WO2019037071A1

WO2019037071A1 - Device and method for feedback and control using optical fibers in catheters

Info

Publication number: WO2019037071A1
Application number: PCT/CN2017/099004
Authority: WO
Inventors: Weyland CHENG
Original assignee: Cheng Weyland
Priority date: 2017-08-25
Filing date: 2017-08-25
Publication date: 2019-02-28

Abstract

A catheter (1) using one or more optical fibers (2) to produce multiple feedback signals located along the length of the catheter (1). The feedback signals may include strain, pressure, temperature or other variables obtainable by optical fibers (2). Sequentially, the feedback signals are used in an analytical model to further derive new data information, such as the catheter's tip position, geometrical body shape, tip orientation, bodily vibration, and so on. The collected data from the fiber optic catheter (1) may also be used to optimize the control of the catheter (1) or to achieve automated movement.

Description

DEVICE AND METHOD FOR FEEDBACK AND CONTROL USING OPTICAL FIBERS IN CATHETERS

Field of the Invention

The present invention relates to generating new feedback data in a fiber optic catheter by modeling multiple feedback signals along the length of one or more optical fibers.

Background of the Invention

Catheters are used for a variety of diagnostic and therapeutic procedures throughout the body allowing for minimally invasive operations. It is optimal to have finer control of the catheter to minimize the procedural duration, surgical mistakes, and the skill and training requirements of the catheter operator. Remote catheters using magnetic resonance or robotic control have been introduced to allow for faster operation times. Remote systems also allow the operator to work from a separate workstation rather than by the patient’s bedside where x-ray fluoroscopy is often used to provide visual feedback images. At a distant location, operators are no longer required to wear protective heavy lead suits that often lead to chronic back injuries.

Robotic catheter procedures rely on internal and external feedback to provide optimal control and to reveal a clearer picture of the operating environment to the physician. However, quality feedback is still difficult to obtain due to the size and safety requirements of intrinsic sensors and the expense, speed and efficiency of external sensors such as image analysis systems.

Following the remote control of catheters is the refinement and improvement the system’s performance. Due to the variety of shapes, lengths, material properties and forces that come into play when manipulating a catheter, robotically maneuvering a catheter can still be highly non-intuitive where the distal end does not move in accordance with the operator’s directions. [J. Jung, Penning, R., N.J. Ferrier, M.R. Zinn, “A Modeling Approach for Continuum Robotic Manipulators: Effects of Nonlinear Internal Device Friction, ” IEEE/RSJ International Conference on Intelligent Robots and Systems, San Francisco, CA, September 25-30, 201. ]

Attempts to automate or model the movements of the catheter have resulted in limited success in terms of accuracy, 3-dimensionality, speed and reaction time, and dynamical adjustments to the moving environment. One issue in remote control is the lack of immediate feedback. Image analysis or magnetic resonance systems that can determine the catheter’s position have a time-delay when providing feedback to the control system. [Cheng W, Law PK. Conceptual Design and Procedure for an Autonomous Intramyocardial Injection Catheter. Cell Transplantation. 2017； 26 (5) : 735-751. ]

Optical fibers have been implemented in catheters, relaying in-vivo feedback signals of from the proximal end of the catheter, such as the contact tip force, oxygen saturation of the blood, fluid concentration, temperature, pressure, etc. Optical fibers are optimal in that they require no electrical activity within the catheter, produce fast response signals, and can be relatively small in size. Additional existing technologies in optical fibers includes the ability to receive multiple feedback signals from a single-mode fiber. For instance, using wavelength division multiplexing (WDM) , different wavelengths of light are multiplexed into a single optical fiber. Fiber Bragg grating (FBG) sensors and out-coupling taps are written within the core of the fiber to generate temperature, strain, pressure, chemical or interferometric feedback from the various wavelengths. [Morey WW, Dunphy JR, Meltz G. Multiplexing fiber bragg grating sensors. 1991； 10 (4) : 351-360. ]

Using such technology, these optical fibers can be placed into the length of the catheter wall. Using the multiple feedback signals from the fibers, a mathematical model can be implemented to the feedback data to indirectly derive new and different data points that regular sensors cannot or have difficulty obtaining in catheter procedures.

Summary of the Invention

The invention uses one or more optical fibers within a catheter. Preferably three to four optical fibers. The optical fibers are preferably symmetrically embedded within the catheter wall. Each fiber consists of multiple sensors such as stress, strain, displacement, , contact force, pressure, temperature, vibration, chemical, etc. Each optical fiber may consist of only one type of sensor or a combination of these sensors. These sensors may be intrinsic or extrinsic or a combination of both. The optical fiber may be a single-mode fiber or multi-mode fiber. More preferably a single-mode fiber. Multiple sensors within one optical fiber is achieved through any form of optical splitting and combining techniques such as wavelength division multiplexing, time division multiplexing or frequency division multiplexing. More preferably wavelength division multiplexing.

An interrogation unit is located at the proximal end of the catheter where it emits and receives the wavelength signals. The signals from the optical fibers are then relayed to an operating system or microprocessor where new data is derived by implementing mathematical or statistical models such as artificial neural networks, a machine learning algorithm. The newly derived data may consist of the catheter’s relative tip coordinates, the coordinate position of the entire catheter body, the tip angle or orientation, the orientation of the catheter body, the vibration of the catheter body, the momentum, speed or acceleration of the catheter’s movement, and so on.

The optical fiber sensor system may also be implemented in a control system to robotically control or automate the catheter. The control system may use the newly derived data or it may directly use the multiple feedback signals from the fibers. Common control methods may be used, such as proportional-integral-derivative (PID) control or state space models. Less common methods may also be used, such as in machine learning techniques like deep learning. In a machine learning model, a mass amount of data is collected, depicting the catheter in thousands of scenarios and shapes. The collected data of fiber optic feedback signals and other features are processed through the machine learning algorithm to calculate a target output, which could be the distance, speed or acceleration that the robotic actuators need to actuate.

Brief Description of the Drawings

FIG. 1 is a depiction of the optical fibers and its sensors within the catheter；

FIG. 2 shows a radial view of how three optical fibers may be installed within the catheter；

FIG. 3 is a depiction of how the catheter would be shaped in various configurations to amass data for the machine learning algorithm；

FIG. 4 is a depiction of how the sensors can placed in different patterns within the optical fibe；

FIG. 5 is overall schematic of how the optical fiber sensor system can be used；

FIG. 6 is a schematic of an example experimental set-up with actuators in order to collect data for a machine learning algorithm；

Detailed Description of Exemplary Embodiments of the Invention

In the following description of the embodiments, references to the accompanying drawings are by way of illustration of an example by which the invention may be practiced. It will be understood that other embodiments may be made without departing from the scope of the invention disclosed.

One embodiment uses machine learning algorithms to derive the positional catheter tip coordinates from multiple strain sensor signals spread evenly across three symmetrically placed single-moded optical fibers 2 within the catheter 1 as in FIG. 1. One of the three optical fibers also includes a temperature sensor in the event that the values of the strain sensors are temperature dependent. The optical fibers run from the proximal end of the catheter to the distal end and are positioned near the outer surface of the catheter or embedded within the catheter wall, radially forming triangular points and allowing a three-dimensional platform as in FIG. 2. Multiple strain sensors 3 are implemented evenly along the length of the fiber from the proximal end to the distal end. Fiber Bragg gratings 3 are used to form the strain sensors. The optic signals are processed through a multiplexer and demultiplexer 5 at the proximal end 4 of the device using wavelength division multiplexing to achieve multiple signals within a single-mode fiber. The feedback data is sent to a processor or operating system where it is fed to an established algorithm or mathematical model that translates the strain data to the relative coordinate position of the entire catheter body.

The positional coordinates of the catheter body can then be translated to a graphical display screen to give visual feedback for the physician or to a control system to robotically or remotely control the catheter through proximal actuators. Alternatively, instead of being processed into new data, the multiple strain values can be directly sent to the control system where it uses these values in its control models. An overall schematic of the system is displayed in FIG. 5.

To achieve remote or robotic control, the possible catheter control mechanisms include, but are not limited to, pull wires, smart material-actuated catheters, hydraulically driven catheters, ionic polymer-metal composites, and magnetic resonance control.

Other embodiments process the multiple feedback signals using models such as the lumped parameter model, finite element analysis, statistical models or other machine learning algorithms. The control system may also use different control models such as PID control, PID control with inverse kinematics, state space, fuzzy logic, deep learning or neural networks, etc.

Another embodiment incorporates contact force sensors at the catheter tip to account for obstructions. The new shape or position of the catheter is still derived using the same data modeling methods when the tip experiences contact.

Another embodiment measures the vibration of the catheter caused by robotic actuation or pressure from the dynamic environment.

Yet another embodiment has the sensors located in different patterns throughout the optical fibers or in a specified pattern as in FIG. 4. The sensors in FIG. 4 are arranged closer together near the distal end where more curves or deflection of the catheter may occur.

Practical implementation of a machine learning algorithm to convert the feedback strain values to the catheter body’s position can be accomplished with the following embodiment. A prototype catheter as seen in FIG. 6 is automated to change shapes using external 6 and proximal 7 actuators to randomly move its body into various configurations. Examples of these configurations are seen in FIG. 3. The tip 9 of the catheter may be deflected in any random direction using actuators at the proximal end 4 to pull on four pull wires 8 within the catheter. The four pull wires allow for omnidirectional deflection.

Thousands of random shapes and configurations of the catheter are automated and two orthogonal stereo cameras 10 capture the shape of the catheter body. Using image analysis tools that detect colored tags on the catheter, multiple positional coordinates of each catheter shape is recorded along with the feedback strain values from the optical fibers. The amassed data is then processed in a artificial neural network where the input features consistent of the strain values and the target outputs are the Cartesian coordinates of the segmented points along the catheter body.

Although the present invention has been described hereinabove by way of specific embodiments thereof, it can be modified, without departing from the spirit and nature of the subject invention as defined herein.

Claims

A catheter for diagnostic and therapeutic procedures, comprising:

at least one optical fiber installed within the catheter body and runs axially along the catheter length, wherein the at least one optical fiber comprises at least one type of sensor；

a multiplexer and a demuliplexer or interrogator arranged at the proximal end of the catheter, configured to process the feedback signals received from the at least one type of sensor to obtain multiple feedback signals；

wherein, the obtained multiple feedback signals are relayed to an operating system or microprocessor to derive new data information through algorithms or models.
The catheters of claim 1, wherein the sensor comprised in the optical fibers detects the signals comprising at least one of the following types of signals: stress, strain, displacement, contact force, pressure, temperature, vibration, and chemical.
The catheters of claim 1, wherein the derived new data information includes at least one of the following: catheter’s relative tip coordinates, the coordinate position of the entire catheter body, the tip angle or orientation, bodily orientation, the physical vibration of the catheter body, and the momentum, speed or acceleration of the catheter’s movement.
The catheters of claim 1, further comprising one or more external and proximal actuators configured to accurately control, automate or autonomize the catheter based on the obtained multiple feedback signals.
The catheters of claim 4, wherein the multiple feedback signals along the catheter length or derived new data information is used in a control model or algorithm to control the catheter.
The catheters of claim 5, wherein the control model or algorithm comprises at least one of the following: PID control, PID control with inverse kinematics, state space, fuzzy logic, a machine learning algorithm .
The catheters of claim 4, wherein at least one of the following control mechanisms is used to control the catheter: pull wires, smart material-actuated catheters, hydraulically driven catheters, ionic polymer-metal composites, and magnetic resonance control.