Utility of Non-Contrast CT and MRA Fusion Images for Depicting Occluded Vessels in Mechanical Thrombectomy

Abstract

Objective

Mechanical thrombectomy (MT) for acute ischemic stroke usually requires blind procedures when endovascular devices are advanced distal to the occlusion site. Visualization of vascular structures distal to the occlusion could potentially enable safer procedures and more effective recanalization. We report the usefulness of fusion images using non-contrast CT and MRA with Ziostation REVORAS (Ziosoft, Tokyo, Japan).

Case Presentation

Fusion images were created for 8 consecutive patients who underwent MT at our institution (Tsuruoka Municipal Shonai Hospital). Non-contrast CT and time-of-flight MRA data were processed using the 3D analysis function of Ziostation REVORAS. Fluoroscopy-like images and vascular structures distal to the occlusion were depicted using non-contrast CT, and vessels proximal to the occlusion site were visualized using MRA. These datasets were used to create fusion images displaying vascular structures distal to the occlusion with bone information. The images enabled simulation of the working angle and visualization of the thrombus extent, and were applicable to both anterior and posterior circulation. The mean image creation time was 3 min 26 s, and image creation was completed within 4 min in all cases. The vascular anatomy depicted on the fusion images was concordant with post-MT DSA in all cases.

Conclusion

Fusion images may represent a useful adjunctive imaging tool for achieving safe and effective recanalization during MT.

Introduction

Mechanical thrombectomy (MT) is an established and effective treatment for acute large vessel occlusion (LVO), and the recent expansion of treatment indications is expected to increase treatment opportunities.^1,2) Vessel perforation is a serious complication of MT and has been shown to be associated with reduced successful recanalization rates and poor prognosis.^3,4) Because most perforations occur due to blind manipulation distal to the occlusion site,³⁾ preoperative understanding of the vascular anatomy distal to the occlusion could potentially enable safer and more effective recanalization. Previous studies have reported the usefulness of MRI variable refocusing flip angle-turbo spin echo (VRFA-TSE)^5–7) and middle cerebral artery (MCA) anatomical scanning MRI (MAS-MRI)⁸⁾ for this purpose. However, these methods cannot be performed on all scanners and are not universally available. In this study, we created vessel images and fluoroscopy-like images from non-contrast CT and MRA using Ziostation REVORAS (Ziosoft, Tokyo, Japan), a medical imaging workstation. We describe cases in which these fusion images were useful for image evaluation during MT procedures.

Case Presentation

At our institution (Tsuruoka Municipal Shonai Hospital), non-contrast CT and MRI are performed as imaging examinations for acute ischemic stroke within 24 h of onset. We created fusion images by editing these data. Plain CT was acquired with an Aquilion ONE scanner (Canon, Tokyo, Japan), using the following imaging parameters: tube voltage, 120 kVp; tube current, 250 mAs; slice thickness, 1 and 4 mm; beam pitch, 0.813; rotation time, 0.75 s. MRI was acquired with an Ingenia Ambition 1.5T unit (Philips, Amsterdam, the Netherlands), and time-of-flight MRA was obtained using the following imaging parameters: repetition time, 24 ms; echo time, 6.9 ms; field of view, 200 × 200 mm; matrix, 332 × 200; slice thickness, 1.1; scan time, 2 min 35 s.

The acquired image data were edited using the 3D analysis function of Ziostation REVORAS (Ziosoft). First, using non-contrast CT (1 mm), fluoroscopy-like images were created from the “Preset/Template” menu (Fig. 1A), and vascular structures distal to the occlusion were depicted by manual plotting using the “Path creation” function in the “Vascular measurement” menu (Fig. 1B and 1E–1G).

Fig. 1. Creation of fusion images in Ziostation REVORAS. (A) Creation of fluoroscopy-like images from non-contrast CT with a slice thickness of 1 mm. (B) Vascular structures distal to the occlusion were depicted by manual plotting using non-contrast CT with a slice thickness of 1 mm (blue line). (C) Vessels proximal to the occlusion site were depicted from MRA. (D) Fusion images were created by combining (A–C), yielding images comparable to those obtained during MT. (E–G) Vascular structures were identifiable on non-contrast CT (arrowhead) and served as a reference for creating (B).

MT, mechanical thrombectomy; Ziostation REVORAS, Ziosoft, Tokyo, Japan

Second, using MRA, the vessels up to the occlusion site were depicted using the “Head MRA” function in the “Automatic processing” menu (Fig. 1C). Non-contrast CT and MRA were synchronized using the “Alignment” function in the “MDF” menu. Fusion of these images enabled visualization of the vascular anatomy distal to the occlusion; furthermore, the fluoroscopy-like images served as a reference when determining the working angle during MT (Fig. 1D). In cases with a hyperdense sign, the hyperdense region could be visualized by the same manual plotting procedure described above, thereby enabling thrombus localization (Fig. 2A). The hyperdense sign was identified based on a hounsfield unit (HU) threshold of >43, as defined by previously published criteria.⁹⁾ The boundary between the vessel and the thrombus was determined qualitatively in principle (Fig. 2B–2G). In cases where the boundary was ambiguous, HU values were measured to assist in the determination.

Fig. 2. Visualization of thrombus on a fusion image. (A) The location and length of the thrombus are visualized by delineating the extent of the thrombus in green and the vessels distal to the thrombus in blue. (B–G) Thrombus in the MCA is identified as a hyperdense sign on non-contrast CT (arrowheads) and then visualized by manual plotting.

MCA, middle cerebral artery

Fusion images were created by the treating physician or 1 of 5 radiological technologists assigned to neuroendovascular procedures, following a standardized protocol. The mean image creation time among all 6 operators was 3 min 26 s, and image creation was completed within the MT preparation period.

Ethics approval for this study was obtained from the Institutional Review Board of Tsuruoka Municipal Shonai Hospital (No. 394).

Case 1

A 79-year-old man was admitted to the emergency department 1 h and 9 min after sudden-onset aphasia and right hemiparesis. His National Institutes of Health Stroke Scale (NIHSS) score was 25. Diffusion-weighted imaging (DWI) showed an acute infarct lesion with a DWI-Alberta Stroke Program Early CT Score (ASPECTS) of 9/11 (Fig. 3A and 3B). MRA showed occlusion of the left MCA M2 anterior branch (Fig. 3C). Electrocardiography revealed atrial fibrillation, and the patient was diagnosed with cardiogenic embolism.

Fig. 3. Case 1 with acute left MCA occlusion. (A, B) DWI shows a lesion with a DWI-ASPECTS score of 9/11. (C) MRA shows a left M2 anterior branch occlusion. (D–G) The occlusion site overlaps the course of the M2 posterior branch on the left side. The fluoroscopy-like images serve as a reference when determining the working angle on the right side. (H–J) The anticipated working angle corresponds to the actual working angle. The vascular course distal to the occlusion on post-MT DSA is consistent with that shown in the preoperative images.

ASPECTS, Alberta Stroke Program Early CT Score; DWI, diffusion-weighted imaging; MCA, middle cerebral artery; MT, mechanical thrombectomy

Preoperative simulation was performed using fusion images created from non-contrast CT and MRA. On the anteroposterior view, the occluded M2 anterior branch overlapped with the course of the M2 posterior branch. Therefore, a working angle was determined that enabled separation of the occluded segment, and its relationship to the fluoroscopy-like images was confirmed (Fig. 3D–3G). Image creation required 3 min 55 s.

During the procedure, the C-arm position was adjusted under fluoroscopy based on the bony landmarks identified at the simulated working angle. The acquired images successfully separated the occluded MCA segment, enabling optimal visualization for treatment (Fig. 3H and 3I). With reference to the fusion images, MT was then performed using EMBOTRAP III 5 × 37 mm (Johnson & Johnson, New Brunswick, NJ, USA) and 6-Fr SOFIA FLOW Plus 125 cm (Terumo, Tokyo, Japan), achieving complete recanalization after 1 pass (modified treatment in cerebral infarction [mTICI] 3). Postoperative DSA showed vascular anatomy concordant with the preoperative fusion images (Fig. 3J). The puncture-to-reperfusion time (PRT) was 50 min. The patient was discharged with a modified Rankin Scale (mRS) score of 2 without perioperative complications.

Case 2

An 84-year-old man who had been hospitalized for heart failure developed disturbance of consciousness and right pupil dilation. His NIHSS score was 34. DWI showed an acute infarct lesion with a posterior circulation ASPECTS score of 8/10 (Fig. 4A–4C). MRA showed occlusion of the basilar artery (BA) top (Fig. 4D). Electrocardiography revealed atrial fibrillation, and the patient was diagnosed with cardiogenic embolism.

Fig. 4. Case 2 with acute BA top occlusion. (A–C) DWI shows a lesion with a posterior circulation DWI-ASPECTS score of 8/10. (D) MRA shows occlusion of the BA top. (E, F) The BA deviates to the right just proximal to the occlusion, suggesting displacement of the thrombus toward the right PCA P1 segment. (G, H) The vascular course distal to the occlusion on post-MT DSA is consistent with that shown in the preoperative images.

ASPECTS, Alberta Stroke Program Early CT Score; BA, basilar artery; DWI, diffusion-weighted imaging; MT, mechanical thrombectomy; PCA, posterior cerebral artery

Preoperative simulation was performed using fusion images in a similar manner to Case 1. Bilateral P1 segments were well developed, and intraoperative adjustment of microcatheter navigation was deemed feasible as needed. Image creation required 3 min 20 s (Fig. 4E and 4F).

Using the fusion images for reference (Fig. 4G), a CEREGLIDE 71 125-cm catheter (Johnson & Johnson) was advanced to directly face the thrombus, and MT was performed using a direct aspiration first-pass technique, achieving complete recanalization after 2 passes (mTICI 3). Post-MT DSA showed vascular anatomy concordant with the preoperative fusion images (Fig. 4H). PRT was 27 min. The patient was discharged with an mRS score of 3 without perioperative complications.

Table 1 summarizes the 8 consecutive cases in which fusion images were created and MT was attempted. During the study period, fusion images were created for all consecutive patients who underwent MT. The median age was 83.5 years. Stroke etiology included cardioembolism in 7 cases and large artery atherosclerosis in 1 case. Occlusion sites included the internal carotid artery in 1 case, MCA in 5 cases (2 of which were medium vessel occlusion [MeVO]), and BA in 2 cases. Mean image creation time was 3 min 26 s overall. By occlusion site, mean image creation time was 3 min 24 s for the anterior circulation (3 min 38 s for MeVO) and 3 min 32 s for the posterior circulation. Image creation was completed within 4 min in all cases, and no substantial difference in image creation difficulty was observed according to the occlusion site.

Table 1. Patient characteristics and fusion image creation data.

Case	Age	Sex	Occlusion site	Diagnosis	Vessel depiction	Hyperdense sign	Thrombus depiction	Creation time	Concordance*
1	83	F	Rt. ICA top	CE	+	+	+	3 min 16 s	+
2	95	F	Rt. M1	CE	+	+	+	3 min 15 s	+
3	84	M	BA top	CE	+	+	+	3 min 20 s	+
4	79	M	Lt. M2 (MeVO)	CE	+	+	+	3 min 55 s	+
5	74	M	BA mid	LAA	+	–	–	3 min 45 s	+
6	84	M	Lt. M1	CE	+	–	–	3 min 25 s	+
7	76	F	Rt. M2 (MeVO)	CE	+	+	+	3 min 22 s	+
8	94	F	Lt. M1	CE	+	+	+	3 min 16 s	+

*Concordance was defined as an agreement between the vascular anatomy depicted on fusion images and that observed on post-MT DSA.

BA, basilar artery; CE, cardioembolism; F, female; ICA, internal carotid artery; LAA, large artery atherosclerosis; Lt., left; M, male; MeVO, medium vessel occlusion; MT, mechanical thrombectomy; Rt., right

Discussion

Vascular perforation is a serious complication of MT. Although its incidence is relatively rare, perforation leads not only to poor functional outcomes but also directly to mortality.^3,4) The incidence is 1.0% in LVO but increases to 2.4% in MeVO, with higher rates observed in more distal vessels.³⁾ Vascular perforation occurs during microcatheter navigation distal to the occlusion in 30.7% of cases and during retraction of stent retrievers/aspiration catheters in 35.5% of cases.³⁾ Because blind manipulation distal to the occlusion is a major risk, preoperative understanding of the vascular anatomy distal to the occlusion may contribute to risk reduction.

Several MRI-based methods have been reported for visualizing vascular structures distal to the occlusion, including T2-weighted imaging and proton density-weighted imaging using the VRFA-TSE method,^5–7) and MAS-MRI, which is an application of basi-parallel anatomical scanning.⁸⁾ Although these methods are useful for depicting vascular structures distal to the occlusion, imaging can be challenging depending on the scanner model. Therefore, we created fusion images that display reconstructed vascular images on fluoroscopy-like images using the 3D analysis function of Ziostation REVORAS. This fusion image approach has 4 main advantages. First, it allows simultaneous visualization of the occluded segment, vascular structures distal to the occlusion, and bony landmarks. Because the native view of DSA displays only bony landmarks, preoperative simulation using fusion images to determine the working angle enables rapid positioning of the C-arm based on these landmarks. Second, the fusion images can be created using only non-contrast CT and MRA, eliminating the need for specialized additional imaging. The creation of fusion images requires approximately 5 min and can be completed during MT preparation, allowing treatment to begin within the conventional timeframe. Third, thrombus visualization is possible in most cases (Fig. 2A). In situations where the thrombus is visible as a hyperdense sign on non-contrast CT, it can be visualized through manual plotting using the same procedure used to depict vessels distal to the occlusion (Fig. 2B–2G). Thrombus visualization assists in appropriate device navigation, particularly when a long-segment thrombus extends from major arteries into branches. Fourth, the same procedure can be applied to both anterior and posterior circulation occlusions. Previously reported methods have primarily targeted MCA occlusions and require modification of the imaging parameters, such as changing the imaging range, to accommodate posterior circulation occlusions. Vascular visualization distal to the occlusion using non-contrast CT is applicable to the anterior circulation other than the MCA and to the posterior circulation, demonstrating high versatility.

Lee et al. reported that vascular images obtained from non-contrast CT were comparable to those obtained by postoperative DSA,⁹⁾ suggesting that non-contrast CT is sufficient for visualizing vascular structures distal to the occlusion. In addition, whereas previously reported MRI-based methods had difficulty depicting vessels beyond M3 due to time constraints, CT can visualize M3–4 segments, making it potentially useful for preprocedural simulation even in MeVO cases. In the present series, manual plotting on non-contrast CT enabled fusion image creation in all cases, including those with MeVO and posterior circulation occlusions. This likely reflected the fact that most patients were older and had cortical atrophy with consequent widening of the subarachnoid space, which facilitated identification of even distal vessels. With respect to creation time, MeVO and posterior circulation cases required only approximately 20 s longer than anterior circulation cases, a difference that is unlikely to be clinically significant.

Fusion images were created by the treating physician or 1 of 5 radiological technologists assigned to neuroendovascular procedures. Because MT requires rapid treatment initiation, only a single fusion image was created per case, and inter-operator image consistency could not be evaluated. However, image creation times were comparable among operators, and the semi-automated nature of the process may be considered advantageous in that it minimizes the influence of operator experience.

This report has several limitations. The first limitation is that visualization of vascular structures distal to the occlusion on non-contrast CT relies on manual plotting, introducing an element of subjectivity that cannot be entirely eliminated. In particular, identification of vascular structures is challenging in patients with a narrow Sylvian fissure. Although no young patients were included in the present series, pilot attempts in healthy individuals aged 20–39 years revealed that vessels beyond M1 were difficult to identify and that depiction of vessels beyond M2 was not achievable. The second limitation is that the fusion images of fluoroscopy-like images and vessels may not completely correspond to actual intraoperative findings. Although the non-contrast CT and MRA images are aligned before fusion, subtle misalignment can occur due to automated processing. The third limitation is that the external validity of our findings may not be generalizable beyond our institution, given the single-center, retrospective, observational design of this study. Further accumulation of cases is needed to validate the reproducibility and usefulness of this method.

Conclusion

Fusion images using non-contrast CT and MRA enable visualization of vascular structures distal to the occlusion and, with the incorporation of bony landmarks, serve as a reference for intraoperative positioning. These images may represent a useful adjunctive imaging tool for achieving safe and effective recanalization.

Disclosure Statement

The authors declare that they have no conflicts of interest.