Conceptual Design and Procedure for an Autonomous Intramyocardial Injection Catheter

Abstract

This article discusses existing catheter systems and proposes a conceptual design and procedure for an autonomous cell injection catheter for the purpose of transferring committed myogenic or undifferentiated stem cells into the infarct boundary zones of the left ventricle. Operation of existing catheters used for cell delivery is far from optimal. Commercial injection catheters available are handheld devices operated manually by means of tip deflection and torque capabilities. Interventionists require a hefty learning curve and often encounter difficulties in catheter stabilization and infarct detection, resulting in lengthy operation times and nonprecise injections. We examined current technologies and proposed a design incorporating robotic positional control, feedback signals, and an adaptable operational sequence to overcome these problems. The design provides the basis for robotic catheter construction that is able to autonomously assist the physician in transferring myogenic cells to the left ventricle infarct boundary zones.

Introduction

In 2012, the World Health Organization (WHO) estimated that cardiovascular disease (CVD) was the leading cause of death affecting 17.5 million people across the globe. Of these fatalities, 7.4 million were the result of ischemic heart disease¹. Myocardial infarction (MI), a potential consequence of myocardial ischemia, signifies cardiac muscle damage where the cells undergo necrosis and apoptosis², eventually forming a permanent scar on the ventricular wall. With scar tissue having low electrical conductivity and thus being unable to contract, the heart's functionality and ability to pump efficiently are deteriorated, likely leading to congestive heart failure.

The ideal treatment for MI is still a developing science. The traditional approach of heart transplantation for patients with end-stage heart failure raises the issues of donor shortages and postoperative complications such as the immune system's rejection of the transplant, cardiac allograft vasculopathy (CAV), infection, and side effects caused by medication. Furthermore, with the scarcity of heart donors, candidates for cardiac transplants are filtered, rendering a portion of the potential recipients ineligible for operation. Given the mortal significance of MI and the challenges faced with current treatment plans, extensive research has been invested in novel treatment methods.

In recent years, cell transfer therapy has undergone significant progress and has been studied extensively in clinical trials^3,14. These types of transferred cells consist of embryonic stem cells (ESCs), induced pluripotent stem cells (iPSCs), human umbilical cord cells, fetal cardiomyocytes, skeletal myoblasts, resident cardiac stem cells, bone marrow-derived mesenchymal stem cells (BM-MSCs), and mesenchymal stem cells (MSCs). The underlying principle of committed myogenic or undifferentiated stem cell transfer is in the regeneration of myocardial tissue when the cell solutions are delivered to an infarct area. The method of transfer for these reparatory cells to the damaged myocardial site may be accomplished by means of intravenous (peripheral) infusion, subcutaneous cytokine injection, coronary sinus infusion, intracoronary infusion, and direct intramyocardial injection using an intracardiac catheter (transendocardial) or through open chest surgery (transepicardial).

The method of interest in this article, intramyocardial injection, allows stem or myoblast cells to be directly injected into the infarct boundaries. Transepicardial injections are accomplished by means of thoracotomy. However, thoracotomy is highly invasive and can lead to complications such as systemic embolization, infection, (possibly chronic) postoperational pain, a prolonged postoperative recovery period, pneumonia, and air leaks^15,16. Consequently, significantly less invasive catheterization techniques are highly favorable in comparison. Transendocardial injections introduce the cells into the endocardium by means of an intracardiac needle catheter. Yet, despite the existence of several cell injection catheters, the optimal catheter system and procedural method for transendocardial injection has not been established. Existing systems present various obstructions such as lengthy operation times, imprecise positioning, potential injury of the myocardium during manipulation, risk of exposure during fluoroscopy, and/or vigorous physician training requirements^3,17.

Apart from needle injection catheters, the past decade has also brought about advances in remote catheters. A remote catheter not only allows finer control but also permits the physician to conduct and observe the procedure at a distant workstation, safe from fluoroscopic radiation exposure. As such, we briefly review current needle and remote catheter designs and propose a design and procedure for a remotely controlled, needle-based catheter. Extending this concept a step further, the remote catheter system also has varying degrees of autonomy where surgical tasks are performed through the robotic system rather than the catheter interventionist. In this article, autonomy refers to the catheter system as being self-governed where parts or all of the operation is completed without human participation. The end objective is to create a practical yet effective robotic catheter system that can autonomously perform a transendocardial cell injection procedure in the left ventricle (LV) for the transfer of reparatory cells into the infarct boundary zones.

Design Scope

An autonomous robotic cell injection catheter can provide an effective and time-efficient, yet minimally invasive, method of injecting committed myogenic or undifferentiated stem cells into the infarct boundaries of the L.V. Benefits of a robotically controlled catheter over a manual catheter include having a higher degree of accuracy, requiring the same or a shorter learning curve for the operator, having enhanced stability, and relying less on the experience of the operator for a successful operation. Drawbacks of robotic catheters include the higher cost, longer preparation time, lack of physician-to-patient interaction, and more frequent vascular complications at the groin if a larger sheath size is used. While effective, robotic catheters that incorporate magnetic navigation face additional problems including the higher initial cost, lack of portability, complications with implants such as pacemakers and defibrillators, possibility of inadequate contact with a weak electromagnetic field and soft catheter tip, and shielding requirements for the operation room^18,19.

The transendocardial transfer method in this design focuses on, but is not limited to, myoblast transfer therapies. A myoblast concentration of 10⁸ cells/ml and total volume of 10 ml are used in each patient. The high concentration and volume are necessary due to the initial cell rejection apparent in myogenic cell therapies²⁰. The catheter system should also be designed to maximize cell retention where the proper injection method, flow rate, and needle gauge must be considered. The LV can only be pierced by the needle a maximum of 20 times before the possible induction of heart arrhythmia. Therefore, in a procedure with 20 injections, each injection should emit 0.5 ml of 10⁸ cells/ml myoblast solution^21,22.

Given the low cell retention rate in cell delivery techniques²³, it is desirable to design a catheter that can maximize the efficacy of each injection. A study on the regulation of cell distribution and fusion via myoblast injection methodology showed that oblique injections (i.e., cells injected into muscle fibers at a diagonal angle while retracting) yielded superior cell distribution and a higher cell fusion rate compared to other injection methods²⁴. Myoblasts that were injected perpendicularly into muscle fiber while the needle was retracting showed a partial distribution of cells, whereas myoblasts that were injected longitudinally or only in one spot showed poor distribution. Accordingly, the autonomous cell injection catheter should deliver the solution in a manner where cell distribution and fusion are optimal. The catheter should be able to insert the needle either perpendicularly or obliquely into the endocardial wall and inject the cell solution while it is retracting. As the wall thickness of an infarcted LV myocardium can be as thin as 5 mm, it is also vital to avoid penetration of the myocardium during injection.

The guideline to this design also emphasizes on minimizing the monetary cost of the catheter, reducing operation time, and reducing the necessary learning curve for the physician. The catheter system should be spatially aware of its position in the LV, and it should avoid causing any myocardial, chordae tendineae, or valvular injuries. Autonomy of the system also includes the ability to identify infarct boundary zones. Upon cell injection, the catheter should be able to maintain full, continuous contact with the myocardial wall, be aware of possible slippage, and ensure needle penetration. As cell retention is vital, backflow of the injectate into the ventricular cavity should be minimized. Furthermore, overlapping of injections should be prevented.

With the novelty of an autonomous device and as no similar catheter systems presently exist, the design may include several variations to yield a wide array of modules. Where possible, the catheter should be made to be compatible with existing systems or technologies, and manual or remote robotic control should be available if autonomy cannot be achieved.

Catheter Systems

Several endocardial injection catheters exist for the purpose of cell delivery in myocardial repair. Remote catheter systems that use robotic or magnetic control have also been established, although they are designed for catheterization procedures that lie outside of cell transfer therapy. These systems provide a foundation upon which an autonomous robotic cell injection catheter can be built. This section presents an overview of existing remotely controlled and endocardial injection catheters.

Endocardial Injection Catheters

As currently known, there are five percutaneous intramyocardial catheters capable of direct endocardial cell injection that have undergone clinical trials: MyoCath® (U.S. Stem Cell Inc., Sunrise, FL, USA), Myostar® (Biosense Webster, a Johnson & Johnson company, Diamond Bar, CA, USA), Stiletto® (Boston Scientific, Natick, MA, USA), Helix® (BioCardia Inc., San Carlos, CA, USA), and C-Cathez® (Celyad, Mont-Saint-Guibert, Belgium). The five catheters use the transfemoral approach to access the LV where the injectate is introduced via the endocardium. These aforementioned devices have been referred to in several reviews and textbooks^3,6,25-28.

MyoCath® may perform multiple injections within the myocardium wall where the needle extends and retracts from the catheter tip by means of a manual control mechanism at the proximal end. It is specifically designed to deliver U.S. Stem Cell Inc.'s (formerly BioHeart Inc. prior to a reverse stock split) MyoCell® product, which is a cultured solution composed of myoblast cells. MyoStar® is similar to MyoCath® in its maneuverability, using rotation and deflection with variable curve lengths to control the catheter. It is also composed of an integrated system with the support and core catheter as a single unit. Both catheters lack the presence of a guidewire lumen and hence require the use of a navigation system to advance the catheter from the femoral artery to the LV. MyoStar® can be used in conjunction with radiographic imaging or an electromechanical mapping (EEM), as in the 3D NOGA® electromagnetic cardiac mapping system (Biosense Webster). It also has been used in conjunction with the Stereotaxis Inc. (St. Louis, MO, USA) magnetic navigation system to remotely guide the NOGA mapping system in porcine trials²⁹.

Unlike the two former catheters, the core and supporting catheters for Stiletto® are separate units consisting of two steerable, preshaped support catheters. The tip orientation is manipulated through steering the guide catheter. An additional independently controlled component is the inner, spring-loaded nitinol needle with a fixed extension length of 3.5 mm. Stiletto® requires the use of fluoroscopy, although studies have also shown it could be adaptable to be feasible with magnetic resonance (MR)-guided techniques^30,31.

The Helix® catheter system also consists of a core unit and a separate supporting guide catheter. The core Helical Infusion Catheter is radiopaque for fluoroscopic imaging, and two fluid ports exist at the proximal end for the therapeutic injectate and contrast solution to confirm needle anchorage. The helical needle provides active fixation at the endocardium by means of rotating the needle into the injection site. The infusion catheter is used in conjunction with BioCardia's Morph® deflectable guide catheter, where it accesses the LV by passing over a guidewire.

The C-Cathez® injection catheter possesses a preshaped nitinol needle curved at 75® with perfusion holes at the side that reduces interstitial pressure and eliminates backflow, leading to an approximately threefold increase in cell retention³². The tip of the catheter may be deflected in a single direction by a thumbwheel located at the handle, and the needle length is altered by an adjustment dial. C-Cathez® comes with an aortic arch simulator and ruler to assist in measuring the protrusion length of the needle.

Remote Catheter Systems

Remotely controlled robotic catheter systems that have undergone clinical trials or are commercially available include the Niobe® magnetic navigation system and the Vdrive® robotic navigation system (Stereotaxis), the Sensei®< robotic system and the Magellan® robotic system (Hansen Medical, Mountain View, CA, USA), the CorPath® system (Corindus Vascular Robotics, Natick, MA, USA), the Amigo® (Catheter Robotics, Budd Lake, NJ, USA), and the catheter guidance control and imaging (CGCI) system (Magnetecs, Inglewood, CA, USA)^33,45. Remotely controlled catheters may be separated into two categories: magnetic and active catheters. Magnetic catheters control the catheter tip using a magnetic field, while active catheters are manipulated through a pull-wire system.

The Niobe® magnetic navigation system (MNS) is capable of controlling the position of a catheter inside a patient's body by means of two permanent magnets that are mounted on mobile pivoting arms on each side of the table. The catheter tip must be embedded with small magnets or contain a microguidewire with a magnetic tip in order to be manipulated by the magnetic field of 0.08-0.1 T generated outside of the body. The system is accompanied by a modified C-arm fluoroscopy unit allowing for single-plane imaging where right anterior oblique (RAO) and left anterior oblique (LAO) X-ray images may be taken and stored. Coupled to the MNS and X-ray is the navigational software, Navigant® (Stereotaxis), which displays a graphic user interface at a workstation where the operation could be controlled by a touchscreen, mouse, or joystick. The software system is typically integrated with CARTO® 3 System RMT (Biosense Webster) to provide enhanced navigational services. Through Navigant®, the RAO and LAO images can be combined to produce a virtual 3D map of the coronary anatomy, and the orientation of the catheter can be established by adjusting a magnetic field vector. The response time of the catheter to physician-given commands is 125 ms. Manipulation of the catheter to the target site tends to require two to three attempts to reach an optimal result where each manipulation can be performed in under 20 s^33,34

Advancement and retraction of the catheters, however, are accomplished by the Vdrive® robotic navigationsystem (RNS) and not the Niobe® MNS. The Vdrive® RNS can be mounted to the rail of the operation table and uses custom-designed disposable components to accept various catheter handles. One custom-designed component is the V-Sono™ (Stereotaxis) catheter manipulator, which controls the SoundStar™ catheter (Biosense Webster) and the ACUSON AcuNav™ ultrasound catheter (Siemens, Munich, Germany). The V-Loop™ (Stereotaxis) variable loop catheter manipulator controls the LASSO™ 2515 circular mapping catheters and the LASSO™ 2515 NAV catheters (Biosense Webster). Last, the V-CAS™ (Stereotaxis) catheter advancement system provides simultaneous control to the magnetic catheter body and a standard fixed-curve sheath. Setup of the device first involves the insertion of the catheter through a support tube and then through a standard sheath where it is locked and held in place by the handle clamp tray. The support tube is then attached and locked in place to the Vdrive® where the adjustable arm containing the Vdrive® is also locked. Catheter control through the Vdrive® system is achieved using a remote controller that allows for the advancement, retreat, deflection, rotation, or change in loop size diameter of the catheters³⁵.

The Sensei® X2 robotic system is compatible with both the CARTO® and EnSite Velocity® (St. Jude Medical, Saint Paul, MN, USA) 3D mapping systems and is capable of integrating multiple imaging or recording information from the catheterization procedure including sources such as 3D mapping, intracardiac echocardiography (ICE), fluoroscopy, and electrocardiogram (EKG). The Artisan® Extend control catheter (Hansen Medical) consists of two steerable catheter guides and works in conjunction with the Sensei® X2 robotic catheter system through a robotic catheter manipulator (RCM) to perform electrophysiology procedures. The Artisan® Extend has an inner guide capable of 275® of deflection and may be deflected in all directions with four pull wires. The outer guide is capable of 90® of distal deflection and 8® of proximal deflection. The RCM manipulates the Artisan® Extend using mechanical actuators, pulling on four internal pull wires of the inner sheath and two pull wires on the outer sheath. The catheter may also be moved forward and backward by means of an actuator. An included feature is the Intellisense® Fine Force Technology, which returns contact force feedback from the catheter tip. If the contact force exceeds a certain limit (typically 30 g), catheter advancement is prevented, and the physician is alerted through an optical alarm displayed on the user interface, thereby preventing perforation. Knowledge of the catheter tip contact force also assists in improving the efficacy of the operation through enhancing the intracardiac mapping accuracy and ensuring dense ablation lesions^34,36-39.

The Magellan™ robotic system is designed to deliver stable and precise distal control of the catheters for peripheral vascular procedures. The robotic system consists of a remote physician console where the physician may work at a safe distance from the fluoroscopic radiation and comes in three sizes (6, 9, and 10 Fr). These catheters are controlled by a mobile robotic arm that may be locked onto the patient table. The 6-Fr catheter guide is suitable for procedures that require smaller diameter access and procedures in smaller vessels. The distal end is capable of robotically deflecting at two sections along the guide and also contains an inner lumen that allows for the passage and control of a guidewire. The 9-Fr catheter guide has an inner lumen that gives access to a 6-Fr catheter leader. Both the guide and leader have deflection and rotation capabilities. The catheter leader also contains lumen for a guidewire that may be controlled by advancement, retraction, or rotation. Each of these devices may be independently and robotically manipulated. The 10-Fr catheter guide has similar capabilities as the 9-Fr guide, but consists of a slightly larger lumen, allowing access for devices requiring a 7-Fr inner lumen guide.

The CorPath® vascular robotic system is used for percutaneous coronary interventions (PCIs) by remote control of coronary guidewires and balloons or stents. CorPath® consists of an interventional cockpit at the foot of the bed where the physician may observe angiographic and hemodynamic information or images through computer monitors. The interventional cockpit is lead-lined to protect the physician from radioscopic exposure, and a study found the radiation exposure to the primary operator to be 95.2% less compared to traditional procedural methods⁴⁰. The cockpit also has a control console where the operator uses two joysticks and a touchscreen to control a robotic drive that maneuvers the guidewires, balloons, or stents. At the patient table, the robotic drive is mounted to an articulating arm attached to the bedrail. The invasive devices are loaded onto a single-use cassette, which is then attached to the robotic drive. The devices are loaded by, first, manually connecting a guide catheter to a Y-connector, which is placed in a Y-connector holder in the cassette. Second, the guidewire is introduced via the Y-connector and fixed to the cassette. Last, rapid exchange devices, such as a coronary angiography balloon or stent delivery system, may be loaded and unloaded as necessary during the procedure. Through this robotic system, translational movement of the devices is achieved at 1-mm increments, and the guidewire may be rotated at 30° increments. The guidewire may be moved by axial and rotational transmission forces, while the angioplasty devices are manipulated axially by a set of rollers^41,42.

Amigo® is designed to facilitate the control and positioning of cardiac electrophysiology (EP) catheters. It is a two-component system consisting of the Amigo® remote catheter system (RCS) and the Amigo® remote controller. The Amigo® RCS is a remotely controlled mechanical device containing a sled, track, and turret, which may be attached to the rails of the EP table. The RCS comes with a sterile, disposable kit consisting of a docking station to attach the catheter to the RCS and a spreader to guide the catheter into the track. The kit is compatible with Blazer™ catheter handles (Boston Scientific) and EZ Steer® handles (Biosense Webster). Two other disposable kits are also used—a sterile cover kit and a track kit to maintain sterility. The remote insertion and withdrawal of the catheter are accomplished by pressing one of two buttons on the Amigo® remote controller, which allows an actuated sled to move forward or backward. The catheter may also be remotely rotated by turning a knob at the end of the remote control or deflected by turning a dial in the middle of the remote. The RCS allows for bidirectional deflection, indicating a two pull-wire system^43,44.

The CGCI system uses magnetic manipulation to perform ablation procedures for cardiac arrhythmias or, more commonly, atrial fibrillation. Imaging systems observable at the operation console include a real-time electroanatomical mapping system, an EP recording system, ICE, and a fluoroscopic C-arm. Unlike other MNSs, altering and shielding the operating room are unnecessary as CGCI contains its own shielding encasement. It uses eight coiled electromagnets with a magnetic field strength of 0.1-0.2 T, and it is magnetically inert when its current-regulated amplifiers are turned off. The magnetic field can be manipulated to provide torque and force to the distal end of the catheter where three permanent magnets are embedded. The catheter is also attached to a linearly actuated advancement mechanism. The operator can remotely control the catheter through a joystick or by pointing and clicking target locations on a 3D cardiac map. In the latter method, the CGCI system automatically determines a pathway for the catheter to travel until continuous contact is made with cardiac tissue. Continuous contact and any possible slippage of the catheter tip are indicated by a tissue contact-sensing filter^43,45.

Catheter Design

The ensuing section outlines the essential components of the catheter system consisting of a robotic handle, a catheter guide, and an operating catheter containing the injection needle, recording electrode, contact force sensor, and active pull wires.

Conceptual Ensemble

The catheter tubing comes in two parts consisting of an outer catheter guide (OCG) with a lumen large enough to act as a stabilizing mechanism and passageway for an inner operating catheter (IOC) to pass through. The distal end of the IOC consists of four main components to allow a feedback-driven control system: a recording electrode at the catheter tip to detect electrophysiological signals, four orthogonal pull wires for tip deflection, a fiber optic-based force sensor to measure the contact force of the tip, and a needle to inject the cell solution. Optionally, an electromagnetic position sensor may be added to assist in determining tip location given the appropriate navigation system. The force sensor and electrodes at the distal end (catheter tip) relay feedback signals to the promixal end of the catheter where a microprocessor sits within the catheter handle. The microprocessor responds to the feedback signals by controlling a series of actuators that dictate the movement or deflection of the catheter and needle. An additional lumen also runs from the proximal to the distal end where heparin solution can be injected near the needle opening of the catheter tip in order to prevent thrombosis during operation. These components of the IOC will be further elaborated below.

At the proximal end of the operation, a robotic arm is locked to the rail of the patient's bed (Fig. 1). The arm contains the mechanical actuators that control the movement and operation of the catheter. Female connectors on the arm connect the electrode lead cables and fiber optic sensor to the circuit board and microprocessor situated inside the arm. A plastic, disposable connector unit may assist in attaching the catheter handle to the arm. The use of a connector unit is convenient if the arm is to accept structurally varying catheter handles. As the robotic arm has only one fixed shape, multiple connector units of different shapes may be manufactured to suit different catheter handles.

Figure 1.

(a) Structures and devices used to hold and manipulate the catheters during operation: 1, robotic arm controlling the inner operating catheter; 2, disposable connector linking the inner catheter to the robotic arm; 3, inner operating catheter; 4, stabilizing arm used to hold the outer catheter guide in place; 5, outer catheter guide. (b) Larger view of the robotic arm: 6, female connectors for the electrode, optical fibers, pull wires, needle, and heparin channel; 7, mobile platform in the arm that can slide backward and forward by an actuator; 8, holder and contraption for the syringe; 9, rotatable joint connecting the handle to the arm.

Outer Catheter Guide (OCG)

The purpose of the OCG is to provide stability and passage for the IOC to maneuver while in the LV. The diameter for the inner lumen of the guide should be slightly larger than the outer diameter of the IOC (8 Fr). The OCG design and its method of introduction to the LV may consist of several modalities:

• 1.A regular, nonsteerable catheter guide is passed over a steerable or nitinol preshaped guidewire that is inserted from the femoral artery to the aortic arch.
• 2.The catheter guide consists of a nitinol-lined catheter preshaped to reach the LV with two stabilizing contact points in the aortic arch.
• 3.A steerable catheter guide, where it may be manually rotated or deflected until it reaches the LV.
• 4.A robotically controlled catheter guide where the physician may steer the guide by remote control.

In the first option, a steerable guidewire is first inserted through a sheath and guided to the descending and ascending aorta toward the LV. The guide catheter is then passed over the wire and navigated through the aortic valve opening into the LV chamber. Once the catheter is in place, the guidewire is removed. However, with the lack of steerability or shape memory, there is no stabilizing mechanism that minimizes movement of the catheter within the body. The position of the outer guide can thus be undesirably shifted by the movements of the LV and aortic valve or by the pulling and pushing forces of the IOC.

A guide with shape memory assists in positioning the distal end of the catheter to the desired intravascular location, and it also can be designed to help minimize unwanted movement during operation. Nitinol is a shape memory alloy that can return to its original form after-stress or a load is no longer applied. It is a nonferromagnetic material, making it suitable for procedures requiring magnetic navigation. By implementing a nitinol-based braid, a segment of the guide can be formed to naturally curve when it is situated in the aortic arch. The curved shape of the guide should allow two contact points along the catheter where it can reside steadily on the aortic wall. An example of a preshaped OCG would be a modified form of the Judkins left, where the primary curve of the catheter is set to reach the LV instead of the left coronary artery (Fig. 2). As the tip of the outer guide is not required to come into contact with any tissue or the endocardial wall, the catheter has a soft distal end that minimizes the risk of traumatization. However, the two curved sections of the guide should have slightly higher rigidity to limit movement while the inner catheter is operating.

Figure 2.

Example of a preshaped outer catheter guide with two built-in curves stabilizing the catheter in the aorta.

To further enhance controllability of the guide, pull wires may be added to allow deflection and steerability. The guide can incorporate one or two pull wires for unior bidirectional deflection and torque may be applied to provide distal rotation. During the procedure, the OCG must remain stable and unmoving when the inner guide is operating. Thus, a manually controlled catheter that is rotated or deflected while situated in the LV must remain that way for the duration of the procedure. As it is implausible for a physician to hold the guide in position for the entire operation, a lock mechanism should be in place at the handle, preventing the spring mechanism of standard steerable catheters from causing the distal end to deviate back to its original form. Alternatively, the steering mechanism can be robotically controlled by means of actuators at the proximal end, allowing the physician to remotely steer the catheter tip. While a robotic outer guide would be more costly and involves more consideration for its spatial arrangement at the proximal end, it also allows the physician to make any minor adjustments to the OCG from his remote workstation when necessary.

When the OCG has been properly moved into the LV and is in position to accept the inner catheter, the outer guide must be immobile. Locking the OCG in place can be realized using an extendable arm (seen in Fig. 1a), securing the handle of the outer guide. The clamp itself is stabilized by attaching to the bedrail and/or being strapped to the patient's leg.

Inner Operating Catheter (IOC)

The IOC is the main recipient of robotic and autonomous control as it contains the necessary equipment to stabilize, detect, and inject cell solution into the endocardial wall of the LV. The IOC contains four pull wires, a needle, a contact force sensor, and a recording electrode at the tip. In the robotic handle of the IOC, feedback signals are received by a microprocessor, which also controls a series of actuators that dictate the movement of the catheter and needle.

Most commercialized, steerable catheters use unidirectional or bidirectional deflection capabilities accompanied by distal torque to allow the catheter to have 360° of reach. Exceptions, such as the Artisan Extend, use four pull wires so that the torque is not required for catheter control. Adding two to three more pull wires, however, increases the radial space used in the catheter, where smaller diameter catheters are more preferable to minimize risks such as coagulation of the blood. The drawback of using rotation to manipulate the catheter tip is that the further away the tip is from the handle, the ratio of the torque of the tip to the applied torque at the handle decreases, resulting in weaker control and consistency compared to a four-pull-wire catheter. As a robotically autonomous catheter requires a high level of accuracy, a four-pull-wire system is more suitable for IOC.

Four pull wires are orthogonally attached to the inner lumen of the catheter at the same distal location. The catheter's deflection may carry a semicircular or semiellipsoid curvature with a reach of 3 to 4.5 cm. Prior to the operation, the physician chooses from catheters with different reaches based on the patient's LV size. Design considerations for the pull wires may also yield up to eight pull wires where the wires are attached at two different axial locations at the distal end, allowing the catheter to bend in an S-like shape. However, because of the desire to minimize radial size, this design focuses on a four-pull-wire system.

Omnidirectional deflection is achieved through pulling on the orthogonally placed pull wires in a certain manner. Individually pulling on one pull wire deflects the catheter tip in a direction alongside one axis. Simultaneously pulling on two adjacent pull wires deflects the tip in the quadrant in between the two relevant axes (Fig. 3a). By adjusting the distance each wire is pulled, the tip may be deflected toward any position within a semiellipsoid.

Figure 3.

(a) Deflection of pull wire based on manipulation at proximal end. The radial view displays the catheter as if looking directly at its tip. Top row: One pull wire is pulled in the x-axis direction. Bottom row: Two adjacent pull wires are pulled in between the x- and y-axes. (b) Preshaped curved needle for diagonal injection.

Furthermore, only two adjacent pull wires may be pulled simultaneously as pull wires at opposite ends may not be manipulated at the same time. Accurate control is attained by mathematically approximating the distance that the proximal actuators must move to achieve a desired position for the catheter tip. This model may be empirical or a predictive model can be attained by accounting for friction, backlash⁴⁶, and the influence of compression and tension forces on the outer guide. As the shape and size of a patient's body can vary, the parameters of the model must be calibrated at the beginning of every procedure.

The IOC contains a needle at the distal end where it may protrude from the catheter tip in order to inject cell solution into the myocardial wall. The needle is linked to a tube that runs within the catheter body until it reaches the proximal end. The cell injectate resides in a syringe stored in the catheter handle that connects to the needle tubing. As the most optimal method of injection is by inserting the cell solution diagonally while retracting²⁴, the needle can be preshaped to curve upon insertion using nitinol, as shown in Figure 3b.

Much like modeling the position of the catheter tip, the needle's movement is based on its actuator. The protrusion length of the needle compared to the distance of actuation can be premeditated using empirical or predictive models. As the minimum wall thickness of an infarcted LV can be as low as 5 mm, the protrusion length can only reach a maximum of 5 mm. Another option is to vary the needle's protrusion based on the wall thickness of the LV, allowing the needle length to exceed 5 mm when necessary. In this scenario, the operating software should have a record or feedback information of the minimum wall thickness at each LV segment.

A study in 2010 showed the effect of needle diameter and flow rate on human mesenchymal stromal cell viability⁴⁷. It was observed that there was no significant effect on cell viability in flow rates up to 500 ml h-¹, and whereas shear stress played a significant role in cell apoptosis, there was no significant relationship with the needle diameter. Although it was not statistically significant, a trend was noticed where cell apoptosis increased with a decreasing diameter. Therefore, it is suggested that the needle diameter should be larger to achieve maximal cell retention (i.e., 24-26 gauge) yet small enough to fit within the catheter lumen.

With an opening for the needle at the catheter tip, the threat of thrombosis is apparent with the possibility of blood entering the catheter lumen. An irrigation tube can resolve this problem where an anticoagulant, such as heparin, is injected from the proximal end and expelled at the distal end inside the cavity containing the needle.

A contact force sensor is necessary and advantageous to determine if the catheter tip is in contact with the LV wall. The sensor can indicate if enough contact force is generated for the electrode to obtain a suitable reading and if the needle is able to penetrate the endocardium layer. The sensor may also be used to indicate if the contact force is too high, possibly damaging the endocardium. In this manner, the ideal conditions for infarct boundary detection and cell injection fall in between an upper and lower contact force threshold.

A few methods can be used to create a contact force sensor consisting of optical fibers, impedance measurements, and magnetic sensors^48,50. Optical fibers are favorable as they have a low expense, are nonconductive and electrically passive, do not rise in temperature while operating, are immune to electromagnetic interference, and can be minimized to as low as 80 um (typically 125 um in silica glass-based fibers) for a single-mode fiber⁵¹. The force sensor is created by attaching one to three equilaterally fiber Bragg grating (FBG) optical fibers to the distal end where it runs axially inside the catheter and extends into the catheter handle at the proximal end. The optical fibers measure the compression of an elastic material (e.g., silicone rubber) or the compression of a spring mechanism caused by the contact force exerted at the IOC tip. The method of measuring this displacement may be intrinsic or extrinsic where the measured light is respectively reflected inside or outside of the fiber. Figure 4b shows an example of an extrinsic sensor. At the proximal end, an optical connector is attached to the terminal end of the fiber where the optic signals are received by a photodiode or interrogator that is connected to the microprocessor.

Figure 4.

Specification and schematic of catheter parts. (a) Pull-wire configuration. (b) Fiber optic contact force sensor with extrinsic displacement. (c) Needle and heparin channel. (d) Electrode configuration with radial views of possible Laplacian and bipolar electrode tips.

Myocardial infarcts in the endocardium are identified through areas of electrical silence, lower electrical activity, or abnormal EKGs. A recording electrode on the tip of the catheter is necessary to detect biopotential emitted from the endocardial wall of the LV. The recording mode of the electrode may be unipolar, bipolar, or Laplacian. A unipolar electrode reads a resting potential of 15 mV for a healthy endocardium, 7-15 mV for hibernating myocardium, and less than 6.9 mV for scarred areas⁶. A bipolar electrode reads a local voltage of over 1.5 mV for healthy myocardium, 0.5-1.5 mV for hibernating myocardium, and less than 0.5 mV for scarred areas⁵².

A study in 2012 describes the advantages, disadvantages, and uses of the three electrode types⁵³. Bipolar electrodes are preferential over unipolar electrodes for local measurements because of the reduction of electrical noise and far-field effects. The unipolar electrode situates the recording electrode at the distal end of the catheter where the target location resides. The recording electrode is connected to the positive input of the differential amplifier while the extracardiac reference electrode, located at a far distance from the recording site, is connected to the negative input. The unipolar electrode measures both low-frequency signals (remote wave fronts) and high-frequency signals (local wave fronts). On the other hand, the two poles in bipolar electrodes are located in close relation to each other (around 2 mm apart) at the distal end, causing any far-field effects to nullify each other, thus allowing the electrode to retain local signals while filtering remote signals. Bipolar electrodes, however, do not record a signal if the wave front is parallel to the distance between the two poles as the voltage reading for both poles would be equal. This issue can be resolved using the Laplacian recording mode. The Laplacian method uses equidistant electrodes on a 2D plane surrounding a target electrode to generate a local signal independent of the direction of the wave front. An example can be seen in Figure 4d, where the surrounding electrodes are represented by a ring or three equidistant electrodes. The calculation for the Laplacian signal is obtained by the difference between the signal received by the central electrode and the mean or weighted sum of the surrounding electrodes.

The location of the catheter may be tracked using an electromagnetic location sensor or through an image analysis algorithm. Using anatomic landmarks, this position can be related to the patient's body using a base image taken by means of an X-ray, computerized tomography (CT) scan, magnetic resonance imaging (MRI), positron emission tomography (PET) scan, or ultrasound prior to the operation. The approximate orientation and position of the catheter with respect to the inner structure of the LV can thus be realized. Disadvantages of electromagnetic tracking include its susceptibility to errors of up to a couple millimeters in accuracy. The presence of robotic devices or ferromagnetic materials can also cause magnetic field distortion, and electromagnetic tracking is not suitable for patients with pacemakers⁵⁴.

The catheter body consists of an 8-Fr tube with an outer layer of thermoplastic urethane (TPU), a braided or coiled stainless steel middle layer, and an inner layer of polytetrafluoroethylene (PTFE). Within the catheter wall, the four pull wires and lead wires for the electrodes are passed from the proximal to the distal end of the catheter (Fig. 5). Inside the catheter body lie five tubes that contain the needle, three optical fibers, and a heparin channel. This lumen configuration can be variable depending on manufacturing capabilities.

Figure 5.

Example of catheter lumen in radial view consisting of three electrode lead wires, four pull wires, three optical fibers, lumen for the needle and heparin solution, and an embedded braid or coil in the outer lumen wall.

Operational Procedure

Given the preceding design, a number of options may be explored in developing an effective method of injecting cell solution into the myocardial infarct boundaries of the LV. This section will elaborate on creating an autonomous procedure with optional design variations.

Starting Position

The operation commences with the standard protocol of accessing the LV by incision via the femoral artery. A guidewire is first inserted through the incision and passed through the femoral artery toward the descending aorta. The OCG is then passed over the guidewire and slid through the descending and ascending aorta through the aortic valve. The outer guide is then situated one fourth of the way into the LV. Once in place, the proximal end of the guide is locked in place by the stabilizing arm (Fig. 1a).

The handle of the operating catheter is attached to a mobile arm above the lying patient. The operating catheter is then passed through the sheath at the proximal end of the catheter guide until it reaches the interior of the LV. The operating catheter tip is placed upon the apex of the LV by the physician, and the mobile arm is locked in place. This arrangement, shown in Figure 6, is designated as the starting position of the automated sequence.

Figure 6.

Starting position of the system's operating sequence. The outer catheter guide is situated in the LV, just beyond the aortic valve. The inner operating catheter is situated at the apex of the LV.

Calibration

Three forms of calibration are required in this procedure. The first calibration determines the value of the electrode readings on healthy myocardium. Knowing the biopotential of viable myocardium allows the system to determine when it comes into contact with lower potential infarct boundary zones. This value may be obtained when the catheter tip comes into sufficient contact with the healthy myocardium within the LV.

The second calibration adjusts the functional parameters that control the robotic actuators to achieve accurate movement of the IOC tip. The software's existing empirical models or compensation functions may need adjusting due to discrepancies caused by friction, backlash, or tensile and compressive forces. The catheter is moved back and forth and deflected (autonomously or remotely) while the position of the tip is recorded by either an image analysis system or a location sensor. The recorded tip position is then compared to the predictive model in the software. If any discrepancies are apparent, the parameters for the existing mathematical functions are changed accordingly.

The third calibration verifies the orientation of the IOC tip in relationship to the anatomical structure of the LV. The purpose of this calibration is to allow the software algorithm to be aware of its own orientation within the LV. The system should be calibrated in both systole and diastole where the tip is deflected to come in contact with a predefined location on the myocardial wall. Once in contact, the value of each actuator's position is recorded. This process is repeated for various allocated positions and the saved coordinates are compared to a 3D surface model of the LV stored in the software. The LV model may be complex in nature, following that of a mathematical model, or it may consist of a rough estimate represented by an ellipsoid or cone. By matching the coordinates of the catheter tip to selected segments of the LV, an estimation of LV structure and its relationship to the catheter's orientation may be established.

By calibrating and associating the coordinates of the deflection of the catheter tip to specified locations of the LV, the operating software may control the catheter in a feedforward manner without feedback from an image analysis algorithm or the physician. However, occasional feedback during the procedure is preferable in order to determine if recalibration is necessary to ensure accuracy of the feedforward control.

Search and Detection

Following calibration, the physician confirms through the user interface that the autonomous procedure may begin. The catheter tip is manipulated by the actuators to follow a gridded outline of the 3D LV model. Once the catheter tip reaches a point on the grid, the tip is manipulated to stay in contact with the LV wall.

Having the catheter continuously maintain contact with a certain spot on the endocardium would be a difficult task given the constant movement of the LV and the possibility of slippage. Varying renditions of the manipulation technique described in the previous paragraph may be explored to meet the operational objective. The simplest approach would have the catheter actuated at one unchanging coordinate where the contact force is adequate in both systole and diastole. The elastic deformation of the catheter would cause the distal end to bend in harmony with the beating LV.

Another method of maintaining contact would be to manipulate the catheter to stay situated on the endocardium surface by moving and deflecting the catheter tip in accordance with the feedback received by the contact force sensor. If the force sensor reads that there is no contact established or if the contact force is too high, the catheter would self-adjust by shifting toward or away from the contact point along a certain vector. The tip may move back and forth along the vector between the systolic and diastolic coordinates in sequence with the beating of the heart (Fig. 7a). An increasing contact force would indicate that the LV is contracting, and a decreasing contact force would indicate that the LV is expanding. Thus, a real-time feedback loop involving the fiber optic contact force sensor and the actuators that maneuver the catheter is required to maintain stable contact.

Figure 7.

(a) The contact force decreases as the LV expands from systole to diastole, resulting in the catheter autonomously adjusting by moving toward the target location and vice versa. This autonomous movement is only necessary if the contact force is outside of the threshold range. (b) The catheter can approach the target site at various angles. The ideal method of approach would be where the tip is orthogonal to the endocardium wall. (c) In order to situate the catheter tip at the target site above the lower threshold force, the actuated coordinate must extend beyond the location of the desired coordinate.

The heart rate of the patient may be reduced using β-blockers or calcium channel blockers. With a steadier heartbeat, the catheter's actuators can move in a continuous, predictive loop to maintain the catheter tip on the ventricular wall instead of entirely reacting to values from the contact force sensor. In the event of unexpected irregular heartbeats, the feedback from the force sensor should cause the actuators to adjust or pull back if the upper force threshold is exceeded so as not to damage the LV. β-Blockers are not recommended for a low selection of patients with hypotension, bradycardia, severe heart failure, asthma, and cardiac rhythmic problems or are taking certain medications⁵⁵.

An additional technique to assist in this step of the procedure would be to adjust the catheter's approach toward the myocardial wall so that the angle of approach is orthogonal. As the actuated coordinate of the tip is beyond the actual myocardial wall, approaching the target site from a more acute angle may hinder the way the catheter tip settles on the endocardium as seen in Figure 7b. Furthermore, it is ideal for the full face of the tip to be in contact with the endocardium, as the electrode requires appropriate contact to achieve an accurate reading.

It should be noted that in order for the contact force to be above the lower force threshold, the actuated position of the catheter tip should be set to a coordinate beyond the desired location of the IOC tip. As a result, manipulation of the catheter yields three variables labeled as the actuated, desired, and actual coordinates of the catheter tip. The desired coordinate is the target location of the catheter. The actual coordinate is the real-time position of the catheter. Ideally, the desired and actual coordinates should be equal. The actuated coordinate is the location where the tip would reside if the LV did not obstruct its pathway. Figure 7c displays an example of this concept where the actuated coordinate is located at a point beyond the position of where the catheter tip is desired to be.

Once stable contact is maintained, the operating system receives the recorded biopotential feedback from the electrode and determines whether the location is an infarct boundary. Determination of an infarct boundary may be accomplished by two possible methods. An infarct area typically has little or no biopotential as the scarred tissue is less conductive than the surrounding myocardial muscle. For instance, scar tissue can have a bipolar reading of less than 0.5 mV, while viable myocardium would read at 1.5 mV or higher⁵². At the infarct boundary, the voltage reading would lie in between this range. Therefore, a clear reading in between these two values would likely indicate the presence of an infarct boundary. However, as the catheter is following an evenly spaced and gridded pattern, it is possible for the catheter to miss an infarct boundary where it resides between two measured locations as shown in Figure 8.

Figure 8.

Two-dimensional schematic of a search grid where each dotted intersection represents a contact point between the catheter and the ventricular wall. A wide distance between contact points gives rise to the possibility of skipping over an infarct boundary.

One solution would be to have a denser gridded pattern, where the distance between each measured site is minimized. Once an injection is made, the catheter tip would not be required to search within the vicinity of that area again. One drawback with this method is that a denser grid may yield a longer operation time. Another solution would be to readjust the catheter if an infarct boundary is missed. Any time the operating sequence detects two adjacent locations consisting of an infarct area and viable myocardium, an infarct boundary must exist in between these two points. The catheter tip accordingly relocates to a location in between these two coordinates where the infarct boundary is present.

Insertion and Detection

Upon detection of an infarct boundary, and while stable contact with the myocardial wall is maintained, an actuator controlling the needle at the proximal end of the catheter propels the needle forward. The contact force must be within the designated thresholds in order for the needle to be inserted. Once the needle is inserted, an actuator controlling the plunger of the syringe injects 0.5 ml of 10⁸ cells/ml myoblast solution into the LV wall.

Injection occurs during diastole as the interstitial space between cells become more compact during contraction, thus increasing the pressure at the needle and cell interface. Injection during expansion would coincide with the decrease in pressure, resulting in less force required to push the plunger and less resistance at the injection point. This method of injection helps limit the backflow of cell solution into the LV cavity and prevent damage to the cells at moments of increased pressure.

An optimal injection would consist of injecting the myoblast solution diagonally within the myocardial muscle. Dispersing the myoblast solution as much as possible in the infarct boundary area would lead to an ideal situation for which the myoblast cells can differentiate along the myocardial tissue and proliferate into the infarct area. Diagonal dispersion is accomplished by using the preshaped, curved needle and injecting the solution as the needle is being retracted.

Ending Sequence

The catheter continuously searches along the gridded pattern until an infarct boundary is detected where a myoblast injection occurs. After every injection, the catheter tip slightly retracts and moves away from the LV wall so the catheter may be readjusted to approach the next site orthogonally. If the algorithm detects two infarct boundaries that are too close to each other, the system only injects once. Once every designated location is inspected, the catheter returns to its starting position where the tip rests at the LV apex and the physician extracts the IOC and OCG.

The designated areas of where the system should search for infarct boundaries are predefined in the software algorithm. It is essential to avoid the chordae tendineae, its point of connection to the papillary muscles, and the aortic and mitral valve in the basal region of the LV. Avoidance of these areas can be accomplished by identifying these structures through image recognition and analysis software. The system would then avoid these areas based on those images. However, there must be a high degree of accuracy in its analysis to ensure the safety of the procedure. Another method is to designate the catheter to only search in certain areas within the LV. As the catheter is calibrated to be aware of its orientation within the LV, it can be programmed to autonomously avoid certain areas. Based on the 17-segment model recommended by the American Heart Association (AHA)⁵⁶, the catheter would consequently search in the apical region (segments 13-16; the apical anterior, apical septal, apical inferior, and apical lateral, respectively) and a portion of the midcavity region (segments 8 and 9, and partially 7 and 10; the mid-anteroseptal, mid-inferoseptal, mid-anterior, and mid-inferior, respectively).

One method of reducing the operating time of this procedure would be to have the physician enter into the user interface of the segments of the LV where infarcts are known to be located. In this manner, the catheter can directly search in the general area of the infarct instead of moving to every designated point in the LV. However, identifying these infarct areas would require diagnostic procedures preoperation.

Discussion

The proposed procedure consists of several unknowns and assumptions that will need to be tested. One assumption is in the consistency and accuracy of the coordinate control system. The constant contraction and relaxation of the heart along with varying pressure within the LV may affect the accuracy of the system. There may be errors in calibration, which would affect the entire operation and force the physician to expend more time recalibrating. Additionally, there may be a delay in the feedback from the contact force sensor, which would jeopardize the ability to ensure constant contact. The electrode also requires full contact with the endocardium wall to provide a viable reading. While these assumptions give rise to uncertainty, the designed system should be adaptable to the extent where various modules of this procedure may be explored (Table 1).

Table 1.

List of Possible Robotic Catheter Modules

Robotic Modules	Description
Completely autonomous	Calibration and procedure performed autonomously with the aid of 3D image analysis algorithm
Partially autonomous	Physician performs calibration remotely
	Locations of infarct areas are manually entered by the physician
	Physician remotely moves catheter tip to infarct area while detection, insertion, and injection are automated
	Physician confirms infarct boundary detection before insertion of needle
Complete robotic control	Physician is remotely in control of all robotic components at all times

One possible module consists of complete autonomous control where the software and robotic system take over once the catheter tip reaches the apex. Calibration, detection, and injection are all automated where calibration occurs by determining the orientation and accurate movement of the catheter tip through an image analysis algorithm or through a position sensor. Once the system is spatially “self-aware” within the LV, it may calibrate the electrode by manipulating the catheter tip to come into contact with viable myocardium.

Otherwise, an interventionist may perform the calibration process, but with the subsequent detection and injection procedure being autonomous. This module is delegated as partially automated. A partially autonomous operation consists of moments that require human interference, such as in calibration, in the event that a complete autonomy is unachievable. The partially automated sequences are essentially deviations from the ideal operation. One deviation exists where the LV is prediagnosed and the infarct areas are already known, in which case the physician may enter the target coordinates so that the catheter would not have to search every assigned area of the LV. Another example is where the physician remotely moves the catheter to desired locations in the LV instead of allowing the system to autonomously search for the infarct boundaries. The system may still autonomously detect the infarct boundary and inject the myoblast solution, but the catheter's movement is controlled remotely. In the event that detection of an infarct boundary is not foolproof, the user interface may ask the physician to confirm that an infarct boundary has been detected before injection, all the while the catheter is autonomously situated at a fixed location. These variations may be incorporated independently or in combination with each other depending on the challenges faced in the development of this catheter.

While different adaptations of this design have been mentioned, the design is flexible enough to welcome additional developments or advances that may assist in the procedure or automation of the operation. For instance, in the incident that the LV wall thickness is known at the injection point, the needle can be adjusted to protrude at various lengths. In this manner, a maximum protrusion length of 5 mm is no longer necessary. Additionally, inserting the guide and operating catheter into the LV could be autonomous if additional actuators are attached to the mobile arms holding the catheters. However, given that this portion of the operation is neither difficult to accomplish nor overly time consuming, it is an unnecessary component.

Conclusion

While this design proposal appeals to catheter cell injection in the LV, it is plausible to apply similar techniques or robotic functions to other cardiovascular catheter procedures. Components in this design, such as the stabilizing OCG, robotic IOC, or the robotic arm manipulating the IOC, can all be developed independently and used to assist with other procedures. Combining these designs to form a novel procedural system resolves several of the issues involving direct intramyocardial cell injections. As CVD is the leading ailment in the world and open heart surgery is a morbid solution for the direct injection of stem or myoblast cells, it is essential to establish an optimal noninvasive solution for this therapy.