Standardized fabrication method of human-derived emboli with histologic and mechanical quantification for stroke research

Abstract

Background

As access to patient emboli is limited, embolus analogs (EAs) have become critical to the research of large vessel occlusion (LVO) stroke and the development of thrombectomy technology. To date, techniques for fabricating standardized human blood-derived EAs are limited in the variety of compositions, and the mechanical properties relevant to thrombectomy are not quantified.

Methods

EAs were made by mixing human banked red blood cells (RBCs), plasma, and platelet concentrate in 10 different volumetric percentage combinations to mimic the broad range of patient emboli causing LVO strokes. The samples underwent histologic analysis and tensile testing to mimic the pulling action of thrombectomy devices, and were compared to patient emboli.

Results

EAs had histologic compositions of 0–96% RBCs, 0.78%−92% fibrin, and 2.1%−22% platelets, which can be correlated with the ingredients using a regression model. At fracture, EAs elongated from 81% to 136%, and the ultimate tensile stress ranged from 16 to 949 kPa. These EAs’ histologic compositions and tensile properties showed great similarity to those of emboli retrieved from LVO stroke patients, indicating the validity of such EA fabrication methods. EAs with lower RBC and higher fibrin contents are more extensible and can withstand higher tensile stress.

Conclusions

EAs fabricated and tested using the proposed new methods provide a platform for stroke research and pre-clinical development of thrombectomy devices.

INTRODUCTION

Rapid and immediate perfusion of ischemic brain tissue with mechanical thrombectomy devices has become the cornerstone treatment in large vessel occlusion (LVO) stroke.¹ Although stent retrievers and suction catheters have proven their value in large clinical trials, they have several limitations, including low rates of first-pass recanalization² and low overall rates of complete recanalization.³ The development of new technologies to more rapidly and completely recanalize LVO in stroke relies on accurate testing phantoms and embolus analogs (EAs). Unfortunately, techniques for the fabrication of standardized, human-derived EAs with characterized histologic and mechanical properties have not been described in the literature.

To date, the compositional variety of human EAs has been limited by the use of only human whole blood^4–6 or plasma,⁷ and no embolus analog was engineered to accurately mimic the mechanical behavior of the different emboli causing LVO strokes. In particular, there is a paucity of models to reproduce the tensile deformation and fracture patterns of emboli, which are critical phenomena driving the success or failure of mechanical thrombectomy with aspiration catheters and/or stent retrievers.^8,9 For example, an embolus can clog the aspiration catheter if the embolus is not fragmented by the vacuum inside the catheter. The stent retriever expands into and engages with the embolus for pulling. While undergoing pulling, the embolus is under tension and may break, leaving residual occlusions. Embolus fragments may also flow downstream and cause iatrogenic embolization and further strokes.¹⁰ In our knowledge, fabrication techniques to generate EAs with known and representative tensile strength that allow accurate analysis of the complex interaction at the artery/embolus/device interface have not been described to date.^11–14

In this study, the EAs are fabricated by mixing various volumetric percentages of human RBCs, fresh frozen plasma, and platelet concentrate obtained from the hospital blood bank before the expiration for human transfusion. Then we provide a histologic analysis of each EA formulation, with the percentage quantification of RBCs, fibrin and platelets. The mechanical tensile strength and stiffness of each EA formulation is then described. Finally, we correlate the histologic characteristics and mechanical strengths of these EAs to a cohort of emboli obtained during mechanical thrombectomy in patients suffering LVO strokes.⁸ Given the limited access to emboli from patients suffering LVO strokes, EAs that can appropriately match the structure, stiffness, and strength of patient emboli are critical to understand the dynamics of LVO and to development of next-generation thrombectomy technology. It is our expectation that this paper, which introduces a well described and standardized method to generate highly realistic human EAs at a minimal cost, will be of value to scientists and physicians trying to better understand the mechanisms of LVO and recanalization in stroke.

MATERIALS AND METHODS

Embolus Analog Fabrication

EAs were molded with different ratios of RBCs, plasma, and platelet concentrate (4.3 × 10¹¹ platelet counts per unit (300 mL)) acquired from University of Michigan Blood Bank. RBCs, plasma, and platelet concentrate were measured by volume and then mixed using a magnetic mixer for 2 min at 300 rpm. The human blood products were used due to its biological similarity to patient emboli, consistent preparing and storing standards across different institutions, and convenience to acquire. Human blood products have another advantage over the conventionally used animal blood which requires animal care and repeated blood draws with higher cost. Synthetic materials (such as silly putty or proprietary formulas) are another EA candidate but the lack of fibrin network prevents them from being used for potential fibrinolytic and anticoagulant drug studies or future devices that may interrogate the make-up of clots or have specific engagement techniques with the fibrin network. Calcium chloride solution (2.3 wt% concentration) was slowly added to the mixture at a 1:10 ratio to reverse the citrate phosphate dextrose and induce coagulation.^15–17 The mixture was injected into a dogbone-shaped mold, as shown in Fig 1a. Specimen dimensions with circular cross-section for a uniform stress distribution between the two bulged ends were adopted from the ASTM D638–10 Type IV geometry.¹⁸ The mold was 3D-printed with a 25-μm layer height using a stereolithography printer (Form 2; Formlabs, Somerville, Massachusetts) with a proprietary resin material (GREY FLGPGR04, Formlabs). The resin surface was rough due to the printer resolution limit. A clear coating (ColorMaster® Acrylic Spray Paint; Krylon Products Group, Cleveland, Ohio) was applied to the mold to smooth the surface. This clear coating also made it easier to remove formed EA specimens without causing damage before testing. The mixture was allowed to coagulate for 1.5 h before testing at room temperature.

FIGURE. 1.

Fabrication and characterization of embolus analog (EA) (a) Molding of a dogbone-shaped EA. (b) Volumetric percentages of the 10 types of EA: i_R, red blood cells; i_M, plasma; and i_P, platelet concentrate. (c) Tensile test setup to quantify the EA strength. (d) A sample pulling force measurement for EA #7. (e and f) Microscopic images of EA #7 with H&E and CD61 stains, respectively.

Ten types of EA specimens with varied volumetric percentages of RBCs (i_R), plasma (i_M), and platelet concentrate (i_P) were fabricated, as shown in Fig. 1b. Five EA specimens were made for each type of EA, resulted in a total of 50 specimen for tensile testing and histologic analysis.

Tensile Test Setup

The tensile test pulled the EA specimen until it fractured to quantify the EA tensile properties: (1) stiffness, the resistance of EA to elongation under pulling, and (2) strength, the maximum force and maximum elongation EA can withstand before fracture. As shown in Fig. 1c, the EA specimen was gripped by two hemostats at the bulged ends. One hemostat was fixed to a linear stage (Model 200cri; Siskiyou, Grants Pass, Oregon) and pulled at a slow and constant speed of 0.3 mm/s for the quasi-static tensile deformation. The other hemostat was fixed to a stationary force transducer (Gamma; ATI Industrial Automation, Apex, North Carolina). The tensile force on the hemostat was measured at a sampling rate of 50 Hz. Fig. 1d shows a sample tensile force vs. pulling distance for one EA #7 specimen. Pulling force increased as the linear stage moved to pull the EA specimen, and the force abruptly dropped to zero when the specimen fractured. EA specimens that fractured at the hemostats or slipped off the hemostats during testing were discarded and not reported. A video camera (model α6000, Sony, Tokyo, Japan) recorded digital images of the deformed EA specimen at 30 frames per second from the top. Two fiducials (Fig. 1c) were marked on the EA specimen to define the gage zone where the elongation and thinning of the EA were recorded during pulling. The positions of these two fiducials were measured to track the elongation of the EA specimen between the fiducials during the tensile test. An open-source image-processing application, Kinovea (v0.8.15), was used to find the distance L between the two fiducials to calculate the elongation, or engineering strain ε = (L-L₀)/L₀, where L₀ is the distance between the fiducials when the pulling force, F, increases from 0 to a small value (2 mN) at the start of the tensile test. This small value was selected as this transducer’s resolution is 1 mN and it has signal noise around 1 mN even when no force is applied. At fracture, the EA specimen has an ultimate strain (ε_ut), the maximum elongation the EA can achieve under pulling.

The tensile force on the EA specimen was normalized by the cross-sectional area of the specimen to calculate the stress (σ), loading on the EA per unit area caused by force (such as pulling by an aspiration catheter, a stent retriever, or, as in this test, a hemostat). Assuming the cross-section of the EA specimen remains circular during pulling, the diameter of the specimen, d, was measured using an edge-detection algorithm implemented in Matlab (v2017b; MathWorks®, Natick, Massachusetts), and σ = 4F/(πd²). At fracture, the EA specimen has an ultimate stress (σ_ut), the maximum force per unit area the EA can withstand. To evaluate the stiffness of the EA specimen, the stiffness E_0–0.45 (N/m²) was calculated as: E_0–0.45 = σ_0.45/0.45, where σ_0.45 is the tensile stress on the specimen at a strain of 0.45. Stiffness measures the extent to which the EA resists elongation in response to the tensile load (tensile stress). E_0–0.45 measures the averaged stress increasing rate on the EA when it elongates to 45% more than the original length (ε = 0.45). The strain of 0.45 was selected to compare with that in the literature.⁴

Histology Assessment

After tensile testing, a sample of EA was cut for histologic analysis to find the percentages of RBCs, fibrin, and platelets in the following three steps. First, the EA sample was stained with H&E to show RBCs in pink^12,14,15 and with CD61 to show platelets in dark brown.¹² Second, a microscope (Axioskop 2 plus; Zeiss, Oberkochen, Germany) was used at 400x and 600x magnifications to scan the H&E and CD61 stained slides, respectively. The images were captured by a digital camera (AxioCam MRc5; Zeiss) at 2584 × 1936 resolution. Figs. 1e and 1f show the microscopic images for EA #7 with H&E and CD61 stain, respectively.

Finally, the microscopic images were processed using ImageJ (v1.52; National Institutes of Health) to apply color masks to quantify the histologic percentages (area percentages) of RBCs (h_R) and platelets (h_P) under the guide of a senior pathologist. The balance (light pink) was assumed to be fibrin (h_F) and has a histologic percentage h_F = 1 − h_R − h_P.

Statistical Analysis

A multi-variable linear regression model was built to estimate the volumetric percentages of RBCs, plasma, and platelet concentrate based on the histologic compositions of the EAs. This model can be used to fabricate EAs with desired compositions. The histologic results of all EAs (n = 50) were used to fit the model, and the goodness of fit was evaluated by calculating the coefficient of determination (R²).

The Pearson correlation coefficient was calculated and used to evaluate how the EA compositions affected tensile properties (stiffness and ultimate strain and stress). This statistical analysis was carried out using SPSS v24.0 (IBM, Armonk, NY). Values of p < 0.05 were considered statistically significant.

RESULTS

Tensile Test and Histology Results

Ten types of EA specimens with varied volumetric percentages (i_R, i_M, and i_P) were successfully fabricated. Fig. 2 shows the tensile stress-strain curves for all 50 EA specimens and the histology results for each type of EA. For all EA specimens, stress increases with strain and increases faster as the strain becomes larger. This means the EA behaves stiffer as elongated. For every EA specimen, the fibrin network and platelet aggregates are mostly homogeneously distributed throughout the EA. In this study, EAs were composed of 0 to 96% RBCs, 0.78% to 92% plasma, and 2.1% to 22% platelets, levels similar to those found in stroke patients.^11–14

FIGURE. 2.

Stress-strain curves (with the average of the curves represented by the dashed lines) and histology (H&E and CD61 stain) of embolus analogs (EAs) #1–10. For each EA type, the five stress-strain curves represent five tests on five samples with the same ingredients. Scale bar = 0.1 mm.

Fig. 3 shows the aggregate averaged stress-strain curves for all 10 types of EA compared with the dashed curves for six patient emboli previously reported by our group,⁸ which were selected because they have more homogeneous structure indicated by a single fracture under tension, like the EAs in this study. The EAs have a large range of E_0–0.45, from 11 to 114 kPa, and a large range of tensile strengths, 0.78 to 1.36 for ε_ut and 16 to 949 kPa for σ_ut (a variation of 59 times). This is related to the highly variable EA compositions. EA #9 has the highest h_P and the largest ε_ut. EA #10 has the highest h_F and the largest E_0–0.45 and σ_ut.

FIGURE. 3.

Tensile and histology results of the embolus analogs (EAs) compared with 6 patient emboli (P1-P6) from the literature.⁸ The EA results are averaged across 5 samples for each EA type. Lower graph is a magnification of the area within the dashed rectangular box in the upper graph. h_R, red blood cells; h_F, fibrin; h_P, platelets; E_0–0.45, stiffness; ε_ut, ultimate tensile strain; σ_ut, ultimate tensile stress.

The EAs mimic the tensile properties of patient emboli, with some variances. For example, patient embolus P4 (black dashed line) has a stress-strain curve that is very similar to that of EA #6 (yellow solid line). P4 and EA #6 also have very similar histologic compositions, as shown in Fig. 3. Compared to P4, the ε_ut of EA #6 is about 6% higher, σ_ut is 20% lower, and E_0–0.45 is 50% higher (42 vs. 28 kPa).

Statistical Results

The multi-variable models to estimate the volumetric percentages of RBCs, plasma and platelet concentrate in order to make EAs with desired histologic compositions were fitted using the histology results:

$$i_R=0.367h_R-0.056h_F+0.124h_P$$ (1)

$$i_M=0.358h_R+0.833h_F+0.040h_P$$ (2)

$$i_P=0.276h_R+0.223h_F+0.836h_P$$ (3)

The model-estimated volumetric percentages have errors of 6.11%, 6.27%, and 6.62% and R² values of 0.878, 0.864, and 0.397 for RBCs, plasma, and platelet concentrate, respectively. The low R² value for i_P might be due to platelet aggregation in the formed EA, resulting in an inaccurate histologic quantification. For all three equations, the fittings are very statistically significant, with p < 0.001. This model can be used to make EAs with desired compositions to mimic different patient emboli.

Table 1 shows the correlation results between the tensile properties (E_0–0.45, ε_ut, and σ_ut) and EA composition (h_R, h_F, and h_P) based on the analysis of 50 EA specimens. The stiffness (E_0–0.45) is strongly and negatively correlated with RBC histologic percentage (r = −0.609, p < 0.001), strongly and positively correlated with fibrin histologic percentage (r = 0.613, p < 0.001), and moderately and positively correlated with platelet histologic percentage (r = 0.349, p = 0.013). The ultimate tensile strain (ε_ut) is moderately and negatively correlated with RBC histologic percentage (r = −0.536, p < 0.001) and moderately and positively correlated with fibrin histologic percentage (r = 0.555, p < 0.001). The ultimate tensile stress (σ_ut) is strongly and negatively correlated with RBC histologic percentage (r = −0.696, p < 0.001), strongly and positively correlated with fibrin histologic percentage (r = 0.702, p < 0.001), and moderately and positively correlated with platelet histologic percentage (r = 0.395, p < 0.001).

Table 1.

Correlation analysis results of the embolus analog composition and tensile properties (n = 50)

	hR		hF		hp
	r	p	r	p	r	p
E 0–0.45	−0.609	<0.001***	0.613	<0.001***	0.349	0.013*
ε ut	−0.536	<0.001***	0.555	<0.001***	0.240	0.093
σ ut	−0.696	<0.001***	0.702	<0.001***	0.395	0.004**

^*p < 0.05

^**p < 0.01

^***p < 0.001.

h_R, red blood cells; h_F, fibrin; h_P, platelets; E_0–0.45, stiffness; ε_ut, ultimate tensile strain; σ_ut, ultimate tensile stress.

DISCUSSION

In this study, EAs were successfully fabricated by mixing various volumetric percentages of human RBCs, fresh frozen plasma, and platelet concentrate obtained from the hospital blood bank. Then, the histologic percentages of RBCs, fibrin and platelets were analyzed, and each EA formulation underwent mechanical tensile tests to measure stiffness, strain, and strength. The EAs were found to accurately represent a broad range of emboli retrieved in LVO stroke, both histologically and mechanically, and therefore provide an advanced platform for translational research in stroke and thrombectomy technologies.

In this study, we elected to mechanically test the EAs employing a uniaxial tensile test as this is the driving force applied to the occluding embolus by current thrombectomy technologies. During LVO revascularization, the pulling force generated by the aspiration catheter or stent retriever on the embolus fights against the friction and adhesion between the embolus and the vessel wall and the blood flow. If the tension generated overcomes the impaction forces, and the embolus tolerates the load and remains cohesive, the embolus is removed by the device.⁸ However, during tensile loading the embolus elongates, becomes thinner and may fracture, leading to residual occlusion or embolization.⁸ Specifically for the stent retriever, the additional compressive forces and 3D shear force between the tines and clot may also contribute to the emboli fragmentation. A detailed analysis of the emboli-device interaction is included in the Appendix. Other groups have characterized the mechanical properties of emboli with shear wave elastography, which measures the small motion response of an embolus under propagating shear waves induced by ultrasound and can be used in vivo.^19–23 However, like other testing methods such as rheometer or compression test, shear wave elastography measures only the elastic modulus or viscoelasticity, which determines how the embolus will deform under force but not how it will fracture.

To fill this gap, we performed tensile tests to pull the EAs until fracture and quantify both the stiffness and tensile strength (the maximum elongation and force load the embolus can withstand before fracturing). The fabrication technique here described appropriately represent a large range (88% or 14/16⁸) of retrieved emboli causing LVO strokes in stiffness and tensile strength. We found that the tensile properties (stiffness and tensile strength) of EAs are correlated to the EAs’ compositions and mainly depend on the histologic percentage of fibrin content (with the highest r values). EAs with higher fibrin content are stiffer and have larger elongation and higher ultimate tensile stress. Such correlations can be explained by the mechanical properties of individual fibers within the EA’s fibrin network.

EAs with higher fibrin histologic percentage are stiffer (with higher E_0–0.45). The elastic modulus of fibrin fiber is about 14.5 MPa,²⁴ much higher than that of EAs, so it is reasonable to attribute the stiffness of the EAs to the fibrin fibers. Also, EAs exhibit greater stiffness as strain increases, showing a strong strain-hardening property, and such a phenomenon was also reported in tensile testing of EAs in the literature.^5,25 This can be explained by the strain-hardening property of individual fibrin fibers.²⁶

EAs with higher fibrin content can elongate more (with higher ε_ut). Fibrin fibers are very stretchable and can elongate to about three times the original length.²⁶ Combining this large elongation capacity with the high stiffness of fibrin fibers, EAs with higher fibrin histologic percentage therefore can withstand higher tensile stress (with higher σ_ut) before fracture. In this study, σ_ut ranged from 16 to 949 kPa, compared to only 6 kPa for EAs made from porcine whole blood⁵ and 30 kPa for EAs made from bovine whole blood.²⁵ The porcine- and bovine-derived EAs are significantly weaker than patient emboli (σ_ut ranging from 63 to 2396 kPa ⁸) and are likely fail to represent the stiff fibrotic emboli when used to test thrombectomy devices or to appropriately reproduce the hemodynamic conditions in an LVO. On one side, weaker analogs may not remain wedged in arterial bifurcations and withstand pulsatile physiological flow in cerebrovascular phantoms. Although this could be compensated by diminishing (or completely stopping) the flow, the occurrence of distal embolization would be underestimated. In addition, structurally weaker EA would be more likely to be “ingested” by suction catheters overestimating the rates of first-pass complete recanalization in the bench for a certaininner diameter and vacuum level, while in real clinical situation stronger clots would “cork” the catheters. On the other side, stents developed to incorporate these softer analogs could lack sufficient radial force to expand through emboli in LVO stroke leading to “clot rolling” and failed recanalization. Using the fabrication method described in this study, we made EAs that is strong enough to form stable LVO under physiological flow conditions and therefore provide a true estimation of the iatrogenic embolization. In addition, other studies using these EAs have shown to appropriately demonstrated the EA-device interaction and the device failure mechanism inside cerebrovascular phantoms.^8,9

Although the EAs presented here have stiffnesses and strengths similar to patient emboli and they constitute a significant step forward in LVO modelling, there are some differences to consider. First, whereas patient emboli can be quite inhomogeneous, these EAs (Fig. 2) were homogeneous in nature and did not contain some components related to thromboemboli that derive from ruptured atherosclerotic plaque, such as calcifications and necrotic core material. The pulling of an inhomogeneous patient embolus causes the weaker and less stiff parts of the embolus to peel off first and detach from the main embolus block, leaving the stronger and stiffer part behind to elongate. This embolus fragment can withstand larger elongation, as supported by the much larger ε_ut (up to 4.89) associated with multi-focal fracture compared to that of single focal fracture.⁸ The heterogeneity and topographic pattern (distribution of stronger and weaker parts) of emboli could change the way how emboli fragment during emboli-device interaction. The fragment of emboli could lead to residue emboli and/or distal embolization, as demonstrated in the Fig. A1. Therefore, future studies should quantify the heterogeneity and topographic pattern of patient emboli to guide the fabrication of EAs. Second, the EAs in this study did not age before testing, whereas patient emboli causing strokes may be hours, days, weeks or even months old before the onset of embolization and symptoms, with the constant “feeding” of blood components from the blood flow and potential hemolysis. Embolus aging is a dynamic biochemical process with observed effect on EA elasticity. The mechanical load on the embolus due to the physiologically pulsatile flow within the human body can also change the fibrin network density. In this paper, to standardize and simplify fabrication techniques while maintaining representative histologic and mechanical properties, we proposed an EA fabrication method in a static condition with short and fixed coagulation time (1.5 h). Third, the EAs did not include white blood cells, which were removed in the banked human blood materials, and were manufactured using un-expired blood product, which may limit the availability of banked human blood. In this regard, although we have not conducted mechanical testing of emboli derived from expired human blood product, in bench top simulation with thrombectomy devices we have not observed major differences in the mechanical behavior of emboli produced with products with less than 10 days from expiration. Fourth, ex-vivo-induced coagulation by calcium chloride could differ from the dynamic in-vivo coagulation process and modification of fibrin network.²⁷ Finally, the present manuscript did not examine the effects of coagulation time, calcium chloride concentration, and fibrin cross-linking on EA mechanical properties. The mechanisms determining EA microstructure response to different coagulation parameters and the structural differences compared to patient emboli are relevant topics for future studies. Possible correlations between the mechanical properties and/or histology of emboli and clinical data such as stroke characteristics, treatment strategy (e.g., thrombolytic medications), and patient outcome should also be explored.

In conclusion, this study provides a method of fabricating EAs with varying compositions and tensile properties to represent a broad range of emboli extracted from stroke patients. The volumetric percentages of the ingredients can be estimated by a multi-linear regression model to fabricate EAs with desired histologic compositions to mimic patient emboli. The tensile properties (stiffness, ultimate tensile strain, and ultimate tensile stress) were characterized by uniaxial tensile testing and found to be significantly correlated to the embolus composition, and accurately represent emboli causing LVO stroke. The EAs generated and characterized in this study can serve as valuable tools in facilitating the construction of accurate test beds for stroke research and the preclinical development of next-generation mechanical thrombectomy devices.

Highlights.

• Ten types of embolus analogs were made to mimic the broad range of patient emboli in large vessel occlusion stroke using banked human blood.

• Histology and tensile properties of the embolus analogs are similar to that of emboli from patients with LVO stroke

• Embolus analogs in this study can serve as a useful tool for stroke research and development of thrombectomy technologies

APPENDIX

Removal of emboli in LVO stroke is a process involving the force interactions between the emboli, thrombectomy device, vessel wall, and the blood flow. Both the aspiration catheter and stent retriever apply tensional force to dislodge and remove the emboli en bloc.

For the case of aspiration catheter, the catheter engages the emboli with vacuum suction (Fig. A1(a)). The embolus is elongated and ideally progressively ingested by the suction catheter, or “corks” at the catheter tip clogging the catheter. At this point, the catheter is pulled to apply further tensional forces to dislodge the emboli from the vessel wall (Fig. A1(b)). If the tensional force is bigger than the static friction and adhesion and the emboli is strong enough to withstand the tensional force, the emboli is dislodged and starts to move with the catheter. During this process, the intravascular portion of emboli are elongated, became thinner and weaker, and could fracture, leaving residual emboli behind. The fractured emboli substance is pushed by the partially resorted blood flow to the same or other branch vessels and cause distal embolization.

A similar mechanism is identified for the stent retriever. The stent retriever engages emboli with compression and shear between the emboli and tines (Fig. A1(c)) and then applies tensional force to dislodge the emboli (Fig. A1(d)). During the pulling process, the emboli are elongated, became thinner and weaker, and could fracture, causing distal embolization and residual emboli.

FIGURE. A1.

Force analysis of device-emboli interaction during thrombectomy with the aspiration catheter and stent retriever. (a) The aspiration catheter engages emboli with vacuum. (b) After engagement, the catheter is pulled to apply tensional force on the emboli to overcome the static friction and adhesion force between the emboli and vessel wall. During this process, the emboli is elongated and fractured (black arrow), leading to distal embolization under the hemodynamic stress and residual emboli. (c) The stent retriever engages emboli with compression and shear between the emboli and tines. (d) After engagement, the stent retriever is pulled to apply tensional force on the emboli to overcome the static friction and adhesion force between the emboli and vessel wall. During this process, the emboli is elongated and fractured (black arrows), leading to distal embolization under the hemodynamic stress and residual emboli.