive endoanchor placement in 2020 for type Ia endoleak. Dynamic CTA was performed that diagnosed a type Ia endoleak, and the patient underwent proximal zone ballooning plus received endoanchors to gain more sealing zone for the graft. After the intervention, a dynamic control CTA was performed, acquiring 12 scans under 21 s scan time with 90 kV using 85 mL iodinated contrast material. Qualitative analysis showed a persisting type Ia endoleak illustrated in Figure 5. Quantitative TAC analysis showed a 12.2 s time to peak value for ROIaorta and a 15.4 s to peak value for ROIendoleak creating a 3.2 s time to a peak value (Figure 6). The patient received a fenestrated-EVAR; the procedure was done using 2D-3D image fusion during the procedure.
Patient II
A 62-year-old male patient with a medical history of obesity, stroke, renal insufficiency (creatinine: 2.02 mg/dL), hypertension, hyperlipidemia, and coronary artery disease. The patient received an infrarenal EVAR at an outside hospital in 2018. He was referred to our Institution for a possible type II endoleak on conventional CTA. Dynamic CTA was performed with acquiring 12 scans under 52 s at 100 kV using 70 mL iodinated contrast material. Sac enlargement with a type II endoleak was detected from bilateral L3 lumbar arteries as inflow vessels shown in Figure 7. Time attenuation curve analysis showed a 7.2 s time to peak value for ROIaorta and 24.6 s for ROIendoleak at the level of the L3 vertebra (Figure 8). An additional ROI was selected in the inferior portion of the sac, demonstrating the downward flow from the level of the bilateral lumbar arteries by the delayed time to a peak value (ROIendoleak2 = 30.8 s). The Δ time-to-peak value for the endoleak was 17.3 s. The patient underwent transarterial coil embolization of the aneurysm sac using 2D-3D image fusion as guidance during the procedure.
These two cases are presented to illustrate the technique described in the protocol section. Patients who underwent d-CTA imaging had potential endoleak (Patient selection). Prior image review was done to personalize individual scans such as higher kV than average for patients with a higher body-mass index (BMI), longer acquisition for possible type II endoleak (Patient II), shorter for Patient I with possible type I endoleak. Appropriate kV selection is crucial in ensuring adequate image quality; too low kV can result in suboptimal images (Figure 9A). The timing of the scans was made according to the step 2.4 of the protocol; this is an essential part because later launched acquisitions result in timing error and may influence qualitative analysis (Figure 9B). Image analysis was done in the dedicated software using the Dynamic Angio preset (Figure 1 and Figure 2). The images were analyzed both qualitatively and quantitatively (Figure 5–Figure 8). Intra-operative image fusion was used to guide the intervention. The hybrid OR workstation aligned the fluoroscopic images with d-CTA images (Figure 4), as mentioned in step 4 of the protocol.
Figure 1: Dynamic CTA scan opened with CT dynamic angio protocol. (A, B, C) The sagittal, axial, and coronal plane reconstructions aligned together. (D, E) Reconstructed images of a patient after a fenestrated-EVAR. The Blue arrow on the right shows the dynamic scans that are used for the review. The green arrow on the left shows the motion correction function (align body). This step is the initial when reviewing images. The white arrow on the left shows the timeline of the total scans, which can be changed manually or played continuously using the "watch" function. ROIs for TAC curves can be selected using the "TAC" function (yellow arrow). Please click here to view a larger version of this figure.
Figure 2: Example of a TAC analysis in a patient with a type II endoleak from a lumbar artery as inflow. (A) The selected ROI (yellow above the stent-graft (ROIaorta), green inside the aneurysm sac where endoleak is visualized (ROIendoleak)). (B) This image demonstrates the generated time-attenuation curves for the selected ROIs in panel A. Time difference between aortic and endoleak curves in reaching peak Hounsfield unit is recorded (Δ time to peak value – marked with white) Please click here to view a larger version of this figure.
Figure 3: Layout of the workstation in the hybrid OR to align the biplanar fluoroscopy images with the 3D dynamic scan (2D-3D image fusion). Yellow arrows highlight the wires inside the aorta, blue arrows show the inferior portion of the stent graft. Panel on the right is to manually modify the automatic alignment: visualization of fluoroscopic and d-CTA imaging, different image selection, fine modification of alignment, accepting the alignment. Additional measurements and annotations can be made using the blue box on the right panel. Please click here to view a larger version of this figure.
Figure 4: Image of the overlaid markers on the real-time fluoroscopic image during coil embolization. The patient had a previous chimney-EVAR and a subsequent Ia gutter endoleak which was treated via coil embolization. Yellow arrows highlight the coil. Purple color is the marked endoleak cavity inside the deployed coils. Green circle indicates the fenestration of the implanted stent graft, horizontal green and blue lines are entrance for gutters next to the endoleak, and orange marks the top of the chimney graft. Please click here to view a larger version of this figure.
Figure 5: An image of the 82-year-old male patient referred after an EVAR with possible type I or type II endoleak based on conventional CTA imaging. Sequentially imaged axial and sagittal plane scans are shown in the highlighted timepoint of the scan (the left top corner indicates the timepoint in seconds). A dashed yellow line marks the level of the axial images. The yellow arrow shows the contrast enhancement in the anterior margin of the stent-graft above the aneurysm sac, demonstrating a type Ia endoleak. Please click here to view a larger version of this figure.
Figure 6: Time attenuation curve analysis of the patient shown in Figure 5. Selected ROIs are shown in (A) and (C) axial scans (aortic ROI at the top of the graft with orange and endoleak ROI at the level of contrast enhancement outside the graft). (B) is the TAC corresponding with the selected ROIs. The white box highlights the time to peak values for each region: ROI3=aorta and ROI2=endoleak). The borders of the Δ time to peak value are shown with white dashed lines. The time interval between the two lines is the Δ time to peak value, which was 3.2 s. The short difference between peak values corresponds with type I endoleak. Please click here to view a larger version of this figure.
Figure 7: Sequentially imaged, reconstructed axial and sagittal plane images of a 62-year-old male patient with a suspected type II endoleak. Each time point of the scan is shown in a separate panel (timepoints are shown in the top left corner). The dashed yellow line on the first sagittal image demonstrates the level of the axial images. Dynamic CTA showed sac enlargement with a type II endoleak from bilateral lumbar arteries at the level of the L3 vertebra (blue arrows). Endoleak is highlighted with yellow arrows. Time-resolved sagittal images demonstrate the downward flow inside the aneurysm sac from the level of the L3 lumbar vertebra. Please click here to view a larger version of this figure.
Figure 8: Time attenuation curve for the type II endoleak. (A) The yellow circle shows the ROI for the aortic enhancement curve, green shows the ROI for the endoleak enhancement curve at the level of the L3 vertebra, and orange shows it at the level of the L4 vertebra. (B) Corresponding analysis of the curves showed a delayed Δ time to peak value for the endoleak (17.3 s) and a more delayed peak for the green region, demonstrating the downward flow. This confirms the presence of a type II endoleak. Please click here to view a larger version of this figure.
Figure 9: This image shows the pitfalls of dynamic CTA image acquisition. (A) A scan was done at 70 kV for a patient with a BMI of 37.4. A high BMI value requires higher radiation exposure for acquiring acceptable images. (B) A timing error of a dynamic CTA. This scan was triggered later, and the aortic curve was already at the peak enhancement point when the acquisition started. The time attenuation curve shows the time to peak value at 0.2 s above the stent graft (corresponding ROIaorta shown in C). TAC can be used to calculate Δ time to peak value even in these cases as well. Please click here to view a larger version of this figure.
Protocol | DynMulti4D |
Total number of volumes | 11-13 scans |
– 2-4 scans @ every 1.5 s | |
– 4 scans @ every 3 s | |
– 2-4 scans @ every 4.5 s | |
Tube voltage | 70-100 kV |
Tube current | 150 mAs |
Rotation time | 0.25 s |
Scan duration | 36±10 s |
Slice thickness | 0.7-1 mm |
Contrast material volume | 70-90 mL |
Flow rate | 3.5-4 mL/s |
Saline flush | 90-100 mL |
Scan range (z-axis) | 23-33 cm |
Pitch | 1 |
Reconstruction parameters | ADMIRE-3, Bv36 kernel |
Dose-length product | 593 (Patient I) and 445 Patient (II) mGy*cm |
Table 1: Parameters of a customized d-CTA endoleak protocol. *Body-mass index for Patient I and II were 26.1 and 21.4 m2/kg.
Dynamic, time-resolved CTA is an additional tool in the aortic imaging armamentarium. This technique can accurately diagnose endoleaks after EVAR, including identification of inflow/target vessels4.
Third-generation CT scanners with bidirectional table movement capability can provide dynamic acquisition mode with better temporal sampling along the time-attenuation curve6. To achieve the highest accuracy in the protocol it is critical to personalize image acquisition: review previously existing imaging set scan parameters according to patient requirement (high BMI – higher kV, cover the whole endograft with the scan, distribute scans based on suspected endoleak) and time the acquisition to cover aortic and endoleak enhancement curves (badly timed scan is shown in Figure 9B). An iodinated contrast agent with 320 mg of Iodine/mL was used in this study. While other contrast agents with lower iodine concentration may be used using this d-CTA protocol, increasing the contrast injection rate or volume might be necessary to achieve at least ~500 HU in the aortic region of interest.
Lower kV imaging comes at its own cost, especially in patients with higher BMI, as illustrated in Figure 9A. Advanced image reconstruction techniques using model-based, statistical methods may help with improving image quality at lower radiation doses, especially during d-CTA imaging.
Mistiming a scan can misrepresent quantitative data along the time attenuation curve (Figure 9B). Although such dynamic imaging techniques can be implemented in most third-generation CT scanners, a learning curve is associated with image acquisition, reconstruction, and post-processing time-resolved datasets.
The apparent roadblock for routine adoption of such dynamic, time-resolved CT imaging techniques concerns radiation and contrast exposure. While the amount of contrast injected is equivalent to triphasic CT imaging, the additional radiation exposure can be mitigated by lowering kV, selecting relevant scan range, and utilizing advanced iterative reconstruction techniques. Recent studies have shown that dynamic CTA can be performed without additional radiation exposure than conventional triphasic CTA5,10,11,12. Minimizing radiation exposure of patients in EVAR surveillance is shown to be an essential and non-negligible factor13. This can be relevant in further CTA scan optimization to reduce scan numbers and subsequent radiation exposure without losing diagnostic accuracy14. Scan range is another crucial aspect that can be a limitation when using d-CTA; in our experience, 33 cm is the maximum length covered. Koike et al. using their different scanner and smaller scan range, published their approach to overcoming this limitation with promising results11.
A previous study compared the accuracy of conventional and dynamic CTA and their impact on the number of digital subtraction angiographies during endoleak treatment5. Dynamic CTA has shown better endoleak diagnosing capability than conventional triphasic CTA5. According to recent papers, traditional CTA surveillance after EVAR may misdiagnose type II endoleaks, and multiple failed treatment attempts should raise suspicion for a different type of endoleaks10. The use of quantitative and qualitative image analysis from d-CTA may help overcome the limitation of diagnosing such misdiagnosed/occult endoleaks using conventional techniques15.
Image post-processing involves reviewing time-resolved dynamic CTA images and 2D-3D image fusion, typically taking ~5-10 min. Inaccuracies during image fusion may arise from the following factors: imperfect alignment of stent-graft from d-CTA with fluoroscopy, patient movement during the intervention, deformation of the aorta with stiff wires/devices. Further automation of image fusion techniques and workflow is required for better, seam-less intra-operative image guidance.
In our experience, d-CTA imaging has also been shown to provide additional image-fusion guidance during endoleak treatment6. Such dynamic time-resolved imaging can also be helpful in future imaging of other dynamic disease processes such as aortic dissection, peripheral arterial disease, arteriovenous malformations, or intramural hematoma16,17,18.
The authors have nothing to disclose.
The authors would like to acknowledge Danielle Jones (Clinical education specialist, Siemens Healthineers) and the entire CT technologist team at Houston Methodist DeBakey Heart and vascular center to support imaging protocols.
Siemens Artis Pheno | Siemens Healthcare | https://www.siemens-healthineers.com/en-us/angio/artis-interventional-angiography-systems/artis-pheno | Other commercially available C-arm systems can provide image fusion too |
SOMATOM Force CT-scanner | Siemens Healthcare | https://www.siemens-healthineers.com/computed-tomography/dual-source-ct/somatom-force | Any commercially available third generation CT-scanner can perform such dynamic imaging |
Syngo.via | Siemens Healthcare | https://www.siemens-healthineers.com/en-us/medical-imaging-it/advanced-visualization-solutions/syngovia | Any DICOM file viewer with 4D processing capabilities can review the acquired time-resolved images, TAC are software dependent. |
Visipaque (Iodixanol) | GE Healthcare | #00407222317 | Contrast material |