Endovascular aneurysm repair: prevention and treatment of complications
Goudeketting, Seline
DOI:
10.33612/diss.98524202
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2019
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Goudeketting, S. (2019). Endovascular aneurysm repair: prevention and treatment of complications.
Rijksuniversiteit Groningen. https://doi.org/10.33612/diss.98524202
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
The Use of 3D Image Fusion
for Percutaneous Transluminal
Angioplasty and Stenting of
Iliac Artery Obstructions:
Validation of the Technique and
Systematic Review of Literature
Seline R. Goudeketting* Stefan G. H. Heinen* Daniel A. F. van den Heuvel
Marco J. van Strijen Michiel W. de Haan Cornelis H. Slump Jean-Paul P.M. de Vries
*Both authors contributed equally J Cardiovasc Surg (Torino). 2018 Feb;59(1):26-36
ABSTRACT
Background: The effect of the insertion of guidewires and catheters on fusion accuracy of the three-dimensional (3D) image fusion technique during iliac percutaneous transluminal angioplasty (PTA) procedures has not yet been investigated.
Methods: Technical validation of the 3D fusion technique was evaluated in 11 patients with common and/or external iliac artery lesions. A preprocedural contrast-enhanced magnetic resonance angiogram (CE-MRA) was segmented and manually registered to a cone-beam computed tomography image created at the beginning of the procedure for each patient. The treating physician visually scored the fusion accuracy (i.e., accurate [<2 mm], mismatch [2–5 mm], or inaccurate [>5 mm]) of the entire vasculature of the overlay with respect to the digital subtraction angiography (DSA) directly after the first obtained DSA. Contours of the vasculature of the fusion images and DSAs were drawn after the procedure. The cranial-caudal, lateral-medial, and absolute displacement were calculated between the vessel centerlines. To determine the influence of the catheters, displacement of the catheterized iliac trajectories were compared with the noncatheterized trajectories. Electronic databases were systematically searched for available literature published between January 2010 till August 2017.
Results: The mean registration error for all iliac trajectories (N=20) was small (4.0 ± 2.5 mm). No significant difference in fusion displacement was observed between catheterized (n = 11) and noncatheterized (n = 9) iliac arteries. The systematic literature search yielded 2 manuscripts with a total of 22 patients. The methodological quality of these studies was poor (≤11 MINORS score), mainly due to a lack of a control group.
Conclusions: Accurate image fusion based on preprocedural CE-MRA is possible and could potentially be of help in iliac PTA procedures. The flexible guidewires and angiographic catheters, routinely used during endovascular procedures of iliac arteries, did not cause significant displacement that influenced the image fusion. Current literature on 3D image fusion in iliac PTA procedures is of limited methodological quality.
INTRODUCTION
Peripheral artery disease (PAD) of the lower extremities affects between 15% and 20% of people aged older than 70 years.1 The iliac artery is involved in up to 30% of patients with PAD.2 Percutaneous transluminal angioplasty (PTA), with or without stent placement, is the preferred treatment option in most of these patients. For a successful technical outcome, the use of iodinated contrast media and ionizing radiation is still necessary in most procedures to create 2-dimensional (2D) digital subtraction angiographies (DSAs) to visualize the location of the lesion and evaluate the severity of a stenosis.2 In recent years, the policy to reduce radiation exposure for patients and physicians, as well as the amount of nephrotoxic contrast, has been embraced globally.
During the preprocedural workup, contrast-enhanced magnetic resonance angiography (CE-MRA) or computed tomography (CT) angiography (CTA) are routinely used to image the aortic-femoral trajectory. The development of new image guidance tools in angiography suites allows for continuous fusion guidance of such preprocedural imaging during endovascular interventions. The technique of 3-dimensional (3D) image fusion relies on performing a rigid registration of preprocedural image data, such as CTA or CE-MRA, to a preprocedurally acquired cone-beam CT (CBCT) or 2 fluoroscopic orthogonal images. Registration takes approximately 5 to 10 minutes and is based on the fusion of bony landmarks and, in case of CTA, arterial wall calcifications. Image fusion significantly reduces the volume of iodinated contrast in endovascular aneurysm repair (EVAR) procedures.3 In addition, it may reduce the total amount of ionizing radiation, fluoroscopy time, and procedure time, especially in more complex cases.3 This is particularly true for the juxtarenal aortic anatomy with relatively fixed orifices of the target visceral arteries. However, the introduction of stiff guidewires and delivery systems during EVAR procedures will result in deformation of the iliac arteries4 and can disturb the accuracy of the rigidly fused images of these access arteries.5 In our clinical experience, this is particularly the case in elongated iliac arterial anatomy. A few studies have described the feasibility of the 3D image fusion technique for peripheral artery interventions.6,7 These studies, however, did not investigate the effect of the insertion of guidewires and catheters on the fusion accuracy during endovascular iliac artery interventions, which is the main subject of the current study. We conducted a systematic review of the available literature regarding the 3D image fusion technique in endovascular iliac artery interventions to evaluate the available clinical information on this subject.
METHODS
Technical validation of 3D fusion technique
Periprocedural analysis
Technical validation of the 3D fusion technique was evaluated in common and/or external iliac artery lesions that were treated during elective percutaneous interventions (i.e., PTA or recanalization, with or without additional stenting) in the angiography suite. All procedures were performed in patients with disabling claudication that did not respond to a supervised exercise program. The local medical ethical committee (METC azM/UM) approved the study protocol, and all patients gave written informed consent before study inclusion.
Besides a clinical examination, measurement of ankle-brachial indices, and treadmill tests, all patients underwent a preprocedural peripheral artery CE-MRA (Intera 1.5T; Philips Healthcare, Best, The Netherlands). A standard T1-weighted fast field echo protocol was used with a field of view of 430 × 300 × 105 mm3 and an acquired spatial resolution of 1.41 × 1.41 × 3.00 mm3. The reconstructed voxel size was 1.22 × 1.22 × 3.00 mm3. The first step for fusion imaging was the segmentation of the preprocedural CE-MRA. The Philips XperGuide software (Version R1.4.0.10030; Philips Healthcare) was used to manually remove all anatomical structures, apart from the aorta and iliac and femoral arteries. The resulting segmented 3D data set is visible during the intervention.
Before the start of the endovascular procedure, patients were placed supine on a vacuum mattress on the angiography table. When the vacuum was created, the mattress formed a mold around the patient. This minimized leg and pelvic movement during the intervention, thus limiting registration errors caused by movement artifacts. A noncontrast-enhanced CBCT of the pelvic area was created at the beginning of the procedure using the flat panel detector of the C-arm angiography system (Allura Xper FD20; Philips Healthcare). The 8-second CBCT was reconstructed to a 3D volume with a maximum spatial resolution of 0.65 × 0.65 × 0.65 mm3 and was automatically transferred to the 3D XtraVision workstation (Philips Healthcare).
A technical physician performed the manual 3D-3D registration of the original preprocedural CE-MRA and CBCT. Bony landmarks, such as the vertebrae and femoral heads, were used for registration. In addition, calcified spiculae of the arterial walls, as visible on the CBCT, were registered to the vessel lumen, as visible on the CE-MRA (Figure 2.1). The image registration linked the CE-MRA to the 3D coordinate system of the C-arm. Therefore, the view angle of the segmented CE-MRA images followed rotations of the C-arm, table movements, and furthermore, adapted to magnification. The time to perform the registration was recorded. The endovascular procedure was performed
under fusion guidance, using additional contrast runs whenever deemed necessary. All patients underwent an initial DSA run to confirm the accuracy of the fusion images. The treating physician visually scored the fusion accuracy of the entire vasculature of the overlay with respect to the DSA directly after the first obtained DSA (Figure 2.2A-B). Fusion quality was categorized as accurate (<2 mm), mismatch (2–5 mm), or inaccurate (>5 mm). If needed, the fused images could be manually translated or reregistered from the current fusion images to achieve a more accurate overlay before the actual revascularization procedure started. The DSAs and fusion images were saved for postprocedural analysis. Before the end of the procedure, a completion DSA confirmed the procedural success, defined as a <30% residual stenosis over the treated iliac lesion. Contrast runs were performed with the 4F Cobra catheter (Radifocus Glidecath, Terumo Europe, Leuven, Belgium); interventions were performed over a 0.035-inch angled Glidewire (Terumo Europe).
Postprocedural analysis and statistics
The image attributes of the initial DSA run were transcribed to the Digital Imaging and Communications in Medicine (DICOM) header of the corresponding secondary captured fusion images. This did not alter the image data of the fusion images but enabled the analysis of the secondary captured fusion images with the Cardiovascular Angiography Analysis System (CAAS) workstation (Release 7.5; PIE Medical Imaging, Maastricht, The Netherlands). The fusion images and DSAs were both analyzed with the Quantitative Vessel Analysis (QVA) of the CAAS software. After image calibration was performed using the known catheter size, the arterial wall contour lines were semiautomatically drawn. The contour coordinates were exported, and 2D centerline reconstructions were performed using Matlab R2015b software (The Mathworks, Inc.,
Figure 2.1. Manually performed rigid registration between the CBCT (red) and the CE-MRA (blue) in the axial, coronal, and sagittal planes. Bony landmarks and calcified speculae of the arterial walls, visible on CBCT, were registered to the vessel lumen of the CE-MRA.
Natick, MA, USA). To perform a fair analysis between the corresponding points at the DSA and fusion image, a cross-correlation method (Appendix I) was used to pair points along the reconstructed centerlines. To overcome the various vessel lengths of patients, vessel length was normalized to a distance between 0 and 1, with 0 being the aorta bifurcation and 1 being the common femoral artery bifurcation. The cranial-caudal, lateral-medial, and absolute displacement were derived from the distance between the centerlines of the DSA and the fusion image (Figure 2.3A-D).
Figure 2.2. Two examples of image fusion as seen during the first DSA run. Panel A shows a patient where the fusion accuracy was scored inaccurate. The overlay can then be manually translated to improve the fusion accuracy. Panel B shows an example of an accurate fusion.
Figure 2.3. Example of displacement calculation between (A) DSA and (B) CE-MRA. Quantitative Vessel Analysis (QVA) software was used to segment the vessel wall, and the luminal centerline was calculated. (C) The left iliac trajectory of the DSA and CE-MRA images were overlaid. (D) The displacement between corresponding points on the centerline was evaluated in the lateral-medial and cranial-caudal direction. The absolute displacement was also calculated, which is represented by the black line (black arrow).
To determine the possible influence of the diagnostic catheter on the vascular anatomy, the centerlines of the DSA and fusion image were translated and aligned at the aorta bifurcation. The fusion image centerline was then rotated to optimize the match with the DSA, and the displacement analysis was repeated. However, different from previous analysis, catheterized iliac trajectories were compared with the noncatheterized iliac trajectories, which served as a reference.
We also analyzed the learning curve for performing the registration by calculating the average absolute displacement for every performed procedure.
Statistical analysis
Normally distributed data are reported as the mean ± standard deviation, and non-normally distributed data are reported as median (range). Statistical significance was analyzed with the independent t-test or Mann-Whitney U-test for normally and non-normally distributed data, respectively. P values were considered significant when α < .05.
Systematic review
Literature search
The PubMed/MEDLINE, Embase, and the Cochrane Database of Controlled Trials databases were searched to identify clinical studies published in English between January 2010 till August 2017 on the use of 3D image fusion technique during endovascular iliac artery interventions. The Medical Subject Heading terms used were “software” OR “computer-assisted surgery” or “multimodal imaging” OR “cone-beam computed tomography” OR “interventional radiology” OR “fluoroscopy” OR “three-dimensional imaging” OR “angiography,” in combination with (AND) “overlay” OR “fusion” OR “roadmap.” The same search terms were used as keywords, and all combinations of these keywords were added to the search as free-text terms, limited to the title and abstract.
Inclusion/Exclusion Criteria
Two authors (SRG and SGHH) independently reviewed the title and abstract of each manuscript that was identified with the broad search. A strict inclusion criterion was used. Manuscripts were included only when the 3D image fusion technique (both 2D-3D or 2D-3D-2D-3D registration) was used for periprocedural guidance during endovascular interventions of iliac artery obstructions. The reference lists of the selected manuscripts were assessed for further relevant publications. The 2 reviewers compared the articles for eligibility, and full papers were retrieved if 1 or both of the reviewers considered the title and abstract eligible.
Data Collection and Quality Assessment
The extracted data included the number of patients, 3D image fusion details, type of endovascular intervention, and study endpoints. The 12-item Methodological Index for Non-Randomized studies (MINORS) scoring system was used to assess the methodological quality of the studies.8 A maximum score of 24 could be achieved (0 = not reported, 1 = reported but inadequate, or 2 = reported and adequate), where a score of ≤14 is considered poor quality, 15 through 22 is moderate quality, and ≥23 is good quality. Two authors (SRG and SGHH) independently scored the included studies. The opinion of a third reviewer (JPPMdV) was sought in case of discrepancies.
RESULTS
Technical validation of the 3D fusion technique
We evaluated 11 patients (7 men [63.6%]; median age, 68 years [range, 53–88 years]) with a total of 15 common and/or external iliac artery occlusion (n = 5) and/or stenosis (n = 10). All suffered intermittent claudication that did not respond to a supervised exercise program. Of the 11 patients, 4 had 2 iliac artery lesions, and 3 of those 4 patients had bilateral lesions. At the time of the initial DSA run, 2 patients were catheterized on both sides. One patient did not undergo PTA because the stenotic lesion observed on the CE-MRA was non-significant on the initial DSA run. The procedures (1 DSA, 9 PTAs with additional stenting, and 5 recanalizations with additional stenting) were successful and uncomplicated. No complications occurred related to the use of the 3D image fusion technique.
The median time between the preprocedural CE-MRA and the PTA procedure was 32 days (range, 17–82 days). The median volume of administered iodinated contrast agent was 28 mL (range, 15–56 mL). The median radiation dose (dose area product) of the procedures was 54.1 Gycm2 (range, 18.0–159.0 Gycm2), and median fluoroscopy time (in minutes:seconds) was 06:03 (range, 01:29–23:36). The median total procedure time from femoral access to arterial closure was 49 minutes (range, 13–111 minutes).
Periprocedural analysis
Image fusion was feasible in all cases with a median time of 6 minutes (range, 2–10 minutes). The interventionalist visually scored the initial image fusion accuracy after acquiring the first DSA run. Fusion quality was judged accurate in 3 patients (27%), mismatch in 6 (55%), or inaccurate in 2 (18%). In case of a mismatch or inaccurate fusion quality, the fusion could be translated to a more accurate fusion quality before the intervention was continued.
In total, there were 22 iliac trajectories to analyze, of which 13 were iliac trajectories with catheter or guidewire, and 9 were iliac trajectories without. Two of the trajectories with catheter or guidewire were excluded from the analysis because of long occlusions present on the CE-MRA and DSA. The remaining 20 iliac trajectories (i.e., 9 without and 11 with a catheter or guidewire) were used to determine the accuracy of the registration, the influence of the catheter on the image vasculature, and the possible learning curve present when performing the registration of the CE-MRA to the CBCT.
Postprocedural analysis
The lateral-medial, cranial-caudal, and absolute displacements were determined by comparing the centerline of the fusion image with the initial DSA centerline (n = 20). The aortic bifurcation was visible on all DSA and fusion images. On average, the centerline of the fusion image was registered more medial and caudal compared with the DSA, –0.6 ± 3.7 mm and –0.2 ± 3.0 mm, respectively (Figure 2.4). The registration error between the DSA and the fusion image increased toward the distal end of the external iliac artery. The mean absolute registration error between the DSA and the fusion image from the aortic bifurcation to the femoral artery bifurcation was 4.0 ± 2.5 mm.
Influence of inserting a catheter or guidewire in the iliac arteries
Catheterized and noncatheterized trajectories were compared to determine the influence of the endovascular equipment on fusion accuracy. The registration errors at the noncatheterized sites were equal to the errors of the catheterized sites (Figure 2.5, A vs D, B vs E, and C vs F). The average displacement errors between the iliac vessel trajectories without (n = 9) and with (n = 11) a catheter were (1) lateromedial, 0.0 ± 0.8 mm versus 0.1 ± 0.9 mm (P = .447), (2) craniocaudal, 0.2 ± 0.6 mm versus 0.4 ± 0.7 mm, (P = .255), and (3) Absolute, 0.9 ± 0.6 mm versus 1.0 ± 0.6 mm (P = .447). Of note, the displacement in the aortic bifurcation (distance 0) is perfect because this point was used to match the DSA and fusion image.
Learning curve
The average displacement error decreased with increasing number of registrations performed by the technical physician, as seen in Figure 2.6. The trend line shows the average displacement error was reduced from 5.6 mm to 2.5 mm after 11 procedures were performed.
Figure 2.4. Displacement of the fusion image relative to DSA for all 20 iliac trajectories that could be analyzed. Displacement is shown in the (A) lateral-medial and (B) cranial-caudal direction. (C) The absolute displacement error increases towards the common femoral artery. Lengths of all 20 iliac trajectories were normalized from 0 to 1, with 0 being the location of the aorta bifurcation and 1 the common femoral artery bifurcation. The blue line represents the mean ± standard deviation of all displacement lines. Notably, some of the displacement lines were interrupted owing to an occluded segment for which no displacement could be determined.
Systematic review
Included studies
The literature search yielded 8441 articles, of which 8437 were excluded because there was (1) no intraoperative use of the 3D image fusion technique and/or (2) no iliac artery intervention was performed (Figure 2.7). The crosscheck of references of the eligible articles did not identify additional manuscripts. The full-text of 4 manuscripts was evaluated, after which 2 articles were excluded. The first was excluded because the report was an EVAR procedure with an iliac side-branch device9 and the second because of the treatment of an internal iliac artery aneurysm.10 Two manuscripts met the inclusion criteria.6,7 The study characteristics are reported in Table 2.1. The methodological quality of both articles was poor, mainly because of a lack of a control group. The studies used both CTA and CE-MRA for the 3D image fusion technique. Both studies used the Allura XtraVision 8.3 workstation (Philips Healthcare).
Figure 2.5. (A) The lateral-medial, (B) cranial-caudal, and (C) absolute displacement of the fusion image relative to DSA for iliac trajectories without a catheter (n = 9) are shown; (D-F) similar displacements are shown for the catheterized (n = 11) iliac trajectories. Lengths of all iliac trajectories were normalized from 0 to 1, with 0 being the location of the aorta bifurcation and 1 the common femoral artery. The blue lines represent the mean ± standard deviation of the displacement lines.
Figure 2.6. The average displacement error between the DSA and fusion image for all 11 performed procedures. The registration of the images became more accurate with increased experience of the technical physician.
Figure 2.7. Flowchart of the literature search.
Description of studies
Ierardi et al.7 evaluated 5 patients with aorto-iliac steno-occlusive disease, of which 4 patients underwent preprocedural CTA and 1 patient underwent CE-MRA. Registration was based on arterial wall calcifications and bony landmarks. The registration process took approximately 3 to 4 minutes and was performed by a skilled interventional radiologist. The operator judged the overlay accuracy and decided whether the fusion was acceptable or not. All fusion images were sufficient, meaning no CBCT scans had to be repeated.
To evaluate the feasibility, precision, and added value of image fusion guidance in peripheral artery interventions, 17 patients were treated under fusion guidance for 14 common and/or external iliac artery and 6 superficial femoral artery and/or popliteal artery interventions in the study by Sailer et al.6 All interventions were performed under local anesthesia in the angiography suite. Registration was performed based on vessel wall calcifications, bony and organ landmarks, and took the interventional radiologists an average of 5 ± 2 minutes. The average maximum difference between the position of the vasculature on angiography and CE-MRA/CTA fusion roadmap was 1.86 ± 0.95 mm after the exclusion of 3 patients with substantial leg and pelvis movement. The fusion roadmap had an added value in 75% of the procedures, and pretreatment angiography series could be omitted in 47% of the patients when evaluated by the executing interventional radiologist. Completion angiograms were performed in all patients to confirm treatment success and to detect possible complications.
2
Search results (n = 8441) 4000 Pubmed/Medline 4378 Embase
63 Cochrane database of controlled trials
Full-text study excluded (n = 2) Reason:
- Endovascular aortic repair procedure with the
addition of an iliac side branch device
- Coil embolization and stent placement of a left
internal iliac artery aneurysm Full-text assessed for eligibility (n = 4)
Duplicate publications and irrelevant articles (title/abstract) excluded (n = 8437)
Table 2.1. Included studies of the systematic review. Study N Study Type Pre-operative Imaging Fusion with
Intervention Endpointsa MINORS
Ierardi 2015 7
5 Feasibility
study
CTA/MRA CBCT Angioplasty and
stent placement in aorto-iliac steno-occlusive disease IC: 20 (20-40) DAP: 60.05 (55.02-63.75) PT: 33 (27-38) FT: 12.42 (10.17-14.25) RT: 3-4 11 Sailer 2014 6 17 Feasibility study CTA/MRA CBCT Recanalization and/or stenting and/or PTA in CIA/ EIA/SFA/popliteal artery/tibiofibular trunk IC: 58 (30-125) PT: 102 (45-260) RT: 5±2 RE: 1.86 (95% CI 0-3.72) 10
Abbreviations: CBCT, cone-beam computed tomography; CI, confidence interval; CIA, common iliac artery; CTA, computed tomography angiography; DAP, dose area product (Gycm2); EIA, external iliac artery; FT, fluoroscopy time (min); IC, iodinated contrast (mL); IIA, internal iliac artery; MRA, magnetic resonance angiography; PT, procedure time (min); PTA, percutaneous transluminal angioplasty; RE, registration error (mm); RT, registration time (min); SFA, superficial femoral artery;
aVariables are given as mean (range) or 95% confidence interval (CI).
DISCUSSION
Our study is the first study to validate the reliability of 3D image fusion technology for interventions of iliac artery obstructions. The visual errors between the fusion images and DSAs were scored <5 mm in 9 of 11 patients. The mean absolute registration error for all iliac trajectories was small (4.0 ± 2.5 mm). On average, the fusion image was registered slightly more medial (–0.6 ± 3.7 mm) and caudal (–0.2 ± 3.0 mm) with respect to the DSA run. The average displacement decreased to 2.5 mm with increasing experience with the technique. The absolute registration error after post-procedural translation and rotation between iliac trajectories without and with catheters was small (0.9 ± 0.6 mm vs. 1.0 ± 0.6 mm), and the difference was not significant (P = .447). Hence, there was no significant influence of the catheters or guidewires on the accuracy of the image fusion technique.
Image fusion accuracy can be influenced by the image quality of the preoperative CTA or CE-MRA. In the current study, a regular preprocedural CE-MRA was used for the image fusion technique, which was of sufficient quality to perform the interventions under fusion guidance. However, fusion imaging has some limitations, such as stent artifacts at the location of the lesion and the occurrence of long-segment occlusions,
which will jeopardize a sufficient roadmap.
Furthermore, differences in acquisition between the CE-MRA and CBCT, such as the patient’s arm position, should be minimized, and it is crucial that the patient does not move throughout the procedure, especially after the image registration is performed. To avoid movement artifacts, patients were placed on a vacuum mattress to minimize leg and pelvic movement, but this can be uncomfortable in extensive procedures. The vacuum mattress was used as a precaution, and the extent of movement artifacts with the use of a regular mattress was not investigated.
A limitation to the technique is that it relies on the use of a rigid registration, which is why differences in acquisition between the CE-MRA and CBCT may result in displacement errors. This study demonstrates that accurate image fusion can be achieved, even though registration of CE-MRA to a CBCT is more challenging compared with CTA because CE-MRA does not depict vessel calcifications. The technique of image fusion may be especially beneficial when the operator is accustomed to using multimodal image guidance during interventional procedures. Nevertheless, the additional radiation dose of the CBCT should be considered. Using 2 fluoroscopic orthogonal shots for the image registration instead of a CBCT is described to be a fast and easy technique. It was estimated that the effective dose can be reduced from 1.53/1.66 mSv for a CBCT and from 0.14-0.20 mSv for the 2 fluoroscopic orthogonal shots.11 The 3D-3D image fusion technique was more precise than the 2D-3D registration during EVAR procedures.12 However, the 2D-3D image fusion technique has not yet been investigated in peripheral artery interventions.
Accuracy assessment of the image fusion technique with respect to the initial DSA was evaluated in 20 iliac trajectories. On average, the medial-lateral and cranial-caudal displacement was small, but we observed that the absolute displacement error increased to the periphery. Registration was first performed based on the lumbar vertebrae, and second, the vessel lumens of the CE-MRA were compared to the vessel wall calcifications of the CBCT. The increasing error to the periphery could be a result of the applied image registration process. Small registration errors at the site of the aortic bifurcation will be amplified more peripherally, hence resulting in larger displacement errors.
We recommend that the image registration be performed using landmarks as close as possible to the artery of interest, for example, first performing the registration based on the vessel outline of the CE-MRA and the calcifications of the CBCT. The postprocedure analysis demonstrated that adjusting the fusion images resulted in an even more accurate overlay. Therefore, visually scoring the fusion quality after contrast runs and adjusting the overlay when needed is important.
Inserting stiff guidewires can result in vessel deformation during EVAR procedures,
whereby discrepancies between the fused images and the actual vasculature can arise.4,5 Simulations by Kaladji et al.4 showed that the largest displacement (10.2 ± 3.3 mm) between preoperative fused images and the DSA during EVAR procedures was located at the iliac bifurcations. Of note, the guidewires and catheters of the current interventions were flexible guidewires, and no rigid delivery systems were introduced. No significant difference was found between registration accuracy in catheterized and non-catheterized sites. The current analysis was performed before the stents were inserted. Stent insertion can deform the arteries, and the fusion images may be less accurate as a result. In addition, the registration accuracy can be influenced due to elongation of (extremely) tortuous arteries.
The average displacement error decreased with increased experience of the operator performing the registration. Because of the presence of a learning curve, image fusion performed by trained personal may decrease the registration error to an average of 0 to 2 mm. However, an automatic alignment algorithm that would evaluate the image fusion quality after every 2D contrast run could improve the image fusion technique. In addition, an automatic registration method would increase the usability, because it would reduce the registration time and could eliminate the need for an extra person to perform the registration. An application that allows the interventionalist to perform the registration at the tableside may also increase usability. New systems are equipped with the latter, and automatic registration methods are under development. However, until this software is available, it is not preferred that the interventionalist performs the registration himself or herself because this will add 5 to 10 minutes to the procedural time.
The systematic review showed that the current literature is of poor scientific quality and lacks a randomized controlled trial that investigates the benefits of the image fusion technique in endovascular iliac artery interventions. A randomized controlled multicenter trial, “Fluoroscopy with MRA fusion image guidance in endovascular iliac artery interventions” (NL47680.068.14), has been set up to investigate whether the addition of the 3D image fusion technique leads to a reduced administered volume of iodinated contrast. The trial will include 106 patients with the following inclusion criteria: (1) peripheral arterial disease of the common and/or external iliac artery for which an endovascular intervention (PTA or recanalization) is indicated, and (2) availability of a diagnostic CE-MRA of the lower extremities without significant artifacts that is not older than 6 months. The patients are randomized to the control group or an image fusion group. The primary endpoint is the amount of administered iodinated contrast agent. Secondary endpoints are technical success, procedure time, fluoroscopy time, radiation dose, and cost-effectiveness. The enrollment of this study is currently ongoing.
Limitation of the current postprocedural analysis
After the 3D-3D registration is performed, the fusion images are projected on 2D fluoroscopy images. Therefore, the analysis could only be performed in 2 directions. The anteroposterior displacement error could not be determined, but may well be of influence on the overall accuracy of the fusion images. Irregularities in the centerline resulting from bifurcations, stent artifacts, or occlusions might also influence the analysis. In addition, the femoral bifurcation was not always present in the fusion images, which could have induced an error in normalizing the data. However, the error caused by the latter is expected to be negligibly small. Moreover, pairing the exact same points on both centerlines is of great importance. A mismatch between the points could arise if the centerlines have a different length, which will result in an overestimation of the segmentation inaccuracy more peripherally.
Limitations of the current study
Limitations of the current study are the small number of patients and the lack of a control group for major end points such as volume of iodinated contrast, radiation dose, fluoroscopy time, and procedure time. In addition, no tortuosity indices of the common and external iliac arteries were calculated for this small patient group. The insertion of guidewires and catheters would be expected to result in a larger deformation in more tortuous arteries.
CONCLUSION
Accurate image fusion based on preprocedural CE-MRA is possible and could potentially be of help in iliac PTA procedures. Flexible guidewires and catheters, routinely used during endovascular procedures of iliac arteries, did not cause significant displacement that influenced the image fusion. Current literature on 3D image fusion in iliac PTA procedures is of limited methodological quality.
REFERENCES
1. Norgren L, Hiatt WR, Dormandy JA, Nehler MR, Harris KA, Fowkes FGR. Inter-Society Consensus for the Management of Peripheral Arterial Disease ( TASC II). J Vasc Surg 2007;45:5-67.
2. Bekken J, Jongsma H, Ayez N, Hoogewerf CJ, van Weel V, Fioole B. Angioplasty versus stenting for iliac artery lesions (Review). Cochrane Database Syst Rev 2015;(5):1-29.
3. Goudeketting SR, Heinen SGH, van den Heuvel DAF, de Vries JPPM, van Strijen MJ, Sailer AM. Pros and Cons of 3D Image Fusion in Endovascular Aortic Repair : A Systematic Review and Metaanalysis. J Endovasc Ther 2017:1-9.
4. Kaladji A, Dumenil A, Castro M, Cardon A, Bequemin JP, Bou-Saïd B, et al. Prediction of deformations during endovascular aortic aneurysm repair using finite element simulation. Comput Med Imaging Graph 2013;37:142-9.
5. Schulz CJ, Schmitt M, Böckler D, Geisbüsch P. Fusion Imaging to Support Endovascular Aneurysm Repair Using 3D-3D Registration. J Endovasc Ther 2016;23:791-9.
6. Sailer AM, de Haan MW, de Graaf R, van Zwam WH, Schurink GWH, Nelemans PJ, et al. Fusion Guidance in Endovascular Peripheral Artery Interventions: A Feasibility Study. Cardiovasc Intervent Radiol 2014;38:314-21.
7. Ierardi AM, Duka E, Radaelli A, Rivolta N, Piffaretti G, Carrafiello G. Fusion of CT Angiography or MR Angiography with Unenhanced CBCT and Fluoroscopy Guidance in Endovascular Treatments of Aorto-Iliac Steno-Occlusion: Technical Note on a Preliminary Experience. Cardiovasc Intervent Radiol 2016;39:111-6.
8. Slim K, Nini E, Forestier D, Kwiatkowski F, Panis Y, Chipponi J. Methodological Index for Non-Randomized Studies (Minors): Development and Validation of a New Instrument. ANZ J Surg. 2003;73:712-6. 9. Ahmad W, Gawenda M, Brunkwall S, Shahverdyan R, Brunkwall JS. Endovascular aorto-iliac aneurysm
repair with fenestrated stent graft and iliac side branch using image fusion without iodinated contrast medium. Ann Vasc Surg 2016;33:231.e5–231.e9.
10. Sadek M, Berland TL, Maldonado TS, Rockman CB, Mussa FF, Adelman MA, et al. Use of preoperative magnetic resonance angiography and the Artis zeego fusion program to minimize contrast during endovascular repair of an iliac artery aneurysm. Ann Vasc Surg 2014;28:261.e1-5.
11. van den Berg JC. Three-Dimensional Image Overlay to Assist Endovascular Procedures. Vasc Dis Manag 2013;10:E179-E184.
12. Stangenberg L, Shuja F, Carelsen B, Elenbaas T, Wyers MC, Schermerhorn ML. A novel tool for three-dimensional roadmapping reduces radiation exposure and contrast agent dose in complex endovascular interventions. J Vasc Surg 2015;62:448-55.
APPENDIX I – A CROSS-CORRELATION METHOD
To perform a fair analysis between the corresponding points at the DSA and CE-MRA a cross-correlation method was used to pair points along the reconstructed centerlines. The cross-correlation method was implemented by first calculating the radial profile of the segmented vessels. The points on the radial profile are directly linked to the points on the centerline. Because the radial profiles of the CE-MRA and DSA exhibit the same profile, they can be used to select corresponding points on the centerline. The radial profiles were shifted along each other, and the root mean squared (RMS) value was calculated at every position by the equation:
eq.1
Here N is the total number of sample points along the radial profile, and X is the difference in the radius between the DSA and CE-MRA at a given point (n) on the line. The minimum RMS indicated the optimal match between both radial profiles. When deemed necessary, the radial profiles could be manually shifted to improve the match. The longest radial profile was then truncated to the length of the shortest profile. Because the points on the radial profile were linked to the points on the centerline, corresponding points on both centerlines were identified and used to calculate the lateral-medial, cranial-caudal, and total displacement.
