Motion Correction of Dual Volume Reconstruction of Three‑dimensional Digital Subtraction Angiography for Follow‑up Evaluation of Intracranial Coiled Aneurysms
Keywords: 3D, aneurysm, dual-volume, motion correction, reconstructionKey Messages: Follow-up image modality of coiled aneurysms is a critical issue. Although extensively replaced by MRA, catheter angiography is still the standard for evaluating the recurrence of coiled aneurysms. Dual volume-3D-VRT with motion correction is a useful complementary modality to evaluate recanalization of coiled aneurysms.
After the successful International Subarachnoid Aneurysm Trial (ISAT), coiling has become a mainstay in intracranial aneurysm treatment. However, coiling poses a greater risk of recanalization and re-bleeding than clipping. Given this, follow-up image modality after coiling of an aneurysm has become a critical and debatable issue in the neurointerventional field.
Much research on magnetic resonance angiography (MRA) has found it to be a useful follow-up tool for coiled aneurysms.,,,, However, conventional catheter angiography is still used because of its advantages in some situations, and it has long been considered the gold standard.,,,,
In catheter angiography, 3D-volume-rendering technique (3D-VRT) images with rotational angiography (RA) provide a good view of spatial relations in a coiled aneurysm and help neurointerventionists increase the detection rate of aneurysm recurrence.,,, The single-volume 3DVRT (SV-3D-VRT) with a single rotation of the C-arm is sufficient when evaluating the pre-coiling state and determining the best working view during coiling for an aneurysm.
However, when performing follow-up examinations of coiled aneurysms, differentiating between vessels, coils, clips, and stents is crucial in evaluating a residual or recanalized sac. The DV-3D-VRT with two rotations of the C-arm can differentiate a remnant or recanalized sac from coil struts and provides a more precise recurrence evaluation. However, we found that DV-3D-VRT images of subtracted 3D-RA were quite vulnerable to motion artifacts.
In studies of time of flight MRA, research has been performed which has suggested that applying a motion correction technique can reduce the incidence of motion artifacts and improve image quality. However, we found no extant research on motion correction in DV-3D-VRT images for follow-up DSAs of coiled aneurysms. Hence, we applied the motion correction technique to DV-3D-VRT images and evaluated the usefulness of this technique in the detection of recurrence after coil embolization.
This was a retrospective study approved by our institutional review board (IRB, number: 2014-11-005). The requirement for written, informed consent from patients was waived.
From October 2014 to April 2015, 70 patients underwent consecutive follow-up digital subtraction angiography (DSA) examinations including 3D-RA with DV-3D-VRT after coiling of intracranial aneurysms. For this study, we chose 59 of those 70 patients, constituting a total data set representing 64 coiled aneurysms, who had undergone both TOF MRAs and the DSA. The remaining 11 patients who had only DSA except MRA were excluded from this study.
In the 59 patients with 64 coiled aneurysms (one patient: three-coiled aneurysms; three patients: each with two coiled aneurysms), 16 patients were men and 43 were women. The patients' ages when they received coiling ranged from 26 to 74 years (mean age: 58.6 years old). Among the 64 coiled aneurysms, 53 underwent a first follow-up DSA after coiling, 10 underwent a second follow-up DSA, and one underwent a third follow-up DSA. Follow-up periods after coiling were 2 to 67 months, and the mean follow-up period was 15 months. Among the 64 coiled aneurysms, 3 were located at the anterior cerebral arteries, 13 at the anterior communicating arteries, 5 at the middle cerebral arteries, 4 at the basilar top, 6 at the posterior communicating arteries, and 33 at the distal internal carotid arteries (ICA). Eighteen patients had ruptured aneurysms and 46 had unruptured aneurysms. Among the 64 aneurysms, 34 were coiled with stent assistance [Table 1].
All DSAs with 3D-RA were performed using a biplane angiography system of Artis-zee (Siemens, Germany) with conscious sedation. 2D-DSAs with anterior-posterior (AP) and lateral projection and working projection were obtained after selection of the ICA or vertebral artery (VA) and injection of contrast medium (Visipaque 320, GE Healthcare, Ireland) (ICA: total volume of 6 ml at a flow rate of 4 ml/s, VA: total volume of 12 ml at a flow rate of 3 ml/s). Subtracted 3D-RAs were obtained using the frontal plane of biplane C-arms. This is performed through the first C-arm rotation in the range of 200° (–100° to 100°), after which the rotation returns to the starting position and the contrast is injected (total volume of 18-21 ml at a flow rate of 2.5-3 ml/s) into the selected ICA or VA. Two seconds later the second rotation started again and which ensured that the arteries would be completely filled during the C-arm rotation. After the RA procedure, the source data were transferred to a workstation (Siemens Syngo-Inspace) where the MC(+) DV-3D-VRT and MC(-) DV-3D-VRT images were acquired.
The principle of motion correction technique in Syngo-Inspace workstation of Artis Zee biplane system is fitting the differences between the mask run of the first rotation image and fill run of second rotation image through flexible pixel shift [Figure 1]. In other words, the C-arm is rotated twice, and the system corrects the pixel shift between the mask (baseline) run of the first rotation images and the fill run of the second rotation images.
TOF MRAs were performed using the Skyra 3.0T system (Siemens, Germany), Achieva 3.0T or Achieva 1.5T (Philips, Netherlands). MRAs in Skyra 3.0T were obtained using 3D fast low-angle shot (FLASH) sequences with TR/TE 22/3.9, flip angle 18°, field-of-view (FOV) 250 × 160, matri × 512 × 288, and an acquisition time of 5 minutes and 49 seconds. With an Achieva 3.0T, MRAs were obtained using 3D T1-weighted fast field echo (3D-T1-FFE) sequences with TR/TE 29/3.9, flip angle 208°, FOV 210 × 190, matri × 600 × 272, and an acquisition time of 5 minutes and 45 seconds. MRAs in an Achieva 1.5T were obtained using 3D-T1-FFE sequences with TR/TE 25/6.9, flip angle 208°, FOV 200 × 200, matri × 500 × 250, and an acquisition time of 6 minutes and 33 seconds.
Image analysis and statistics
Two radiologists independently reviewed the MC(+) DV-3D-VRT and MC(-) DV-3D-VRT images to determine recanalization status of the coiled aneurysm in the workstation (Siemens Syngo-Inspace). Following the initial independent reviews, any disagreement between the two radiologists was resolved through consensus to reach final conclusions.
The recanalization status of the MC(+) DV-3D-VRT and MC(-) DV-3D-VRT images were compared to the reference assessments. In the comparison, the reference assessments were made comprehensively using DSA 2D images (AP andlateral view and working projection) and TOF MRA with source images in PCAS system [Figure 2]. The McNemar test was used to evaluate the recanalization status which is higher agreement with the reference assessments. The recanalization status of coiled aneurysms was divided into four categories, i.e. stable (no flow or no change of the minimal remnant neck), minor recurrence (contrast filling of the aneurysm neck), major recurrence (contrast filling of the aneurysmal sac), or undetermined recurrence. If the image quality was inappropriate for evaluation, we considered it as a “undetermined recurrence.”
Inter-observer agreement between the two radiologists was evaluated after the review of the recanalization status. The kappa statistic of 0.75–1.0 was considered an excellent agreement, 0.4-0.74 as good agreement, and 0-0.39 as poor agreement. A P value less than 0.05 was considered statistically significant. All statistical analyses were performed using SAS software (version 8.0; SAS Institute, Cary, NC, USA).
Among the MC(-) DV-3D-VRT images of the 64 coiled aneurysms, 24 aneurysms were identified as stable. Nineteen aneurysms exhibited evidence of minor recurrence, 14 exhibited major recurrence, and 7 exhibited undetermined recanalization. In the MC(+) DV-3D-VRT images of the 64 coiled aneurysms, 36 were identified as stable. Twenty aneurysms exhibited evidence of minor recurrence, seven exhibited major recurrence, and one exhibited undetermined recanalization [Figure 3] and [Figure 4]. Among the reference assessments, which were made comprehensively with DSA-2D images and TOF MRA with source images, 47 coiled aneurysms were identified as stable, 13 exhibited evidence of minor recurrence, and 4 exhibited major recurrence [Table 2].
In the review of MC(-) DV-3D-VRT images, inter-observer agreement between the two radiologists was good (ĸ = 0.69). In addition, the inter-observer agreement was excellent (ĸ = 0.78) regarding the review of MC(+) DV-3D-VRT images.
Among the 64 aneurysms reviewed in the MC(-) DV-3D-VRT images, 23 aneurysms were changed grade of recanalization in the MC(+) DV-3D-VRT image (36%). The tendency for changes from MC(-) DV-3D-VRT to MC(+) DV-3D-VRT images is characterized by a decreased exaggeration of recanalization.
When evaluating the agreement rate with reference assessments, 33 coiled aneurysms showed agreement with the reference assessments (51.6% agreement) in the MC(-) DV-3D-VRT images. In the MC(+) DV-3D-VRT images, 50 coiled aneurysms showed agreement with the reference assessments (78.1% agreement). The MC(+) DV-3D-VRT images revealed a higher agreement rate (P = 0.035, McNemar test).
As extant research describes a significant rate of recurrence after the coiling of intracranial aneurysms,, various modalities including simple radiography, computed tomography angiography, MRA, and catheter angiography have been used to detect recurrence. Among these modalities, MRA and catheter angiography are the most commonly used techniques. MRA has two critical advantages over the other methods. First, it is noninvasive, radiation-free, and safer than catheter angiography. Second, TOF MRA does not require the use of a contrast agent. Many reports have mentioned that MRA is adequate and can replace the DSA in follow-ups for coiled intracranial aneurysms.,,,, However, conventional catheter angiography is still commonly used as follow-up modality after coiling, as it can show blood flow characterization and has a good temporal and spatial resolution,,,,, and catheter angiography can serve 3D-RA. The evaluation of raw data of 3D-RA in 2D multiple planes can give good information of coil compaction or neck remnant. In particular, in the case of stent-assisted coiled aneurysm, catheter angiography should be performed at least once during the follow-up period to evaluate in-stent stenosis.,
In catheter angiography, 3D VRT images with RA have proven useful in evaluating aneurysmal occlusions after coiling by allowing neurointerventionists to observe the parent arteries and aneurysms from arbitrary views.,,, In addition, DV-3D-VRT images enable extraction of coil struts from coiled aneurysm. Given its broad range of advantages, the DV-3D-VRT image has become an essential tool in angiographic follow-ups for patients with coiled aneurysms. However, a problem with the DV-3D-VRT image is the occurrence of radiographic blurring, including sample artifacts and motion artifacts that can mimic recanalization.
We noted that MC(+) DV-3D-VRT images reduced motion artifacts and produced clearer images than MC(-) DV-3D-VRT images. We attempted to confirm our observations through the statistical comparison of images obtained with and without the use of the motion correction technique. The coiled aneurysm is usually followed by either MRA or catheter angiography, and it is not common for both MRA and catheter angiography to be performed. However, there are some of our patients who have undergone both MRA and catheter angiography because recurrence cannot be confirmed in only one imaging method. In addition, it enabled us to collect the kind of data necessary to make comprehensive reference assessments. In an appropriate working view, 2D-DSA images represent a useful imaging modality for detecting recanalization because of their high spatial resolution (1024 × 1024 matrix). However, the use of 2D-DSA alone is insufficient because it does not provide a three-dimensional view., The combined interpretation of the proper working view of DSA and TOF MRA with the source image may provide the best detection of recanalization after coiling. Reference assessments were compiled using DSA 2D images and TOF MRA with source images to compare MC(+) and MC(-) DV-3D-VRT images in this study. In the TOF MRA of coiled aneurysm with stent, the stent is known to distort the local magnetic field. Hence, we interpreted source images of TOF MRA in stent cases. The usefulness of the source image in TOF MRA is well-known through research, especially for stent-assisted coiling.,
In this study, the MC(+) DV-3D-VRT images demonstrated a higher agreement coefficient with the reference assessments than MC(-) DV-3D-VRT images (78.1% vs. 51.6%). In addition, inter-observer agreement is also better in MC(+) DV-3D-VRT images (ĸ = 0.78 vs.ĸ = 0.69).
This is a retrospective study, and hence it has some limitations that should be mentioned. First, in spite the strong statistical relationship demonstrated by the results, the sample size was small and obtained from a single clinic. Second, we performed this study using the Siemens biplane angiography system and, as far as we know, no other supplier angiographic system can compare MC(+) and MC(-) DV-3D-VRT images.
In conclusion, MC(+) DV-3D-VRT images demonstrated higher agreement with the reference assessments that consisted of DSA-2D images and TOF MRA with source images, as compared to MC(-) DV-3D-VRT images. Therefore, the use of the motion correction technique seems to improve the accuracy of DV-3D-VRT images in the follow-up of intracranial coiled aneurysms.
All procedures performed in the studies involving human participants were in accordance with the ethical standards of our institutional review board (IRB, number: 2014-11-005) with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.
For this type of study, formal consent is not required.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2]