Closed-loop Control of Steerable Balloon Endoscopes for Robot-assisted Transcatheter Intracardiac Procedures

AI summary

Overview

• Research area: Medical robotics — robot-assisted transcatheter intracardiac procedures, specifically steerable balloon endoscopes ("cardioscopes") for direct optical visualization inside the beating heart.
• Technical level: Intermediate. The robotics and control concepts are presented clearly, but the paper assumes some familiarity with catheter-based intervention, closed-loop control, and image processing.
• Scope: The paper presents the design, fabrication, closed-loop control, and bench validation of a single-input, two-output steerable balloon cardioscope tailored to aortic leaflet laceration.

What This Paper Is About

Fluoroscopy and ultrasound, the standard imaging tools for transcatheter heart procedures, either fail to show tissue or produce noisy images, which makes precise tool navigation inside the beating heart difficult. Cardioscopy — direct optical imaging from the catheter tip — addresses this, but existing cardioscopes are either too rigid and large for vascular delivery (approximately 8–11 mm fixed-diameter devices) or are single-lumen balloons that force the camera and tools to share one channel and require continuous saline flushing. The goal of this work is a cardioscope that is collapsible for vascular delivery, contains separate imaging and working channels, and is steerable, with closed-loop control of the steering angle so the tip holds its orientation while tools are inserted and removed.

Key Contributions

1. A steerable balloon cardioscope with separate imaging and working compartments. The camera and LED sit inside the balloon, while a distinct through-lumen working channel terminates at the face of the optical window, so the working channel is visible in the image and can be centered over a target.
2. Single-input, two-output design using graded balloon wall thickness. Balloon inflation volume independently controls optical face diameter (field of view) and bending angle, with thinner wall sections expanding before thicker ones, eliminating dead zones at low pressure and over-expansion at high pressure.
3. Image-based closed-loop steering control with no additional sensor. Working-channel motion in the camera frame is converted to a pixel ratio that maps to bending angle, and a multi-threshold bang-bang law drives a syringe pump to hold or change steering angle.
4. A tunable, 3D-printed mold fabrication process. The mold-based casting workflow (five components, 4-hour room-temperature cure) can be re-parameterized for other intracardiac tasks by changing wall thickness, section length, and diameter values.

Main Findings

• Decoupling confirmed across ten trials: Varying infusion volume from 0 to 4 mL in 0.2 mL increments, the optical face expanded first over 0–0.8 mL, then bending increased with higher volumes up to approximately 100 degrees at 4 mL while face diameter leveled off. Vertical markers in the plots show D2 ≥ 8 mm before bending begins, and 8 mm ≤ D2 ≤ 11 mm, satisfying the decoupling and field-of-view requirements simultaneously.
• Design targets met: The collapsed outer diameter is 4.63 mm (requirement: ≤ 5 mm), fitting standard steerable sheaths as small as 15Fr inner diameter; maximum bending exceeds the 60-degree target; and the working channel inner diameter is 1 mm, above the 500 micrometer requirement.
• Tool insertion shifts steering angle by up to 13 degrees: With a 360 micrometer electrosurgical wire in the working channel, the same inflation volume produced up to a 13-degree difference in steering angle, which the authors cite as motivation for closed-loop control. The shift was small enough that compensating by increasing volume only slightly increases optical face diameter.
• Controller step response: Averaged over five trials, a commanded 60-degree step produced an overdamped response settling in under 6 seconds, with an average tip velocity of approximately 14 degrees per second, meeting the ≥ 10 degrees per second requirement.
• Image-based calibration: Pixel ratio as a function of bending angle, gathered in red-dyed water to mimic a blood-filled heart, was fit with a 4th order polynomial over 3 trials. The calibration curve was the same with and without a tool in the working channel, although it varied slightly between balloons.
• Automatic compensation during tool exchange: The controller maintained tool-induced steering angle error below 2 degrees, meeting the stated precision requirement, by increasing infused volume when the tool straightened the balloon and reducing volume when the tool was removed.
• Imaging performance: The cardioscope clearly resolved the six elements of group 1 on a 1951 USAF resolution test chart, with the smallest resolved features being 0.14 mm wide lines spaced 0.14 mm apart. It also enabled manual centering of the working channel over a bullseye pattern and clear visualization of calcifications on a heavily calcified human aortic leaflet submerged in porcine blood.
• Extended range noted: The authors state the fabricated balloon is already capable of bending greater than 90 degrees, beyond the 60-degree requirement, which opens other applications.

Methodology in Plain English

The team cast a balloon from a transparent silicone (Ecoflex 00-45 Near Clear) in a 3D-printed mold with five parts: two outer negatives, two end caps, and an internal pin that defines the balloon's interior shape. They tuned the wall thicknesses and section lengths in the mold so that, as saline is pumped in, the optical face inflates first and the balloon then bends toward its thicker side. A 1 mm chip camera with an integrated LED sits inside the balloon; a separate 1 mm inner diameter silicone tube serves as the working channel and terminates flush with the optical face. 3D-printed clips hold the camera aimed at the face and let the optics follow tip deflection. The proximal end attaches to a Pebax tube that can be extended and rotated through a steerable sheath, producing a trumpet-shaped workspace when combined with rotation.

For control, they built a syringe pump driven by a stepper motor (up to 42 Ncm torque, 0-450 rpm, 0-3 mL/s) with an Arduino-interfaced driver dividing the motor's 1.8-degree step into 32 subdivisions. To sense bending angle without extra hardware, the camera image is processed: a region of interest around the working channel is brightened by a factor of 3.5 (capped at 255), red blood pixels are removed in HSV space (hue ≤ 10 or ≥ 160, saturation > 15), near-black pixels are removed (grayscale < 5), pixels outside the region of interest are zeroed, and OpenCV's findContours isolates the working channel. The ratio of channel pixels to total pixels (with 160k total pixels) maps to bending angle through the polynomial fit. A multi-threshold bang-bang law (motor velocity of 100, 25, 5, or 0 rpm in sign of the pixel-ratio error, switching at error thresholds of 0.006, 0.002, and 0.001) closes the loop.

Validation was done in an acrylic tank filled with water for visual monitoring, with a manual vector-based image analysis method for ground truth, and a Savitzky-Golay filter (2nd order polynomial, window size seven) to smooth the recorded data. Image quality was tested with a laminated USAF target in a water-filled beaker, a bullseye targeting pattern, and a calcified human leaflet in porcine blood.

Why This Matters

This work points toward transcatheter procedures that no longer depend on the limited views offered by fluoroscopy and ultrasound, and toward robotic systems in which a tendon-actuated sheath handles tip positioning while a cardioscopic balloon handles tip orientation. It also provides an accessible fabrication route: the mold-based process lets the same design template be re-tuned for different intracardiac tasks by adjusting the wall thickness, length, and diameter parameters.

Real-world applications identified in the paper:

• Aortic leaflet laceration prior to transcatheter aortic valve replacement, where the steerable balloon orients an electrosurgical wire into the ventricle to avoid perforating the heart wall, and cardioscopic vision helps avoid calcifications.
• Coronary sinus access, enabled by the balloon's demonstrated ability to bend beyond 90 degrees.
• Delivery of a wide variety of instruments through the working channel, including optical fibers, stents, occlusion devices, and microcatheters.
• Broader robot-assisted transcatheter procedures, since the image-based controller is agnostic to the device being delivered and integrates with robotic catheter systems.

Industry relevance: the design reduces the need for an additional concentric steerable sheath, requires no extra steering sensor, and directly addresses catheter navigation accuracy and X-ray exposure concerns that motivate commercial robot-assisted catheter platforms. Funding from the National Institutes of Health (grants R01HL167925, R01HL124020, T32HL007572) indicates active clinical-translation interest.

Future Directions

• Adapting the mold parameters for other procedures. The authors explicitly propose re-selecting the wall thicknesses, section lengths, and diameters to produce cardioscopes for tasks beyond aortic leaflet laceration.
• Expanding tool compatibility. The paper illustrates only an electrosurgical wire; testing with optical fibers, stents, occlusion devices, and microcatheters through the 1 mm working channel remains open.
• Improving calibration consistency. The calibration curve varied slightly between balloons, and the paper does not report a method for standardizing it across manufactured units.
• In vivo validation and integration with a full robotic catheter system. All reported testing was benchtop (water tank, water-filled beaker, blood-submerged leaflet); the paper does not report animal or clinical evaluation.

Target Audience

This paper is most useful to medical robotics researchers and engineers working on catheter and endoscope design, closed-loop control of flexible instruments, and image-based sensing. It also serves interventional cardiologists and cardiac surgeons interested in direct visualization and steerable tool delivery inside the heart, as well as device developers looking for a mold-based fabrication approach that can be re-parameterized for new intracardiac procedures.

Authors’ abstract

To move away from open-heart surgery towards safer transcatheter procedures, there is a growing need for improved imaging techniques and robotic solutions to enable simple, accurate tool navigation. Common imaging modalities, such as fluoroscopy and ultrasound, have limitations that can be overcome using cardioscopy, i.e., direct optical visualization inside the beating heart. We present a cardioscope designed as a steerable balloon. As a balloon, it can be collapsed to pass through the vasculature and subsequently inflated inside the heart for visualization and tool delivery through an integrated working channel. Through careful design of balloon wall thickness, a single input, balloon inflation pressure, is used to independently control two outputs, balloon diameter (corresponding to field of view diameter) and balloon bending angle (enabling precise working channel positioning). This balloon technology can be tuned to produce cardioscopes designed for a range of intracardiac tasks. To illustrate this approach, a balloon design is presented for the specific task of aortic leaflet laceration. Image-based closed-loop control of bending angle is also demonstrated as a means of enabling stable orientation control during tool insertion and removal.