Abstract
Understanding detailed hemodynamics is critical in the treatment of aneurysms and other vascular diseases; however, traditional digital subtraction angiography (DSA) does not provide detailed quantitative flow information. Instead, 1000 fps High-Speed Angiography (HSA) can be used for high-temporal visualization and evaluation of detailed blood flow patterns and velocity distributions. In the treatment of aneurysms, flow diverter expansion and positioning play a critical role in affecting the hemodynamics and optimal patient outcomes. Patient-specific aneurysm phantom imaging was done with a CdTe photon-counting detector (Aries, Varex). Treatment was done with a Pipeline Flex Embolization Device on a 3D-printed fusiform aneurysm phantom. The untreated aneurysm and two treatment stent expansions and positions were imaged, and velocity calculations were generated using Optical Flow (OF). Pre- and post-treatment images were then compared between different HSA image sequences and evaluated using OF with different stent positions. Differences in flow patterns due to changes in stent placement characteristics were identified and quantified with OF velocimetry. The velocity results within the aneurysm post-treatment showed significant flow reduction. Differences in stent placement result in substantial changes in velocities. The peak velocities found in the aneurysm dome show a reduction with the widened stent placement compared to the narrowed placement and both are reduced compared to the untreated aneurysm. The stent placements were compared quantitatively with the adjusted widened stent clearly better diverting the flow away from the aneurysm with decreased velocity in the aneurysm dome compared to both the narrowed stent placement and the untreated aneurysm. Providing this information in-clinic can help improve treatment and patient outcomes.
Keywords: 1000 fps, Aneurysm, Flow diverter, High Speed Angiography, Hemodynamics, Optical Flow, Invitro, Multi-angle
INTRODUCTION
An aneurysm results from a weakened area in the wall of an artery which causes the blood vessel to bulge; internal carotid artery (ICA) aneurysms are within the brain. Aneurysms ruptures can cause subarachnoid hemorrhage (SAH) which can have catastrophic consequences, including death. Aneurysms equal to the size of 10 mm have an overall risk of rupture per year of 1.9%.1 The diagnosis and treatment of an aneurysm are critical for patients given that cerebral aneurysm disease has been reported to affect 1-6% of the population.2
Endovascular treatment of aneurysms using pipeline stents provides a unique area of research that is not easily modeled. The stent itself provides unique interactions with the flow that cannot readily be calculated so the best way to study the treatment of aneurysms would be to physically treat 3D printed patient-specific models and observe the flow change with high-speed imaging. This work presents the results of evaluation of two stent deployment situations for one complex aneurysm model.
Treatment is usually done using Digital Subtraction Angiography using iodine contrast agent at frame rate of 3-15 fps to visualize the morphology of the vasculature.3 With this study, instead of using 3-15 fps to visualize vessel morphology, 1000 fps High-Speed Angiography (HSA) will be employed to visualize not only vessel morphology but also the fine detail of the contrast flow that is otherwise unseen with the typical frame rate. This examination of complex flow at 1000 fps enables unique understanding of the aneurysm treatment.
DESCRIPTION OF PURPOSE
Endovascular treatment of an aneurysm relies upon precise placement of flow diverters and other treatment devices. To better evaluate aneurysm treatment, 1000 fps High-Speed Angiography (HSA) can be used to acquire detailed blood flow information before and after treatment to show decreased blood flow within the aneurysm. Optical Flow (OF) can be used on the HSA images to quantitatively evaluate blood flow velocities within the vessel to evaluate treatment.
METHODS
A 3D-printed patient-specific Internal Carotid Artery (ICA) aneurysm phantom was used to replicate in-vivo flow for evaluation of anatomically relevant hemodynamics. The model was printed using a Stratasys J750 Digital Anatomy 3D printer. The model has a saccular aneurysm of size about 16 mm x 17 mm and its vessel length is 90 mm (Fig. 1). The diameter of the inlet of the model is 5.5 mm, and the outlet from the aneurysm is 4.5 mm. The outlet branches of the model are smaller, one 2.5 mm and the other 3.5 mm (Figure 1). The lateral view of this patient-specific model shows the vessel inlet to the aneurysm connected near the bottom following a curve in the inlet vessel and the outlet situated more on top of the aneurysm, on the side opposite to the inlet. Due to the aneurysm morphology, the hemodynamic information within the aneurysm is complex and the analysis of its image sequences benefits from multi-angled projection views.4 A Pipeline™ Flex Embolization Device was deployed for the treatment of the cerebral aneurysm model. A balloon catheter was used to adjust the positioning and to enlarge the narrowing of the initial placement of the device. HSA was used to compare the fluid dynamics of the (i) untreated model, and the model with (ii) a narrowed stent placement and (iii) an adjusted, widened stent placement.
Figure 1:
Patient-specific aneurysm. geometry frontal and lateral views (left and right respectively). Black lines across the vessel denote locations of diameter measurements.
A flow loop system was used to create anatomically relevant flow within the aneurysm phantom (Fig. 2). The flow system consists of a programmable flow pump, flow tubing, reservoir tank, and automatic contrast injector. The blood-mimicking fluid used is a mixture of 40% glycerol and 60% water solution by volume which was heated and mixed to mimic physiological fluid dynamics. The programmable flow pump used is a CardioFlow 5000 MR (Shelley Medical Imaging Technologies, Canada). The flow pump provided an anatomically relevant pulsatile carotid flow waveform shown in Figure 3 with peak flow rate of 60 mL/s and period of 828 ms. From the output of the flow pump, two lines are created, an internal carotid line, which is where the ICA phantom model is, and an external line which can be used to modify the flow in the internal line to be a specific flow rate. The average flow rate at the inlet of the phantom model was monitored at 245 ml/min or 4.083 ml/s using a TS420 transit-time perivascular flowmeter (Transonic Systems Inc.). Two ml of Omnipaque contrast media was injected over 1000 ms using an automatic contrast media injector system (Mark V ProVis, Medrad).
Figure 2:
Photos of flow system. Left: CardioFlow Pump with carotid waveform. Right: Tabletop set up with flow loop, reservoir tank, phantom, detector system and catheter which is attached to the automatic contrast injector (not pictured).
Figure 3:
Carotid waveform used in programmable flow pump with peak flow set to 60 ml/s.
RESULTS
Detector
A 1000 fps CdTe photon-counting detector (Aries, Varex) was used for image acquisition (Fig. 4). The Aries detector has a field of view (FOV) of 768 x 512 pixels, or 7.5 cm x 5 cm. The Aries detector was mounted to the frontal unit of a Canon C-Arm system to utilize its x-ray source from the mounted arm and enable easy rotation and acquisition at varying angles.
Figure 4:
1000 fps detector with 100 μm pixel pitch.
As opposed to traditional x-ray machines where the typical frame rate of angiographic image sequences ranges from 3-30 fps, when acquiring images at 1000 fps, the x-ray source will be required to be on for one continuous pulse. Given the excellent detail of 1 ms images, the x-ray duration can be significantly less than that of a multi-minute image acquisition for traditional angiography. In this study, the x-ray image acquisition and the contrast injection were manually synchronized so the duration of x-rays was set to one continuous pulse of about 1 second. The x-ray paramaters used for this study were 86 kVp, 200 mA, and 19 mm Al filtration.
Optical Flow
Once the images are acquired, an Optical Flow (OF) method was used to derive flow velocities for quantitative comparison. OF derives velocities in a two-step process; it begins with the Horn-Schunck5 method using image intensity differences based upon gradients of brightness, followed by application of the method from Liu and Shen6 which uses conservation of mass and boundary conditions to determine velocity. Because the OF method relies upon brightness gradients, OF excels in locations with large contrast gradients such as a contrast bolus edge. In other cases where there is no contrast or unchanging streamlines of contrast, it will not detect accurate velocities.
For treatment comparison, Optical Flow (OF) was used to calculate velocity when the contrast is located within the aneurysm. Under the assumption that any change of brightness is due to underlying motion, the OF velocity determination algorithm relies upon spatial (pixel-by-pixel) and temporal contrast intensity changes between consecutive 1 ms HSA frames. Initially, a gaussian blur and a small temporal filter is used for improved temporal continuity at 1 ms time intervals. Then, regions within the vasculature are masked either manually or using a k-means clustering if the specific run has sufficient contrast for clear differentiation from background.
The frontal anterior-posterior view of the 3D printed aneurysm model was carefully chosen to be the 0-degree view and HSA images were taken for each condition tested at 0, 45, 90 degrees RAO. Figure 5 shows the conditions tested for the model: untreated, narrowed stent placement and widened stent placement. The magnified locations in the frontal view (Fig 5b and 5c) clearly show the differences in the narrow versus widened stent placement. To highlight one difference between image 5b and 5c, a magnified region was created to show the narrowing and pinch in b compared to c where the stent is more open. This can also be seen in 5e and 5f; in Figure 5e, which corresponds to the lateral view of 5b, the stent does not extend to the bottom of the aneurysm and instead is pinched at the curve from the inlet to the aneurysm; in Figure 5f, that region where the stent goes from the inlet to the aneurysm is much more open, extending the stent almost to the bottom of the aneurysm. In the first placement of the stent, there is the narrowing of the stent at the inlet to the aneurysm and the stent is generally not fully opened throughout the aneurym and outlet (5b and 5e). The stent was adjusted using a ballon catheter to widen the stent throughout the entire placement (5c and 5f). For 5c and 5f, the more open stent position is espeically widened in the entrance to the aneurysm. Because the stent was widened, the total length of the stent decreases slightly, which can be seen at the beginning and end of the stent.
Figure 5:
HSA images (not flat-field corrected) showing two views (frontal and lateral) of the aneurysm model for the three tested conditions: Frontal view (a) untreated aneurysm, (b) narrowed stent and (c) widened stent placement and Lateral view (d) untreated aneurysm, (e) narrowed stent, and (f) widened stent. These images were created by averaging image frames to see the 3D printed vessel morphology and the stent placements in the two stented models. The corresponding lateral views help illustrate the stent location in depth for better understanding. Stent conditions are highlighted. In f, the outlet vessel slightly overlaps with the aneurysm and the darker shaded rectangles in the lateral views are due to the model holder material. The light lines across the images are from the inter-detector-module lines.
In all pictures and videos following, the HSA images are DSA images, and the flow direction goes from left to right. The anatomical orientation of the patient-specific aneurysm would be rotated 90 degrees to the left, where flow goes from bottom to top on the clinical display, but for better visualization and spacing in the manuscript all image sequences of this aneurysm model will be displayed with flow going from left to right.
HSA videos depicting the detailed fluid flow will be shown for each condition. The Figure 6 video begins with an edge of contrast that flows into the aneurysm. As the fluid flows, that collection of contrast swirls clockwise and slowly disperses through h the aneurysm with small streamlines can be seen. The region on the upper part of the aneurysm is where the inlet contrast directly aims and begins the circulation around the aneurysm. The outlet of the aneurysm is on the bottom part of the aneurysm in the image and some contrast exits the aneurysm but much of it continues circulating within the aneurysm itself.
Figure 6:
Video 1, 1000 fps HSA video for the Unstented 0-degree frontal view played back at 50 fps. The sections where contrast visually speeds up corresponds to the systolic peak in the carotid waveform. In this video, an edge of contrast can be seen starting in the top right section of the aneurysm. All videos shown in this manuscript have a playback rate of 50 fps. In this video, flow goes from left to right. http://dx.doi.org/10.1117/12.3006922.1
For visulization, 1 ms frames are shown every 20 ms in Figure 7. Note that data was collected and analyzed at each 1-ms frames, but visualization would be shown best in video. The 1000 fps allows visulization of the complex vortices within the pooling of the contrast in the unstented model (Fig. 7).
Figure 7:
Unstented 0-degree frontal view frames from the video sequence in Figure 6 illustrating a 1-ms frame every 20 ms. All image arrays are organized in a sequential left-to-right, top-to-bottom arrangement. The circulation inside the aneurysm is clockwise and much of the contrast continues circulating causing the aneurysm to darken as time continues. The detail circulation and flow dynamics are better visualized with the 1000 fps video in Figure 6.
For better understanding of the 3D morphology frontal and lateral views of the patient-specific model HSA projection images are shown in Figures 6 and 8, respectively. One important understanding gained from the multi angled view is the orientation of both the inlet and the outlet. The inlet to the aneurysm enters the aneurysm on the bottom (Fig. 8) of the aneurysm and is angled toward the wall of the aneurysm. The outlet is on the opposite side of the aneurysm, in the top of Figure 8, but also on the opposing side in the frontal view. In both views, if the aneurysm was not there, the inlet and outlet would connect like a y = x3 line. From the lateral image, it can be seen that the aneurysm is over the inlet and under the outlet overlap regions shown in the frontal view. The outlet of the aneurysm is on the top part of the aneurysm in the lateral view and as flow circulates, some contrast exits the aneurysm but much of it continues circulating within the aneurysm itself (Figure 9). The region in the center of the video has foreshortening in the inlet. Once treated, the remade vessel not only has to transverse the lateral distance between the inlet and the outlet but also must compensate for the vertical distance. This geometry adds to the complexity of treatment for this aneurysm. When the stent is deployed in this model, it must go along a general ‘S’ shape from the inlet to the outlet, especially in the lateral view. This curvature makes properly deploying the stent more complicated.
Figure 8:
Video 2, 1000 fps HSA video of the Unstented 90-degree lateral view played back at 50 fps. In this video, an edge of contrast can be seen starting in bottom section of the aneurysm. The edge swirls around and hits the right side of the aneurysm. As the fluid flows, that collection of contrast swirls inside the aneurysm and slowly disperses. The region on the right side of the aneurysm is where the inlet contrast directly aims and begins the circulation around the aneurysm. Flow goes from left to right. http://dx.doi.org/10.1117/12.3006922.2
Figure 9:
Unstented 90-degree lateral view of the model. Each image is a 1-ms frame every 20 ms taken from the video shown in Figure 8. In the lateral view of the unstented model, contrast flows in from the bottom right, towards the right side of the aneurysm where it deflects off the wall and begins to swirl towards the left side of the aneurysm. There some of the contrast exits the top of the aneurysm, but most continues to disperse inside the aneurysm. The most definition to the contrast edge occurs in the beginning of the image sequence.
Frontal and lateral views of this model provide depth information that is otherwise unseen with a single orientation. Additionally, a third projection was obtained at 45 degrees, halfway between the frontal AP view and the lateral view; a 45 degree view of the model provides more information on the vortex of the contrast within the aneurysm (Figure 10).
Figure 10:
Video 3, 1000 fps HSA image sequence video of the unstented 45-degree view In the 45 degree view, where the projection is halfway between the frontal and lateral views, the circulation inside the aneurysm is more complex. After the contrast hits the side wall of the aneurysm, there is some circulation in a clockwise motion. In this figure, the flow goes from left to right. http://dx.doi.org/10.1117/12.3006922.3
Regarding the frontal view of the unstented aneurysm model, circulation in the aneurysm goes about in a clockwise manor as seen in video Figure 6. Most of the contrast flows in from the left side along the edge of the aneurysm dome, then down, hitting the aneurysm wall so that part of the contrast bolus flows out the aneurysm and the rest continues to circulate in one vortex around the aneurysm. But that only shows part of what is happening. Using the lateral view in Figure 8, an additional depth component can be visualized. The contrast flows in along the bottom of the aneurysm, then hits the right wall were part of the ciculation happens. It can now be seen that the outlet from the aneurysm is towards the opposite side of the aneurysm which means that the contrast must flow from bottom to top along the aneurysm dome edge. The circulation inside the dome can be seen in the lateral view as it circulates around the entire side of the aneurysm, not just in one section of the aneurysm. In the 45-degree view, the circulation of contrast begins towards the viewer, then also goes along the bottom edge of the aneurysm as it is directed back towards the inlet of the aneurysm (Figure 11). In the last image of Figure 11, the contrast starts to make a clockwise vortex inside the aneurysm at this 45-degree angle projection.
Figure 11:
Unstented 45-degree view showing 1 ms frame images every 50 ms. This flow is demonstrated in Figure 10 video.
All the above videos and image seuqnces were gathered for the unstented, untreated 3D printed patient-specific model. The next sequences are for the narrow-stented model.
In the first stent treatment of the aneurysm, the stent itself seems slightly pinched and not open fully as shown in Figure 5b and 5e and so will be referred to as narrowed stent treatment. The narrow stented treatement model shows in the frontal view that while some of the contrast exits the aneurysm, quite a bit of contrast builds up inside the aneurysm. There is a region in the inlet where contrast is entering the aneurysm, and there is a region in the center of the aneurysm where the stent darkens from the iodine contrast slowly dispursing into the aneurysm (Figure 12). While the second mechanism for contrast entering the aneurysm is expected, the first entering in the aneurysm at the inlet is not expected. Figures 12 and 13 show that in the frontal view, contrast jets into the aneurysm on the top side of the stent. The contrast that enters the leftover aneurysm region begins to swirl clockwise but does not seem to exit the aneurysm well and thus as time goes on, the aneurysm fills up. The region where the stent created the new vessel inside the aneurysm is best seen in one of the first images of Figure 13 but there is a significant escape from the stent near the inlet of the aneurysm, which makes the region where the stent is harder to see in the later images.
Figure 12:
Video 4, 1000 fps HSA video Narrow stent 0-degree frontal view (Left) with a frame-averaged, subtracted and un-flat-field corrected image without iodinated contrast illustrating narrow stent placement (Right). Flow goes from left to right. The frontal view of the narrow stent placement shows the beginning of injection as it enters the aneurysm and some of the contrast continues into the aneurysm where the stent is not placed. Some of the contrast goes through the stent where the vessel location is, while the rest enters the aneurysm at a high rate and begins to circulate around the aneurysm. This illustrates an effect of the stent that is not effective for all of the flow. http://dx.doi.org/10.1117/12.3006922.4
Figure 13:
Narrow stent 0-degree frontal view, showing 1-ms image every 20 ms corresponding to video of Figure 12.
In the lateral projection of the narrow stent, there is a jet of contrast that enters the aneurysm and flows along the bottom of the aneurysm. This is illustrated in both the frontal and lateral views. Figure 14 video shows the beginning of injection as it enters the aneurysm and some of the contrast continues into the aneurysm where the stent is not placed. Some of the contrast goes through the stent where the vessel location is, while the rest enters the aneurysm at a high rate and begins to circulate around the aneurysm. This illustrates an effect of the stent that is not effective for all of the flow. When contrast first going through the stent, most of it exits the aneurysm, and then some starts to drip and disperse from the stent. That jet goes along the bottom of the aneurysm and reaches near the top of the aneurysm on the right side next to the stent going out. It does not seem like the contrast within the aneurysm has a clear place to go and by the later images, the section of the right side of the aneurysm, next to the stent is filling with contrast. Additionally, a neat effect can be captured at 1000 fps in Figure 15 where contrast droplets leave the stent by falling through the stent as systole passes.
Figure 14:
Video 5, 1000 fps HSA video sequence of the Narrow stent 90-degree lateral view. In this run, the narrow stent placement goes from bottom to top of the aneurysm. Flow goes from left to right. http://dx.doi.org/10.1117/12.3006922.5
Figure 15:
Narrow stent 90-degree lateral view images corresponding to the video in Figure 14. Unlike the other image sequence arrays, this figure shows one image every 50 ms corresponding to video in Figure 14, the very beginning of contrast going through the stent.
The 45-degree projection half way between the frontal and lateral views provides a unique look into the narrow stent treatment. In Figure 16’s video, some of the contrast does flow directly along the stent but there is also some contrast that is able to escape the stent and flow quickly into the aneurysm. The region where this occurs is at the turn right before the entrance to the aneurysm. Contrast also disperses from the stent in the bend of the stent in the middle of the aneurysm, in the video it looks almost like droplets of contrast coming out of the stent. Towards the end of the video, much of the contrast inside the aneurysm swirls around inside the aneurysm without seeming to exit the aneurysm at all. Much of the contrast can be seen entering the bottom side of the aneurysm and swirling up around. Where the stent enters the aneurysm, the narrowing of the stent creates a region where contrast can jet into the aneurysm outside of the stent path during systole. Once contrast is inside the aneurysm, most of the contrast cannot escape in the time frame of the video (Figure 16).
Figure 16:
Video 6, 1000 fps HSA video Narrow stent 45-degree view played back at 50 fps. The view halfway between the frontal and lateral views of the narrow stent-deployed model provides an interesting cohesive understanding of the flow inside the aneurysm. Flow goes from left to right. http://dx.doi.org/10.1117/12.3006922.6
Next the stent was widened using a balloon specifically at the regions where the stent seemed to be pinched. This enabled a third condition to be studied, where using the exact same stent and model, the stent was opened or widened at both the inlet and outlet of the aneurysm.
The frontal view of the aneurysm with the widened stent treatment, the third condition visualized in this study, shows a much more effective diversion of flow through the aneurysm (Figure 18). Most of the contrast goes directly through the vessel, with edges of contrast passing though as the contrast breaks apart slightly at the bend where the inlet and aneurysm meet. The main contrast edge goes directly though the stent path and there is only slow diffusion of contrast into the aneurysm region outside the stent. By the end of the video of Figure 18, the entire aneurysm has not filled with contrast. Figure 19 shows images every 20 ms from the video in Figure 18. Even by the end of Figure 19, the entire aneurysm is not visible or filled with contrast.
Figure 18:
Video 7, 1000 fps HSA video of widened stented 0-degree frontal view video sequence played back at 50 fps (Left), with summed image illustrating widened stent placement (Right). For this video the entire 7.5 x 5 FOV is shown. The injection of contrast starts off by following the stent placement which is expected. There is a slower diffusion entering the aneurysm and slowly defining the edge of the aneurysm. Flow goes from left to right. http://dx.doi.org/10.1117/12.3006922.7
Figure 19:
Widened stent 0-degree frontal view every 20 ms, shows slow buildup of contrast in the aneurysm. The contrast that is entering the aneurysm seems to be much more diffuse and not have a direct edge to it as it fills the aneurysm.
In the widened stent model’s lateral view (Figure 20), much more of the contrast is following the path of the stent than in the previous narrow stent model’s video (Figure 14). For better understanding of the flow inside of the aneurysm, two segments of time were used to create the lateral view image array displays, Figures 21 and 22. Figure 21 shows the beginning of the contrast injection going through the aneurysm where only towards the later part of the sequence does some of the contrast begin to fill the aneurysm. Figure 22 follows that sequence where the contrast is slowly filling the aneurysm. As intended with flow diverter treatment, the blood inside the aneurysm segment will slowly fill and the blood stasis will cause thrombosis and stop aneurysm growth.
Figure 20:
Video 8, 1000 fps HSA video of widened stent 90-degree view played back at 50 fps. Flow goes from left to right. There is a slow filling occurring in the aneurysm but much of the flow is diverted by the stent exiting the top of the aneurysm. In this video there is a small jump of 100 ms in the middle to demonstrate the flow in the beginning and later. The slight line on the right side of the image is due to an artifact from the model’s outlet structure supports. This is also where the half circle artifact on the right side of the image came from. http://dx.doi.org/10.1117/12.3006922.8
Figure 21:
Widened stent 90-degree lateral view displayed 1-ms frames every 20 ms. In the beginning, the contrast mostly goes through the stent, and only some begins to fill the aneurysm. Only during systole does the aneurysm begin to fill. Over time, the aneurysm does start to begin to slowly fill.
Figure 22:
Later segment of the cycle for the widened stent 90-degree lateral view displayed every 20 ms. The first frame begins 500 frames after the first image of Figure 21, or about 200 frames after the last image of Figure 21. This next segment of aneurysm pictures was chosen to show the flow filling of the aneurysm. This filling of the aneurysm occurs much slower than that of the narrow stent and the unstented model. With 20 ms between each frame, a total of 300 ms change between the first and last image, illustrating the magnitude of the speed of filling of the aneurysm with contrast.
Next is the 45-degree view of the widened stent aneurysm model (Figures 23 and 24). In this view, the stream of contrast through the stent is best visualized because much of the contrast entering the aneurysm goes along the bottom edge as it fills, and this does not overlap as much with the intended flow path through the stent compared to other views.
Figure 23:
Video 9, 1000 fps HSA video of the widened stent 45-degree view. Flow goes from left to right. There is a slow filling occurring in the aneurysm but much of the flow is diverted by the stent exiting the top of the aneurysm. In this video there is a small jump of 100 ms in the middle to demonstrate the flow in the beginning and later. http://dx.doi.org/10.1117/12.3006922.9
Figure 24:
Every 20 ms 1-ms frames from the widened stent 45-degree view. Much more of the contrast goes directly along the stent path in the top of the images. The contrast filling of the aneurysm begins at the bottom region of the aneurysm first. These images correspond to the video in Figure 23.
The HSA videos corresponding to the unstented model, narrow stented model, and the widened stented model demonstrate the complexities of flow in each configuration. Comparison between videos of contrast flow with the narrow stent model and the widened stent model show that treatment of the same aneurysm with the same stent, but with just a minor change of stent placement results in drastically different blood flow conditions. Besides the HSA video demonstration of these flow details, quantitative information can be gained and compared evaluating the flow details.
OPTICAL FLOW RESULTS
OF calculates the velocity distributions based on contrast difference between pixels and between frames. It excels where there is a large bolus of contrast with a clear edge. In regions where there is no contrast, or regions of streamlined contrast, OF outputs velocity measurements of zero. Therefore, a maximum velocity over a relevant segment of an HSA sequence is used to compare aneurysm treatment. The temporal maximum velocities over a 300 ms sequence for all three conditions: unstented, narrow stented, and widened stent is shown in the following figures.
The OF velocity results of the untreated aneurysm demonstrate a clockwise vortex around the aneurysm dome at a relatively high velocity of 30 cm/s with max velocity near the outlet with peak velocity at the edge of contrast in the bottom region of the aneurysm around 50 cm/s (Figure 25a). In the unstented model video sequence the contrast bolus begins to swirl around the outer edge of the aneurysm dome and then hits the bottom portion of the aneurysm dome and swirls back towards the inlet of the aneurysm (Fig. 6). In the narrowed stent placement, there is a vortex within the aneurysm with a smaller magnitude compared to the untreated aneurysm, but the velocity near the top of the aneurysm outside the stented region has a mean velocity around 16 cm/s with peak velocity around 23 cm/s, while within the stented region the mean velocity is only 5 cm/s (Figure 25b). Within the aneurysm there is a region of higher velocity towards the top left, due to the leak of contrast through the narrowed stent (Figure 12). Additionally, that leak into the aneurysm is represented with the higher velocity in the top right (Figure 25b). In the widened stent placement, the flow is diverted through the stent and the contrast leakage into the aneurysm dome is much less than the narrowed stent (Figure 18). This is evident in the HSA images as well as the OF resulting temporal maximum image where the OF results inside the remaining aneurysm region outside the stent is a mean of 5 cm/s where the contrast reaches but less than that near the edge of the aneurysm and the velocity through the stent is 17 cm/s with some values closer to 32 cm/s at the curve (Figure. 25c). In the frontal view, the reduction of the flow inside the aneurysm between the unstented model and narrow stented view is 14 cm/s, and between the widened-stented view and the narrow-stented view is a difference of 12 cm/s in the stent region. Because the widened stent diverts flow better than the narrow stent, the velocities within the widened stent region are higher than within the narrow stent. Compared to the other two views, the narrow stent model shows a lower velocity range, but this might be due to the fact that in the other two there is a clear bolus of contrast that goes in one general direction, while in the narrow stent view the contrast splits between going through the stent and into the aneurysm.
Figure 25:
OF results of the 0-degree frontal views for the aneurysm in the three conditions: the unstented model (a), narrow stented model (b), and widened stented model (c). These correspond to the videos in Figures 6, 12, and 18 respectively. In the unstented view, the curve of higher velocity is traveling clockwise inside the aneurysm. General clockwise flow occurs in all three views.
The OF results in the lateral view of the aneurysm shown in Fig. 26 provides a little more detail into the flow differences. While the unstented model does not have a main path for the contrast to flow, the other two OF results for the narrow-stented model and the widened stented model provide better distinction for the intended flow. In the unstented model’s lateral view, the OF velocity results show a mean velocity of 20 cm/s in the swirl right to left inside the aneurysm with peak velocities around 40 cm/s (Fig. 26a). In the narrow-stented model, within the stent the velocity results are about 5 cm/s, but the region outside the stent has a mean velocity of about 10 cm/s with peak values of 20 cm/s (Fig. 26b). For the widened stent, much more of the flow is successfully diverted through the stent with its mean velocity inside the stent region 35 cm/s and the mean velocity outside the stent region only 5 cm/s. This shows a difference of 30 cm/s in the velocity in the stented region, between the narrow stented lateral view and the widened stented lateral view. This shows that in the lateral view, the widened stent successfully diverts much more flow compared to the narrow stent model.
Figure 26:
OF results of the 90-degree lateral views for the aneurysm in the three conditions: the unstented model (a), narrow stented model (b), and widened stented model (c). These correspond to the videos in Figures 8, 14, and 20 for the unstented, narrow, and widened stent respectively. While the flow mainly goes from left to right, in the unstented view 26a, inside the aneurysm the contrast is mainly going from right to left due to the swirl vortex.
One additional feature to point out from the HSA image sequences in comparison to the lateral OF results (Fig. 26) is that the magnitude of time that passes for the circulation of contrast within the aneurysm is on a different scale for each flow condition. In the unstented model as expected, the swirl vortex of contrast flow inside the aneurysm happens the quickest. In the lateral view, once systole occurs, contrast jets into the bottom of the aneurysm and then begins to swirl around out-of-plane. There is also a jet of contrast occurring in the narrow stent view where the contrast goes along the bottom of the aneurysm. The OF image corresponding to the narrow lateral view shows the flow into the aneurysm at the bottom region has a velocity of 10 cm/s but it also does not have as clear an edge as it goes along the bottom of the aneurysm which might result in a slight underapproximation of the velocity. In the widened stent view, on the same time scale as the other two, it barely looks like any of the contrast is filling in the aneurysm at all (Figure 27).
Figure 27:
HSA lateral views of the aneurysm phantom 1-ms frames every 15 ms for the (a) untreated, (b) the narrowed stent placement, and (c) the widened stent placement. Note the framerate chosen is higher than standard DSA for best viewing. Corresponding videos showing 1000 fps with consecutive 1-ms frames are recommended for best demonstration. A direct line up of the three conditions gives great perspective on the time scale it takes for the aneurysm to fill with contrast.
The OF results for the 45-degree view show a mean velocity of 20 cm/s in the unstented view with contrast swirling around in the aneurysm with peak velocity of 45 cm/s (Fig. 28a). In the narrow-stented model, the velocity of 15 cm/s was shown in the aneurysm region outside the stent on the bottom part (Fig. 28b). In the widened-stented model, the velocity of 15 cm/s was shown in the stented region of the aneurysm and 5 cm/s in the area below that outside of the stent path (Fig. 28c).
Figure 28:
OF results of the 45-degree views for the aneurysm in the three conditions: the unstented model (a), narrow stented model (b), and widened stented model (c). These correspond to the videos in Figures 9, 16, and 23 for the unstented, narrow, and widened stent respectively. Only the velocity magnitude is shown in these images, for example while the velocity in the unstented inlet is going left to right, inside the aneurysm, the swirl is going against that direction, so the velocity shown inside the aneurysm is mainly facing right to left directionally which can be seen in video Figure 9.
The flow velocity within the stent changed from 7 cm/s in the narrow-stented model to 15 cm/s in the widened stented model. The flow in the region that is outside the stented region went from 15 cm/s for the narrow-stented model, and 5 cm/s with areas less than that for the widened stented model. There is a more direct path for fluid flow to go through the stent in the widened stent model compared to the narrow stent model demonstrated by the increase flow inside the stent and decrease in flow in the remnant aneurysm region. Though both show significant flow differences compared to the pre-treatment unstented aneurysm model, the widened stent model demonstrates the most successful diversion of flow from the aneurysm.
DISCUSSION
For each of the three conditions, three different angled projections were acquired. Within those conditions, systole was manually segmented in 300 ms to create OF plots. Ideally, the injection of contrast happens during diastole about a fraction of a second before systole occurs so that the contrast has enough time to begin to pool so that when systole does occur, the increase in velocity will push a large bolus of contrast through the vessel which will then be used for best OF calculations. Though in these videos there is variation during the carotid cycle when contrast is injected, each provides optimally chosen runs. The angled view of the aneurysm at 1000 fps provides more information on the flow inside the aneurysm that would otherwise be unknown. It can be seen in more detail how contrast flows in response to two placements of the same stent.
CONCLUSION
A detailed study was performed on a patient-specific aneurysm model with treatment along with velocity analysis. A clear drastic difference is demonstrated between different treatments by the same flow diverter deployed within the aneurysm. HSA images compared through OF temporal maximums provided depiction of dynamic flow patterns. Stent placement is shown to have a significant impact on the flow velocities within the aneurysm, with the widened stent able to divert significantly more flow through the stent, and away from the aneurysm, as illustrated in both the HSA and OF images.
Figure 17:
Narrow stent 45-degree view image array, which shows 1-ms frame of the aneurysm every 20 ms. This image array corresponds to the video in Figure 16. In this view, contrast drips from the stent in the beginning and collects in the bottom of the aneurysm. Then as systole passes through the vessel, more contrast escapes from the stent at the region right where the inlet and aneurysm meet which agrees with the two other views of this model. Then as more contrast escapes into the aneurysm, it begins to swirl in a vortex around the stent.
ACKNOWLEDGEMENTS
This work was partially funded by NIBIB Grant 1R01EB030092 and Canon Medical Systems, Inc.
REFERENCES
- [1].Rinkel Gabriel J. E. et al. “Prevalence and Risk of Rupture of Intracranial Aneurysm,” AHA Stroke, vol. 29, no. 1, pp 251–256 Jan 1998, 10.1161/01.STR.29.1.251 [DOI] [PubMed] [Google Scholar]
- [2].Wardlaw JM, & White PM (2000). The detection and management of unruptured intracranial aneurysms. Brain: a journal of neurology, 123 (Pt 2), 205–221. 10.1093/brain/123.2.205 [DOI] [PubMed] [Google Scholar]
- [3].Pooley RA, McKinney JM, Miller DA., “The AAPM/RSNA physics tutorial for residents: digital fluoroscopy,” Radiographics, 21 521–34 (2001). 10.1148/radiographics.21.2.g01mr20521 Google Scholar [DOI] [PubMed] [Google Scholar]
- [4].Vanderbilt E, Wu X, Shields A, Setlur Nagesh SV, Ionita C, Bednarek DR, Rudin S, “Multi-angled simultaneous biplane High-Speed Angiography (HSA) of patient-specific 3D-printed aneurysm phantoms using 1000 fps CdTe Photon-Counting Detectors (PCD’s),” Proc. SPIE 12468, Medical Imaging 2023: Biomedical Applications in Molecular, Structural, and Functional Imaging, 124680M (10 April 2023): 10.1117/12.2653136 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Horn BK, & Schunck BG, “Determining optical flow,” Artificial Intelligence. 17 (1-3)185–203. (1981). 10.1016/0004-3702(81)90024-2 [DOI] [Google Scholar]
- [6].Liu T, & Shen L, “Fluid flow and optical flow,” Journal of Fluid Mechanics, 614 253–291 (2008). 10.1017/S0022112008003273 [DOI] [Google Scholar]
- [7].Amini A, “A scalar function formation for optical flow,” (2000). 10.1007/3-540-57956-7_13 [DOI] [Google Scholar]
- [8].Schueler BA, Kallmes DF, Cloft HJ. “3D cerebral angiography: radiation dose comparison with digital subtraction angiography”. AJNR Am J Neuroradiol. 2005. Sep;26(8):1898–901. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8148841/ [PMC free article] [PubMed] [Google Scholar]