Concentric Tube Robot-Assisted Intracerebral Hemorrhage Evacuation: Validation in an Ovine Model

Abstract

Intracerebral hemorrhage (ICH) is a type of hemorrhagic stroke that causes nearly 3 million deaths annually worldwide. Recent clinical trials have indicated minimally invasive surgery (MIS) can improve functional outcomes in patients with lobar ICH. However, despite these promising results challenges persist, namely, tool dexterity and visualization. Previous research has been developing a platform for MR-guided ICH evacuation using a concentric tube robot (CTR), and in this study we present the first-ever in vivo ICH evacuation with an MR-guided CTR. The CTR is a three degree of freedom (DoF) robot mounted to a 4-DoF stereotactic frame. The robot has two non-metallic concentric tubes that are pneumatically actuated. Detailed in this paper are our experimental in vivo workflow, a novel clot production method to be used in ex vivo and in vivo ICH models, and the evacuation outcomes.

I. Introduction

INTRACEREBRAL hemorrhage (ICH) is a type of hemorrhagic stroke caused by the rupture of compromised cerebral blood vessels. These vessels are most often weakened by factors such as hypertension, arteriovenous malformations, or head trauma. Among these, there is a projected 35% increase in the incidence of ICH by 2050 due to hypertension alone [1]. Presently, ICH accounts for over a quarter of all stroke cases worldwide, and it carries a staggering mortality rate of over 50% within the first year [2], [3], [4]. This poor clinical outcome stems from the substantial damage resulting from the pooling blood within the brain parenchyma. This blood forms a clot that increases intracranial pressure and induces significant brain distortion. Additionally, perihematomal neurotoxicity impairs the normal functioning of neurons. The subsequent cortical and neuronal disruption induces severe cognitive dysfunction, curtailing patient independence in daily activities [5].

Historically, treatment of an ICH has focused on medical management. With this approach, patients are monitored in an intensive care unit and pharmacological interventions are administered to reduce secondary complications and support the body’s physiological coping mechanisms. This conservative mode of treatment stems from the debated efficacy of disruptive craniotomies and minimally invasive surgery (MIS). Notably, STICH I (Surgical Trial in Lobar Intracerebral Haemorrhage), STICH II, and MISTIE III (Minimally Invasive Surgery Plus Alteplase for Intracerebral Hemorrhage Evacuation) have failed to consistently demonstrate improvements in functional outcomes when using a MIS approach, fostering reluctance towards surgical intervention [6], [7], [8], [9], [10]. However, due to the dismal prognosis of ICH, surgical endeavors have persisted. Recent success reported in 2023 from the Early Minimally-invasive Removal of ICH (ENRICH) trial [6] has recently confirmed the results by Pantazis et al. [11], indicating MIS can improve functional outcomes if the patient and procedure exhibit key factors [6], [12]. These key factors include an initial hematoma (herein referred to as clot) volume of greater than 50.5 mL, a post-evacuation clot volume below 15 mL, a lobar or basal ganglia location of the hemorrhage, and a mean time to surgery after clinical decompensation of 16.6 hours [6]. While these recent results show promise, several additional challenges persist. These include the restricted dexterity imposed by MIS instrumentation and the limited visualization of the clot when using current visualization techniques [13].

Existing MIS instrumentation for ICH typically relies on a rigid endoscope and aspirator. Consequently, the degrees of freedom are fundamentally constrained by the burr hole drilled into the skull. As a result, the reachable workspace is severely limited, especially for larger clots that exceed 30 to 50 mL [12]. To compensate for this restricted access, surgeons often angle the rigid metal endoscope through the burr hole to widen the operative field [14], [15]. This maneuver increases the risk of tissue damage and produces what is colloquially known as a cone of destruction [14], [16]. Several research groups have attempted to mitigate this problem using continuum robotic systems. These efforts include tendon-driven continuum robots [17], [18], bevel-tipped needles [19], [20], meso-scale SMA actuated continuum robots [21], and concentric tube robots [22], [23]. Among these approaches, concentric tube robots show the most potential for ICH evacuation. They offer elegant modeling, simple and durable device architectures, and the innermost tube can intuitively serve as the aspirator [24].

However, most of these systems have only been evaluated in benchtop environments with external cameras and they lack in vivo demonstrations. When medical imaging strategies are incorporated, they rely on the same visualization modalities used in the current MIS ICH workflow. These modalities include intraoperative endoscopic video or preoperative CT, and each presents important limitations. CT imaging exposes patients to ionizing radiation, which restricts its use to preoperative or postoperative assessment rather than continuous intraoperative guidance [13]. Endoscopic visualization provides real-time feedback, but its field of view is limited and may become obstructed during the evacuation process [25].

To address the aforementioned concerns, previous research has explored development of an MR-conditional platform for MR-guided ICH evacuation. By using MRI as the imaging modality, the patient is not exposed to radiation and visualization of the clot is greatly improved with better 3D soft tissue visualization. The development of the system began with a 3-degree of freedom (DoF) CTR using nitinol tubes and a custom pneumatic motor [26], [27], [28]. The pneumatically driven CTR was selected because it provides practical advantages over cable-driven and magnetically actuated systems. Specifically, pneumatic motors offer a lightweight and MR-safe solution that avoids the friction and backlash introduced by the long cable transmissions required in this setup for cable-driven systems, and they eliminate the MRI radiofrequency noise and image distortion that would arise from magnetic or piezoelectric actuation in the MR-environment. Additional details on the actuator selection can be found in [27], [29]. The system was evaluated in phantom studies where dynamic MR-images were used for surgeon-in-loop evacuation feedback [30]. After validation of the proof-of-concept, the design was improved by reducing the overall size and implementing plastic tubes to enhance dynamic visualization and evacuation efficiency [31]. Following its success in phantom studies that emulate human anatomy [31] and sheep anatomy [32], a stereotactic aiming device was developed to aim the system and it was evaluated in several ex vivo studies [33], [34]. Notably, in both phantom and ex vivo evaluations, the system reduced the clot to a residual volume below 15 mL, demonstrating a promising direction for MR-guided ICH evacuation.

In this paper, we continue to pursue the MR-guided robotic ICH evacuation paradigm by evaluating our system in an MR-guided in vivo ovine ICH evacuation. Through this in vivo evaluation, we demonstrate the clinical feasibility of MR-guided robotic ICH evacuation. Specifically, we show that the system can be integrated into the MRI control room and imaging suite. Notably, the robotic equipment does not spatially interfere with the anesthesia equipment or MRI. Contributions of this manuscript include the following: (i) presentation of the first-ever reported MR-guided ICH evacuation using a concentric tube robot in an in vivo ovine model, (ii) detailed discussion on a method for minimally invasive clot production within an in vivo ovine model using fluoroscopic guidance, and (iii) description of an experimental in vivo workflow for MR-guided ICH evacuation using a CTR.

II. Hardware Overview

The system consists of three primary subsystems, as shown in Fig. 1A. These include (i) a surgical platform for intracerebral hemorrhage robot evacuation (ASPIHRE), (ii) a neurosurgical, interventional, configurable device for effective aiming (NICE-Aiming), and (iii) a control box used for controlling a pneumatic radial inflow motor and encoder (PRIME). In addition to these systems, a method of regulating vacuum from the MRI control room (see Fig. 1B and C) is developed.

Fig. 1.

(A) Robotic system setup in a benchtop setting. The workstation is used to send high-level commands to the control box for control of ASPIHRE. Details on NICE-Aiming and ASPIHRE can be found in [27], [31], [33]. (B) Hand-held evacuation control gun for vacuum regulation in the MR-control room. The vacuum canister remains in the MR-imaging suite next to the provided vacuum supply. (C) Fluidic schematic of the evacuation setup. Only the evacuation control valve is in the MR-control room. Everything else is in the MR-imaging suite.

A. ASPIHRE

ASPIHRE is a two tube, 3-DoF CTR [35]. Each DoF is controlled by PRIME [27] and encoding is enabled using optical encoders (E2, US-Digital, USA). The robot range-of-motion limits are monitored using fiber optic limit switches. The outer tube is a non-metallic tube with 1-DoF (translation) and the inner tube is a nylon tube with 2-DoF (translation and rotation). The outer tube has an OD/ID (outer diameter/inner diameter) of 8/6.35 mm and the inner tube has an OD/ID of 6/4 mm. The inner tube has a pre-curved length of 35 mm and a radius of curvature of 30 mm. In this configuration of the CTR, the system has an insertion depth of 65 mm. This length was adopted to ensure the system would fit within the MRI bore during the in vivo ovine experiments, as the anatomical positioning of the animal and the limited range of motion of its head and neck constrain the available bore space in ways that differ from human use. Note that this insertion depth is sufficient in the ovine model, as the distance from the cranial surface to the skull base is approximately 40 mm. A computer aided design (CAD) representation of both ASPIHRE and NICE-Aiming can be seen in Fig. 2. Note that an obturator is inserted through the inner concentric tube to facilitate gentle insertion through the cortical tissue. The obturator is filled with MR contrast so that appropriate MR slice planes can be selected based on the obturator’s insertion vector.

Fig. 2.

ASPIHRE and NICE-Aiming can be seen disassembled. For additional details, see [33].

B. NICE-Aiming

NICE-Aiming is a 4-DoF remote center of motion (RCM) frameless stereotactic aiming device. The device is manually configured with thumbscrews based on the inverse kinematics detailed in [33]. Two DoF define the RCM point relative to the aiming device’s base and the remaining two DoF perform orthogonal rotations to define the CTR’s insertion vector orientation. Note that while the whole system has 7-DoF, once the aiming device is configured, the system is reduced to the 3-DoF provided by the CTR for position control of the CTR tip. Registration of NICE-Aiming and ASPIHRE to the imaging modality is accomplished using rigid-point registration via fiducials attached to the base of the aiming device (as shown in Fig. 2).

C. Control Box

The control box is a custom-designed housing that contains a set of pneumatic valves (6425K18, McMaster, USA), a custom-designed valve drive circuit, a set of fiber optic transmitters and receivers (IF E22, Industrial Fiber Optic, USA), and a motion controller (DMC-4163, Galil, USA). Inputs to the control box include 115 V outlet power and VGA connections from the motion controller. Outputs include the limit switch fiber optic cables, the pneumatic transmission lines used for controlling PRIME, encoder cables, and an emergency stop switch (as detailed in [33]). The VGA connectors from the motion controller transfer encoder signals, limit switch signals, and valve drive signals. The emergency stop button is used to sever power to the valves, ensuring operation is ceased in the event of an emergency. In this system, the motion controller monitors the robot state in real time while a computer workstation is used to send high-level commands that define the setpoints of the axes.

D. Evacuation Regulation

In this system, a vacuum canister is connected to the inner tube of the CTR to evacuate the clot. The canister is connected to a vacuum regulator located in the MR-imaging suite. Herein, we developed a manually actuated control valve that can be opened and closed within the MR-control room that is also connected to the canister in the MR-imaging suite, as shown in Fig. 1B. In this setup, when the valve is open, the fluidic resistance across the valve is significantly lower than the fluidic resistance associated with suction of the clot through the inner tube of the CTR. This effectively eliminates suction at the tip of the CTR. Details on the fluidic modeling can be found in [36].

III. Workflow and Results

In this study, a large in vivo ovine model was used to evaluate the efficacy of ASPIHRE and NICE-Aiming. This model was adopted as it emulates the cranial mounting (in terms of size and skull thickness) and the brain anatomical characteristics of humans, albeit with a much smaller volume [37], [38]. Further, the MR contrast parameters for multicontrast imaging of large in vivo ovine models are similar to those in human brain imaging. The ICH herein is intentionally induced.

A. Pre-Clinical Decompensation Setup in MRI

At 7:30 a.m., prior to preparation of the in vivo ovine model, a mobile cart was prepared and transported to the MR control room. This cart contained the custom-designed control box, the pneumatic compressor (8010, California Air, USA), and the pneumatic and fiber optic transmission lines for controlling ASPIHRE. After locating the cart in the control room, the pneumatic and fiber optic transmission lines were routed through the MR waveguide and the electronics were provided power through an outlet on the wall. Notably, the cart can be prepared well before clinical decompensation and can remain within the MR control room for prompt implementation. Further, the transmission lines are MR-conditional and can remain routed through the waveguide if permitted. Additionally, during this time, the fiber optic temperature probe was setup in preparation for the in vivo evacuation. The total duration for setup inside the MR control room was approximately 1 hour.

B. Animal Preparation

The in vivo ovine model was prepared for surgery from 9:00 a.m. to 11:04 a.m. The in vivo ovine model was male and had a mass of 59.1 kg. Note that the in vivo ovine model was fasted for 40 hours prior to the surgical procedure to reduce the risk of bloating when anesthetized. The in vivo ovine model was anesthetized via inhalation using a precision vaporizer with 2–2.5% isoflurane and 40% oxygen. Following general anesthesia, the in vivo ovine model was intubated and placed on mechanical ventilation. Heart rate, respiratory rate, end tidal CO2, SPO2, and body temperature were monitored during the procedure and recorded in 15 minute intervals. The in vivo ovine model was placed in the sternal recumbency position and subsequently strapped to the MR-safe transit board. Note that a water-circulating warming pad was used to maintain body temperature. All anatomical surgical sites were subsequently shaved and aseptically prepped. Venal catheterization was performed for administration of pharmaceuticals and arterial catheterization was performed for blood collection, which will be used for producing the clot. Both were placed percutaneously in the ear.

Following the above, the head of the in vivo ovine model was fixed to the fixation head frame and a craniotomy was performed by our neurosurgeon from 11:05 a.m. to 11:27 a.m. The craniotomy consisted of a 5 cm incision made into the skin over the target site along the medial plane. Through gentle dissection, the cranial surface was exposed by retracting the fascia and muscle. Sutures were used to maintain a clear exposure of the cranial surface. A 12.7 mm burr hole was produced using a drill and Kerrison Rongeur forceps were used to shape the margin of the burr hole window. Following the craniotomy, the dural layers were incised using a cruciate incision to reveal the cortical surface beneath. We drew an autologous 15 mL arterial blood sample from the arterial catheter and transferred it into a heparinized syringe. The blood was mixed with X-ray contrast in preparation for the clot production procedure.

C. Clot Production

Following the preparation of the surgical site and craniotomy, a clot was produced under fluoroscopic guidance from 11:28 a.m. to 11:55 a.m. Earlier studies such as the one detailed in [37] have used direct injection of blood into the brain to create an intracerebral clot. However, in our experience with ex-vivo models, this approach offers limited control on the extent and shape of the generated clot. Here, we propose a new procedure to generate a well-defined intracerebral clot. We have validated this method earlier in ex vivo models using endotracheal tubes [20], [34], but here we provide a detailed workflow and use pediatric Foley catheters.

This procedure used a 10 mL pediatric Foley catheter (57–0165PL10) with a metallic insert (as shown in Fig. 3A and D) to create a well-circumscribed intracerebral cavity in which the blood sample will be injected to generate the clot. To create the cavity, the Foley catheter was inserted via the burr hole. Depth graduations on the catheter were used to insert it to the desired clot location determined by fluoroscopic imaging. Following placement of the catheter, a 10 mL sample of water was injected into the balloon of the catheter over the course of 10 minutes (see Fig. 3B and E). The balloon remained inflated for a 10-minute holding period. After the holding period, the water was removed and the balloon was deflated. Fluoroscopic imaging was performed to verify the formation of the intracerebral cavity. Following this, the Foley catheter was retracted so that the tip of the catheter was at the boundary of the intracerebral cavity. The ICH was then generated by injecting the autologous blood through the catheter’s central port until blood was seen ejecting out of the burr hole (13 mL). This injection procedure was drawn out over the course of 5 minutes. Fluoroscopic imaging was used to intermittently monitor the hemorrhage formation, as shown in Fig. 3C and F. After clot formation, the catheter was removed slowly over the next 2 minutes and the aiming device was mounted. This procedure was only performed once. There was not a need for additional burr holes or injections. The mounting procedure lasted from 11:56 a.m. to 12:00 p.m. During clot production, the in vivo ovine model tolerated the procedure well and no sustained problematic changes in vital signs were observed. This suggests that though the catheter does cause damage to the tissue through the path it cuts, it did not affect stability of the in vivo ovine model for purposes of the experiment.

Fig. 3.

(A) Image of the pediatric Foley catheter after insertion into the brain. (B) Image of the pediatric Foley catheter after inflation. (C) Image of the clot in the brain after removal of the pediatric Foley catheter. (D) Annotated image of (A) highlighting the pediatric Foley catheter. (E) Annotated image of (B) highlighting the inflated balloon of the pediatric Foley catheter. (F) Annotated image of (C) highlighting the clot after removal of the pediatric Foley catheter.

D. Animal Transit and Setup

After the clot production, the in vivo ovine model was prepared for transit from 12:01 p.m. to 12:30 p.m. Specifically, the in vivo ovine model remained securely strapped to the transit board and the transit board was moved onto a MR-conditional mobile cart. After placement onto the mobile cart, the mobile anesthesia equipment was connected to the in vivo ovine model for monitoring and maintaining general anesthesia. During transit from 12:31 p.m. to 12:42 p.m., five personnel were used: one for ensuring an unobstructed route, one for driving the anesthesia cart, one for ensuring the anesthesia and monitoring equipment cables were not tangled, and two for steering the MR-conditional mobile cart. Note, during transit, ASPIHRE was connected to the control box and normal operation of the system was confirmed in the MR control room. The anesthesia cart and transport board can both be seen in Fig. 4A-B.

Fig. 4.

(A) The transport board and head frame used for transit and restraint of the in vivo ovine model can be seen. Note that the entire setup is MR-safe. (B) The anesthesiology cart used for transit and the procedure can be seen. Additionally, the cart for robot control is depicted next to the anesthesiology cart. Each system uses a different waveguide.

At 12:43 p.m., the in vivo ovine model arrived at the MRI suite and setup begin. First, the anesthesia conduit was routed through the MR waveguide and connected to the in vivo ovine model. Additionally, the fiber optic temperature probe was attached to the in vivo ovine model. The in vivo ovine model was then placed onto the MRI bed and ASPIHRE was mounted to NICE-Aiming. The robot mounting can be seen in Fig. 5A-B. At 1:12 p.m., the in vivo ovine model was setup inside the MRI bore with ASPIHRE and initial imaging began. Imaging parameters were tuned and robot registration was performed using rigid-point registration with 3-D T₁ weighted TFE (turbo field echo) (FOV: 192×192×200 mm³, 1.0×1.0×1.0 mm³ resolution, TR/TE/TI = 13/4/1060 ms, flip angle = 8°, TFE factor = 150) [39].

Fig. 5.

(A) ASPIHRE and NICE-Aiming can be seen up close while mounted to the in vivo model. (B) The system can be seen setup with the in vivo model inside the MR-imaging suite prior to placement into the MRI bore.

E. Intraoperative Imaging and Evacuation

The MR-guided robotic ICH evacuation procedure began at 2:17 p.m. A preoperative imaging scan of the clot was performed using 3-D T₂ weighted imaging (FOV: 192×192×200 mm³, 1.0×1.0×1.0 mm³ resolution, TR/TE = 2500/235 m, flip angle = 80°, TSE (turbo spin echo) factor = 120). The clot had a volume of 1.91 mL, as depicted in Fig. 6A and Fig. 6C. Note that the clot volume of 1.91 mL is significantly less than the injected 13 mL; a discussion on this discrepancy will be provided in Section IV. During the procedure, dynamic images at a region of interest provided surgeon-in-loop feedback. Images were obtained using three orthogonal slices across the ICH volume. Each slice was 2-D TFE (FOV: 180×180 mm², slice thickness = 2 mm, 1.0×1.0 mm², TFE factor = 5, flip angle = 75°, TR/TE = 7.5/3.7 ms, 3 s per image), and they were viewed in a MATLAB interface. At the beginning of the procedure, the tubes of ASPIHRE were incrementally inserted to a depth of 25 mm while the obturator was included. Following this insertion, the obturator was removed and the evacuation line was connected to ASPIHRE. After the evacuation line was attached and engaged, robot axis commands were performed based on surgeon-in-loop feedback until the clot was evacuated. Following evacuation, 3-D T₂ weighted imaging (FOV: 192×192×200 mm³, 1.0×1.0×1.0 mm³ resolution, TR/TE = 2500/235 m, flip angle = 80°, TSE factor = 120) was performed for visualizing the final clot volume. An example slice and the segmentation of the final clot volume can be seen in Fig. 6B and Fig. 6D. The residual clot volume was 0.56 mL, which is approximately 28.9% of the initial clot volume. The evacuation procedure and imaging ended at 3:25 p.m. and the in vivo ovine model was euthanized using pentobarbital sodium.

Fig. 6.

A T₂-weighted image of the clot (A) prior to and (B) after evacuation can be seen. An image of the clot (C) prior to and (D) after evacuation can also be seen with the segmented clot voxels identified and super-imposed on the image. The segmentation was performed manually using Slicer [40], [41].

IV. Discussions

In this work, we demonstrated for the first time a MR-guided ICH evacuation using a concentric tube robot in an in vivo ovine model. The surgical intervention included evacuation of a 1.91 mL hematoma that was induced in a novel minimally invasive manner via a pediatric Foley catheter. Approximately 71.1% of the initial hematoma volume was removed. In terms of relative scale, the residual volume of 0.56 mL is consistent with the proportionally smaller total brain volume in the ovine model and aligns with clinically meaningful reduction thresholds when adjusted for species size [6]. Notable contributions in this study include: (i) proof-of-concept evaluation of our robotic system in an in vivo study, (ii) a minimally invasive method for producing a hematoma under fluoroscopic guidance, and (iii) an experimental workflow for ICH evacuation using a CTR with MR-guidance. While the in vivo ovine model survived the entire procedure and the system implementation was feasible, there are still several improvements that can be made, both in the system and the experimental in vivo workflow.

One area that necessitates improvement is the clot production method. While the diameter of the clot imaged in the fluoroscopic imaging matched the diameter of the expanded pediatric Foley catheter, it was observed that the injected clot volume (13 mL) did not match the segmented volume that was imaged in the MRI (1.91 mL). This issue did not arise in our ex vivo studies [33], [34]; however, there are several factors that may have contributed to this discrepancy in the in vivo study presented here. First, in the ex vivo studies, the sample was frozen prior to experimentation. While an effort was made to thaw those samples, they still remained well below body temperature which likely contributes to reduced tissue elasticity. Further, the tissue likely stiffened independent of temperature post mortem. Conversely, in this study, there is a significantly higher observed elasticity in the cortical tissue in the in vivo ovine model. Consequently, once the pediatric Foley catheter is removed, the entry path acts as a low resistance pathway for the injected blood to eject. This ejection is likely due to the high intracranial pressure induced by the deformed elastic tissue. While this ejection may be mitigated if the blood is permitted sufficient time to coagulate, in this workflow the blood was heparinized. To improve the clot production workflow, we will implement two notable changes in the future. First, we will not heparinize the blood to facilitate clotting. Second, instead of removing the pediatric Foley catheter immediately after clot production, we will leave it in place during transit to serve as a tamponade. This will prevent the blood from ejecting out of a low fluidic resistance path. Implementing these two changes should improve the likelihood that the clot volume imaged at the MRI is equal to the injected volume.

Another operational concern regarding the tissue elasticity of the in vivo ovine model is the rapidly changing hemorrhage boundary. To reduce the likelihood of evacuating healthy brain parenchyma, we plan to develop an algorithm that approximates the hematoma shape given a small set of dynamic slices. From this volume, we will define virtual fixtures that restrict the motion of the concentric tube robot to avoid applying suction to the changing clot boundary as it is evacuated. Additionally, we aim to implement an electronically controlled valve that will regulate the vacuum applied, preventing the sudden application of high vacuum that occurs when our currently manually actuated valve is closed.

Finally, one area that warrants additional investigation is the development of real-time tracking of the tip of ASPIHRE. Currently, during the initial insertion phase, the contrast filled obturator serves as an identifiable landmark to interpret the tip position of the CTR. However, after the obturator is removed, the tip position of the CTR is dependent upon accurate modeling of the mechanics and kinematics of the inner tube. In the past, we have implemented wireless tracking using compact tracking coils, as detailed in [42]. However, due to the size of the coil, it was limited to interventional devices that could accommodate the large size without concern for an obstructed inner lumen such as biopsy needles. As we continue to develop the system, we will consider improving the size of our wireless tracking coil modality by reducing its overall footprint, and identify a suitable method of integrating it with the CTR that does not obstruct the evacuating lumen.

Although the present work was performed in an ovine model, the system has a clear translational pathway toward human use. For example, the stereotactic frame will likely not require major changes as it was designed with the workspace demands of human use in mind. However, the insertion depth of the system will need to be increased to reach deeper hemorrhage locations in human studies. The current shorter configuration was selected to ensure that the system fit within the MRI bore during the in vivo ovine studies, where the animal’s anatomical positioning and the reduced range of motion of the head and neck limit the available space in a way that differs from human imaging conditions. This limitation is acceptable in the ovine model, given that the distance from the cranial surface to the skull base is approximately 40 mm. Conversely, human anatomy is more accommodating of a longer CTR, which is needed for the increased depths. Specifically, since the human can be operated on in a supine position, and the head can be tilted more dorsally than a sheep, the long axis of the CTR has a larger component projected onto the length of the bore, avoiding interference with the upper part of the bore. Additionally, we will also need to explore optimal concentric tube curvatures and configurations for large volume hematomas. A preliminary study for this can be found in [43]. The planned human cadaver studies will provide the next critical step in the translational process by validating these design modifications and identifying additional clinical constraints prior to future in-human evaluations.

Overall, we believe this study demonstrates the feasibility of MR-guided ICH evacuation using a CTR. Future work will involve investigating the aforementioned modifications and improving the system’s reliability by including appropriate redundancies, such as redundant tracking and encoding [44]. We will also begin evaluating the system in (i) comparative studies that attempt to demonstrate the improved evacuation outcomes compared to existing approaches and (ii) human cadaver studies to identify possible shortcomings that can not be observed in animal studies. Finally, we will also consider clinical integration, including system sterility and reusability, preferred system mounting location, and streamlined system setup in human studies.

V. Conclusion

In this study, we report the first-ever in vivo intracerebral hemorrhage ICH evacuation with an MR-guided CTR. Notably, we show that this interventional procedure is feasible in the MRI bore and have provided an experimental in vivo workflow. Additionally, we provide a method of producing an ICH clot in a minimally invasive way that can be used for emulating in vivo ICH conditions. The current work presents a promising step toward improving ICH patient outcomes through the use of dexterous robotics. Future work will involve improving the workflow, enabling tracking, and developing rapid image segmentation techniques that enable implementing virtual fixtures. Additionally, we will perform accuracy evaluations with the system in an in vivo study.