A Multidimensional Pre-operative Planning Method of Unruptured Vertebral Artery Dissecting Aneurysms Using Three-Dimensional AWE Mapping and Hemodynamic Simulation

Objective: High-resolution magnetic resonance imaging (HR-MRI) can provide valuable insights into the evaluation of vascular pathological conditions, and 3D digital subtraction angiography (3D-DSA) offers clear visualization of the vascular morphology and hemodynamics. This study aimed to investigate the potential of a multimodal method to treat unruptured vertebral artery dissection aneurysms (u-VADAs) by fusing image data from HR-MRI and 3D-DSA. Methods: This observational study enrolled 5 patients diagnosed with u-VADAs, who were scheduled for interventional treatment. The image data of HR-MRI and 3D-DSA were merged by geometry software, resulting in a multimodal model. Quantified values of aneurysm wall enhancement (AWE), wall shear stress (WSS), neck velocity, inflow volume


Introduction
Vertebral artery dissecting aneurysms (VADAs) are commonly observed in adults aged 40-50 years, constituting approximately 3% of intracranial aneurysms.VADAs primarily manifest as subarachnoid hemorrhage and ischemic events [1,2].While the overall prognosis for patients with VADAs is favorable, the prognosis for ruptured VADAs is grim.Compared to aneurysmal subarachnoid hemorrhage, ruptured VADAs have notably higher rates of rebleeding and mortality [3][4][5].Yamada et al [3].reported on 24 cases of conservatively managed ruptured VADAs, with 58% experiencing recurrent bleeding and 46% succumbing to extensive rebleeding.Hence, before the rupture of VADAs, proactive intervention can mitigate the elevated fatality and disability rates.
While DSA remains the standard for VADAs diagnosis, it is essential to recognize that conventional DSA images solely provide morphological insight into VADAs.The details such as intimal tear, lesions vessel of dissected aneurysms cannot be determined precisely from DSA images alone.High-resolution magnetic resonance imaging (HR-MRI) provides the capacity to comprehensively observe the intracranial aneurysm wall and identify other biological characteristics, including atherosclerosis, inflammation, and thrombosis [6].The identification of aneurysm wall enhancement (AWE) serves as a pivotal indicator of aneurysmal instability, including rupture and recurrence [7].Computational fluid dynamics (CFD) enables simulating the blood flow dynamics of aneurysms and plays a crucial role in understanding the mechanism of aneurysmal pathology.Wall shear stress (WSS), neck velocity, and intra-aneurysmal velocity, when exhibiting abnormalities, are closely linked to the growth, rupture, and recurrence of aneurysms [8][9][10][11][12].Recently, Tian et al. reported that it was possible to predict the therapeutic outcomes of aneurysm treatments by combining hemodynamic parameters with virtual stent implantation strategies [13].
For the first time, we applied a multidimensional method by merging the image data of 3D-DSA with HR-MRI to provide more information for unruptured VADAs (u-VADAs).This method involved: (1) mapping the wall enhancement onto the 3D model to precisely identify the lesion area, (2) comparing hemodynamic simulation results of different surgical plannings, and (3) making a treatment strategy with reference to hemodynamic data.Finally, the potential of this method in treating VADAs was visualized by the results of VADAs during follow-up.This study adhered to the principles of the Declaration of Helsinki.All patients and their families were informed about the treatment and provided informed consent, including the signing of the surgical informed consent form.

HR-MRI Protocol
HR-MRI were performed with a 1.5-Tesla scanner (Siemens Healthineers, Model: Avanto I-Class, Germany) on all patients.The DWI scan was conducted in the axial plane with a repetition time/echo time of 3400/102 ms.High-resolution sequences included post-contrast 3D T1-weighted images with a voxel size of 0.7×0.7×0.7mm 3 .Six minutes after the injection of 0.1mmol/kg gadopentetate dimeglumine (Gd, Magnevist; Bayer Schering Pharma AG), post-contrast T1-weighted imaging was performed.

Geometrical Reconstruction
To solve the problem of inconsistent spatial coordinates between DSA and HR-MRI images, we reconstructed blood vessels based on two different images.The rotated 3D-DSA images were segmented using image processing software (Mimics v15.0,Materialize, Leuven, Belgium) to obtain the DSA model.
The reconstruction of the DSA model was based on the vascular lumen dimensions from 2D-DSA images, ensuring that the dimensions of the DSA model matched those of the angiographic images.On the other hand, the HR-MRI images were segmented and reconstructed using ITK-SNAP (http://www.itksnap.org).This process involved both automated reconstruction and further refinement through manual adjustments.
For areas missing or inaccurately delineated during the automated reconstruction, additions or deletions were performed to generate the final segmentation result (MRI model, Figure 1A).The dimensional difference between the reconstructed models was less than 0.46mm, which is approximately equivalent to the size of a DSA pixel.The segmentation outcomes for both models were exported in STL format files.

Matching Model of DSA and HR-MRI
This method has been thoroughly described by Jiang [14].A concise reiteration of this method is presented: As shown in Figure 1C, modifications to the DSA model were executed using Geomagic Studio 12.0 (3D Systems, Rock Hill, SC, United States).The coordinates of the MRI model remained unaltered, and an optimal alignment was employed to synchronize the DSA model with the MRI model.Subsequently, fine adjustments including smoothing, trimming, and other refinements were performed on the DSA model.Ultimately, the boundary distance between the DSA model and the MRI model did not exceed 0.6875 mm (Figure 1C), which was less than the pixel size of the HR-MRI images.

Semi-automated Intimal Tear Localization and Quantified SI
In arterial dissections, blood can enter the arterial wall through an intimal tear, and if the intramural hematoma extends to the outer layer, it can lead to the formation of a pseudoaneurysm [15,16].Therefore, in this paper, we define the portion with the intimal tear as a "tear".In HR-MRI, a "tear" often appears as a double-lumen, intimal flap image.
This study formulated customized scripts using the Visualization Toolkit (VTK-8.Considering that the DSA model mirrors the lumen, we shifted the DSA vessel surface outward by one-pixel size (0.6875 mm).Through a VTK script, The HR-MRI and DSA models were processed simultaneously to extract the aneurysm wall's SI.The SI localization was standardized against the corpus callosum (CC), producing the CC ratio (CCratio) which has exhibited superior normalization reliability, resulting in a multimodal model (Figure 1E).A CCratio surpassing>1 signifies noteworthy AWE, while an average CCratio (AVCCratio) >1 indicates General AWE (GAWE).The maximum CCratio (MCCratio) gauges the degree of localized AWE.AVCCratio, MCCratio, and the proportion of enhanced area with CCratio > 1 relative to the total aneurysm area (EACCratio) were computed.

CCratio = Aneurysm intensity /Callosum intensity
Callosum intensity is the average SI of the corpus callosum, while Aneurysm intensity is the SI of the aneurysm wall surface.

J o u r n a l P r e -p r o o f
This script accurately localized the frame depicting the intimal tear in HR-MRI onto the DSA model, as depicted in Figure 1F.The intimal tear of the vessel or the AWE area resulting from the intimal tear was defined as the proximal lesions area of VADAs.The distal lesions area was determined based on VADAs' morphology or regions of AWE.We measured the lesion's lengths, as well as the proximal and distal diameters, while pinpointing the locations of intimal tears in VADAs on the DSA model referring to the distribution of AWE as shown in Figure 2.

Hemodynamic Simulation
We assessed the dimensions of VADAs, including their length and width by the DSA as shown in Figure 2A.Additionally, along the direction of blood flow, the proximal position was defined as the location where AWE begins or the site of the tear, while the distal position was defined as where the VADAs' morphology ends or where AWE terminates.We measured the lesion lengths, the length of the vessel between the distal and proximal ends, as well as the proximal and distal diameters of the vessel as shown in Figure 2B.
According to the lesion length, proximal diameter, and distal diameter of the vessel, different treatment strategies were proposed for each case as shown in Figure 3A-B.The process of virtual implantation followed the methods proposed in a previous study [17].An entirely unstructured mesh was generated using ICEM CFD (version 2019R3, ANSYS Inc., Canonsburg, PA) with a mesh size of 0.25 mm for pre-operative models without stents and coils.For the models with stent and coil placement within the aneurysm, the mesh size on the stent surface was set to 0.02 mm, and on the coil surface, it was 0.09 mm.The steady laminar Navier-Stokes equations were solved using CFX (version 2019R3, ANSYS Inc., Canonsburg, PA) with blood density of 1056 kg/m³and viscosity of 0.0035 kg m -1 s -1 .The no-slip boundary condition was applied at the vessel wall, and an inlet flow rate of 1.3 mL/s was assumed.The outlet boundary condition followed Murray's law, where the outlet flow is proportional to the cube of its equivalent diameter [18].The convergence criterion for calculations was set at 1e-6.
WSS, velocity at the aneurysm neck (neck velocity), inflow volume into the aneurysm via the neck (inflow volume), intra-stent flow velocity (ISvelocity), and intra-aneurysmal velocity (IAvelocity) were computed in this study to qualitatively assess the flow patterns.The definition of aneurysm neck and stent intraluminal flow area are shown in Figure S1.Hemodynamic simulations were relied upon to assess the therapeutic effects of different surgical strategies as shown in Figure 3C.

J o u r n a l P r e -p r o o f
In this study, we first obtained measurements of the lesion length, proximal diameter, and distal diameter of VADAs using HR-MRI in conjunction with DSA.These measurements were crucial for the precise selection of appropriate stent dimensions, including stent placement location, length, and diameter.Tailored treatment strategies were devised for each patient based on their specific conditions, ensuring that the stents effectively covered the lesion site, matched the vessel diameter, and were securely positioned.
However, given the diversity of stent types and various surgical plannings, even experienced surgeons might find it challenging to assess the treatment outcomes of different surgical plannings.
Therefore, we employed hemodynamic simulations to analyze and predict the efficacy of various surgical planning as depicted in Figure 3.These simulation results, when combined with the expertise of our surgical team, enabled us to choose the surgical planning that yielded the best treatment outcomes.
Finally, by comparing pre-operative and post-operative imaging data, including assessments of AWE and DSA results, we evaluated the treatment effectiveness of the final surgical planning.

Result Patients Characteristics
Five patients with u-VADAs were included in this study, 3 were male and 2 were female, with ages ranging from 42 to 74 years (median age: 53 years).Four u-VADAs were not involving the posterior inferior cerebellar artery (PICA) and the case 5 was involving PICA.

Vessel Segment Measurement and AWE Localization
The wall enhancement of 5 cases was mapped onto the DSA model and visualized in Figure S2.We calculated the lesion length in DSA models and multimodal models, respectively (Figure 2).Proximal diameter and Distal diameter were used for stent diameter selection.The measurement data showed that the actual lesion length of u-VADAs in the multimodal model was longer than those in DSA.In addition, AWE signals in HR-MRI were observed in the multimodal model, while not in the DSA model.

Selecting Treatment Strategies
Three distinct surgical strategies, including treatment with stent only, stent-assisted with coils, and flow diverter (FD), were proposed for each case (Figure S3).Following that, WSS, neck velocity, inflow volume, ISvelocity, and IAvelocity were calculated in three predictive models (Table 2 and Figure S4-S8).We found both stent-assisted coil embolization and FDs implantation effectively reduced neck velocity, inflow volume, IAvlocity, and increasing ISvelocity.In Case 1, neck velocity and IAvelocity J o u r n a l P r e -p r o o f decreased by 43.9% and 28.1%, respectively.In Case 2, the FDs approach resulted in reductions of 53.6% and 6.1% for neck velocity and IAvelocity, respectively.In Case 3, the FDs approach led to a reduction of 33.3% in neck velocity.According to Ouared's research, it is an effective treatment if the intraaneurysmal IAvelocity decreases by 33.5% after treating with FDs [19].Consequently, the FDs approach was excluded in Cases 1, 2, and 3.
Taking into account the comparison of the four hemodynamic parameters along with visualization results and relying on the clinical experience of a neuro-interventional expert with at least 15 years of experience, the final surgical planning was determined.For Case 1, Case 2, and Case 3, stent-assisted coil embolization was selected.In Case 4 and Case 5, both the stent-assisted coil embolization and FDs approaches exhibited favorable therapeutic effects.Considering the patients' conditions and the surgical team's experience, FDs were ultimately chosen for treatment.
All cases were successfully treated with the stent-assisted coil embolization and FDs approaches without intraprocedural rupture.The case 5 treated with FDs achieved complete occlusion of the aneurysm with the preservation of PICA after procedure.The Raymond-Roy Occlusion Classification (RROC) scores was evaluated with DSA.The total occlusion rate was 100% after surgery and during follow-up of 5 patients.

Follow-up Evaluation for Treatment Efficacy
Higher MCCratio, AVCCratio, and EACCratio are related to Inflammatory factors.In the postoperation 6 th month, we found significant decreases were observed in the MCCratio, AVCCratio, and EACCratio for Case 1, Case 3, Case 4, and Case 5 compared to pre-operative values, while Case 2 exhibited a slight increase in MCCratio, AVCCratio, and EACCratio (Table 3).Follow-up results at 4-6 months revealed a reduction of AWE in all cases (Figure S9).Additionally, DSA confirmed complete occlusion in Case 1, Case 3, Case 4, and Case 5, with Case 2 demonstrating near-complete occlusion (Figure 4).Endovascular treatments were performed on patients according to the pre-operative planning strategy.During the perioperative period and post-operative, none of the patients exhibited adverse outcomes.

Discussion
This study introduced a novel multidimensional method of surgical planning for u-VADAs and validated its effectiveness.By evaluating the hemodynamics and AWE of u-VADAs, the final treatment J o u r n a l P r e -p r o o f strategies of five cases were selected.All results have been successfully cured.This multidimensional method provided more diverse and more accurate information for surgical planning than DSA alone, including lesion areas and tears.Besides, quantifying the AWE of the vessel provided more foundation for evaluating follow-up results.
HR-MRI offers a high signal-to-noise ratio and spatial resolution, allowing for a visual depiction of aneurysm wall characteristics such as AWE and wall thickness, which are valuable for evaluating aneurysm stability [20].Gadolinium cannot penetrate the intimal of normal blood vessel walls.However, it can infiltrate the aneurysm wall with inflammatory cell infiltration or pathological neovascularization, manifesting as AWE [21].Numerous studies have demonstrated the correlation between AWE and the growth and rupture of aneurysms [7,10].Histological analyses have proposed a potential association between AWE and disruptions in the internal elastic lamina [22].This association with AWE signifies an increased risk of rupture.Retrospective studies have shown that AWE notably predicts the occurrence of recurrence [23].Consequently, AWE is a non-invasive biomarker of aneurysm instability.Besides, we believe that preventing blood flow from entering the aneurysm sac through the tear in VADAs may effectively prevent post-operative aneurysm recurrence.HR-MRI provides AWE and tear information, but it is challenging to assist in surgical planning because relying solely on HR-MRI makes it difficult to pinpoint the specific locations of AWE and tears.By mapping AWE information on 3D-DSA, this limitation can be addressed, enabling a more accurate distribution of AWE [24].We utilized AWE to select stents and locate the stents.We believe that by covering the AWE with stents or coil, the risk of aneurysm rupture can be reduced, and can accelerate the healing process of the aneurysm.Additionally, covering the tear can prevent blood flow into the aneurysm.
Abnormal hemodynamics plays a crucial role in the development of aneurysms [8,9,12].Increased blood flow within aneurysms can lead to rupture of the internal elastic lamina, and proliferation or reduction of smooth muscle cells, resulting in vascular wall degeneration and remodeling [25][26][27].WSS is a significant risk factor for aneurysm rupture [10].Low WSS induces destructive remodeling driven by inflammatory cells, rendering the aneurysm unstable; High WSS may also contribute to aneurysm enlargement and rupture [25].Simultaneously, WSS, IAvelocity, and neck velocity correlate with aneurysm recurrence.Lower IAvelocity forms areas of blood deceleration within the aneurysm, increasing the likelihood of thrombus deposition, and neck velocity partly reflects the blood flow rate and velocity entering the aneurysm [13,[28][29][30].A decrease in neck velocity suggests a prognosis for the J o u r n a l P r e -p r o o f aneurysm [31].ISvelocity, which intuitively reflects the guiding effect of stents on blood flow as a measure of the flow characteristics within the stent.When ISvelocity is higher, it suggests a more pronounced role of the stent in directing and controlling blood flow, effectively guiding blood flow toward normal directions.Hence, integrating hemodynamic information into surgical planning helps more accurately select the optimal treatment strategy.Reverse flow occurs on the aneurysm wall, creating a vortex region.The high flow velocity near the aneurysm wall is the primary cause of high WSS.Preoperative planning for Case1, Case2, and Case5 indicates that while a conventional stent can reduce WSS and velocity, it cannot alter the flow direction or mitigate vortex formation.On the other hand, stentassisted coil embolization can prevent blood from flowing into the aneurysm, resulting in extremely low or stagnant blood flow, and promoting thrombus formation.For Case3, Case4, and Case5, FDs effectively guided the blood flow, as evidenced by Isosurface Visualization and Velocity-slice images displaying reduced high blood flow velocity.However, in Case1 and Case2, high blood flow velocity remained along certain FD surfaces, resulting in residual high-velocity blood flow within the aneurysm.The result indicated that FDs guide blood flow, reducing turbulence and disturbances within the aneurysm sac, but the effectiveness of this guidance may vary due to the aneurysm's geometry, leading to inconsistent outcomes.It is essential to consider the aneurysm's geometry and hemodynamic parameters to ensure optimal guidance and treatment efficacy.
In this study, AWE was considered a significant reference factor for both treatment and follow-up observation.Based on the pre-operative geometric parameters and the localization of AWE, appropriate stent lengths were selected to ensure coverage of the enhanced wall regions.The selected stents after hemodynamic simulation achieved full coverage of the enhanced wall areas for five cases.The postoperative follow-up data indicated a decrease in AWE to varying degrees.In Case 3, AWE near the intimal tear was relatively inconspicuous, implying a more stable state of the aneurysm with a milder inflammatory reaction.However, in Cases 2, 4, and 5, AWE was not only limited to the vicinity of the intimal tear but also occurred in other areas of the aneurysm wall.This suggested that the inflammatory cell infiltration on the wall of these cases may be more active, increasing the risk of aneurysm rupture.
Except for Case 2, the AWE indicators in all other four cases showed a significant reduction.Nevertheless, 2D-DSA revealed good occlusion effects for all aneurysms.Therefore, AWE in Case 2 might be associated with thrombus formation and Gd retention.This study further supports that the reduction or disappearance of AWE might be related to favorable outcomes in aneurysms.We emphasize that precise J o u r n a l P r e -p r o o f localization of AWE provides accurate information for treatment planning.

Limitations
The limited number of cases and short follow-up duration introduced potential biases in this observational study.We will consider multiple follow-up for patients in the future to demonstrate that this method can reduce adverse outcomes.Based on only five cases, it is not possible to quantify the level of improvement in success rate compared to traditional treatment methods.The association between AWE and other normal vessels is uncertain.The impact of contrast agent retention or noise, these factors have not been eliminated in this method.Furthermore, an increased number of cases with more diversified cases will be required to refine the AWE indicators and hemodynamic parameters in the study for a better prediction of VADAs' recovery.Due to the lack of research data on anterior circulation intracranial dissecting aneurysms, it is not possible to assess whether the specific methods used in this study are applicable to anterior circulation intracranial dissecting aneurysms.

Conclusions
In this study, we presented a multimodal method for surgical planning for u-VADAs.This method combined DSA with HR-MRI and integrated hemodynamic simulation.The collaboration of DSA and HR-MRI enabled precise localization of AWE and vascular segments requiring treatment within u-VADAs and obtained necessary vascular diameter data.This multimodal method provided more information for surgical planning.Surgeons could reference these hemodynamic parameters, along with their clinical expertise, to select the optimal treatment planning.
Criteria J o u r n a l P r e -p r o o f This observational study enrolled patients with unruptured VADAs who were undergoing treatment at our hospital.The inclusion criteria were as follows: (1) Patients diagnosed with VADA, displaying typical imaging characteristics on DSA such as the double lumen sign, linear sign, or pearl-string sign, while excluding stenosis due to atherosclerosis; (2) Presence of aneurysmal dilatation.The exclusion criteria were: (1) Presence of untreated aneurysms in other locations; (2) History of aneurysm surgery at the same location; (3) Severe underlying metabolic disorders, significant liver or kidney dysfunction, malignancies, etc.; (4) The ruptured VADA.
2.0) to precisely point out the location of the intimal tear on the DSA model and quantify the signal intensity (SI).This script, a fusion of VTK 8.2.0 and QT5, streamlined the visualization and manipulation of HR-MRI frames.By loading the DSA model with spatial coordinates aligned to the MRI model, Interactive navigation and frame selection containing the intimal tear were enabled.For obtaining the SI at the central points of each facet of the DSA model, trilinear interpolation was employed for sampling.

Figure 1
Figure 1 Workflow of mapping wall enhancement on the vessel model.(A) Model reconstruction .(B) Model reconstruction in DSA.(C) Geometric alignment of DSA model with MRI model.(D) Fusion of DSA model and MRI images.(E) Signal intensity quantification, and localization of Normal and Lesions Blood Vessels.(F) Semi-automated localization of intimal tear.J o u r n a l P r e -p r o o f

Figure 2
Figure 2 Measurements of VADAs length and width.(A) DSA model.(B) DSA model combined with HR-MRI and positioning of intimal tear with intimal flap image.

Figure 3 Figure 4
Figure 3 Workflow of treatment strategy selection.(A) Defining lesion areas.(B) Virtual implantation of multiple plannings, a: Stent-assisted coil embolization; b: Conventional stent; (C) Comparing hemodynamics result and selecting the strategy for treatment.

J o u r
n a l P r e -p r o o f

Table 1
Geometric parameters of u-VADAs

Table 3
Pre-operative and follow-up AWE indicator

Table 2
Hemodynamic results of different surgical planning J o u r n a l P r e -p r o o f