Method for surgical navigation utilizing scale-invariant registration between a navigation system and a localization system

[0001] This invention relates to a system and methods for interventional medicine, and more generally to navigation and localization systems for directing and controlling movement of medical devices through the body.

[0002] Interventional medicine is the collection of medical procedures in which access to the site of treatment is made through one of the subject's blood vessels, body cavities or lumens. For example, angioplasty of a coronary artery is most often performed using a catheter which enters the subject's arterial system through a puncture of the femoral artery in the groin area. The procedure is referred to as PTCA, or Percutaneous (through the skin), Transluminal (through the blood vessel), Coronary (in the vessel of the heart), Angioplasty. Other interventional medical procedures include assessment and treatment of tissues on the inner surfaces of the heart (endocardial surfaces) accessed via peripheral veins or arteries, treatment of vascular defects such as cerebral aneurysms, removal of embolic clots and debris from vessels, treatment of tumors via vascular access, endoscopy of the intestinal tract, etc.

[0003] Interventional medicine technologies have been applied to manipulation of instruments which contact tissues during surgical procedures, making these procedures more precise, repeatable and less dependent of the device manipulation skills of the physician. Some presently available interventional medical systems for directing and manipulating the distal tip of the medical device by remote actuation use computer assisted navigation and an imaging system for providing imaging of the catheter and blood vessels and tissues. The system may also be configured to include a localization system for determining the position of the catheter or medical device in the localization system's own frame of reference translatable to the displayed image of the imaging system. In the absence of an explicit link between the localization system and the imaging and navigation systems, the localized device data is not readily available to control navigation of the device by computer-controlled means. The present invention provides a method for obtaining a registration between the display of a localization system and a navigation system, and a means for using the display directly to control device navigation.

[0004] According to the principles of the present invention, a system and method are provided for direct control, from the display of a localization system of a navigation system for deploying a catheter or medical device within the body, such that a physician can input a displacement or a target location that may be converted into a global specifications for a navigation system to enable navigation directly from the display of a localization system. The present system and method provide a means for determining and establishing a scale invariant transformation or “registration” of the coordinate systems of a localization system and a navigation system. One embodiment of this method comprises the identification of at least a minimum number of points or anatomical landmarks projected in the localization system display, of which the three dimensional coordinates relative to the navigation system are known and can be used to estimate a transformation between the navigation system coordinates and the coordinates as determined from the localization system display. The system provides for marking or identifying the various points in the localization projection display, and determining the coordinates of these points relative to the display for use in determining distances between the known points. The method of transforming the coordinates of the localization display projection comprises using ratios of the distances (preferably vectors but possibly also scalar magnitudes) in a cost function to determine a rotation matrix for transforming the points in the display of the localization system to the reference frame of the navigation system. The cost function provides a method for evaluating rotation matrices, to determine an optimum rotation matrix that will provide the most accurate transformation between the local coordinate frame and the global coordinate reference frame of the navigation system. The localization system also has the capability of receiving a user input of a displacement vector or a target location through a joystick or touch screen apparatus, and transforming the displacement or location data to enable navigation control through the localization system. This provides the capability for more precise navigation and operation from a remote location, removing the physician from the procedure site and thereby reducing exposure to X-ray radiation. Remote operation also allows a particularly skilled physicians to operate over a broader geographical area.

[0010] A system for navigating a medical device through the lumens and cavities in an operating region in a subject typically utilizes an elongate medical device, having a proximal end and a distal end adapted to be introduced into an operating region in a subject. The system also connects to an imaging system from which images can be transferred to the navigation system. The navigation system includes a display 20 as shown in FIG. 1 for displaying an image of the operating region, and may include a representation of the distal end of the medical device in the operating region. The imaging system may be an x-ray, MRI, ultrasound, or other imaging system. Most interventional procedures currently use fluoroscopy for imaging of the blood vessels, tissue surfaces, and catheter devices. Dye or contrast agents which absorb X-rays can be injected through the catheter system to obtain a “roadmap” of the vessels or outline of the endocardial surfaces. The medical devices can contain radiopaque materials such as platinum marker bands or polymers loaded with heavy metals to render them visible in the fluoroscopic image. The physician typically views the tissues and the medical device in a single imaging plane, which does not uniquely specify the orientation of the catheter or the direction of tissue targets.

[0011] The navigation system's display can be used to assist in navigating the medical device, and may provide two projections which can be used to identify three dimensional coordinates of points on the display. The navigation system display 20 may display anatomical landmarks or various known points within the operating region of the subject, such as points 22, 24, 26 and 28. The imaging planes may be rotated to two or more angles to generate perspective views. Many imaging systems contain two planes of imaging, often configured for orthogonal views. Some modern fluoroscopic systems have the capability to capture images while rotating around the subject. Three dimensional images are then quickly reconstructed by the imaging system. While these images are not quite “real time” because of the time required for rotation and computing, the three dimensional images are most useful for device navigation.

[0012] The navigation system can be used with a localization system for providing real time coordinate information of the position of the medical device on a constantly updated localization system display 30, which shows the location and movements of the device. The display 30 of the localization system can be used to locate a plurality of various points or anatomical landmarks that have known three dimensional coordinates as obtained from the navigation system by fluoro-localizing locations using two X-ray images displayed therein. The coordinates of these known points may be read relative to the plane of the localization display. Specifically, an apparatus such as a transparent sheet having a fine grid may be placed over the localization system display, or a commercially available touch screen apparatus configured to mount over the display may be used to read the coordinates of the various points upon suitable identification by a user. The coordinate system of the grid or touch screen apparatus are used to determine coordinates of these various points projected from the display plane. While the scale of the localization system display 30 is generally unknown, a local frame of reference may established to read the coordinates of the points as displayed and determine distances between the points.

[0013] A transformation of these points must be made from the local frame of the localization display to the global frame or coordinate system of the navigation system. A vector in one frame may be related to a vector in the other frame by a rotation matrix, which is used in transforming vectors in the global frame to vectors in the local frame. A depiction of a vector 44 in the global coordinate system 40 of the navigation system being projected onto the localization reference frame 42 as vector 46 is shown in FIG. 3. In the system and method of the present invention, a minimum of four known points are required to establish four distance vectors that can be used to determine the three degrees of freedom of a rotation matrix and to exclude dependence on scale. The distance vectors and the rotation matrix can be defined as:
{right arrow over (X)}₁=({right arrow over (X)}_B−{right arrow over (X)}_A); {right arrow over (X)}′₁({right arrow over (X)}′_B−{right arrow over (X)}′_A)
{right arrow over (X)}₂=({right arrow over (X)}_C−{right arrow over (X)}_B); {right arrow over (X)}′₂=({right arrow over (X)}′_C−{right arrow over (X)}′_B)
{right arrow over (X)}₃=({right arrow over (X)}_D−{right arrow over (X)}_C); {right arrow over (X)}′₃=({right arrow over (X)}′_D−{right arrow over (X)}′_C)
{right arrow over (X)}₄=({right arrow over (X)}_A−{right arrow over (X)}_D); {right arrow over (X)}₄′=({right arrow over (X)}′_A−{right arrow over (X)}′_D)  (1)
{right arrow over (u)}′=PR{right arrow over (u)}  (2)
where {right arrow over (u)}′ is a projection in the local frame of vector {right arrow over (u)} in the global frame. Thus, {right arrow over (X)}′₁ (“primed” quantities are defined in the localization display, while “unprimed” quantities are the corresponding quantities defined in the navigation system) would be the projection of the rotated vector R{right arrow over (x)}. This projected vector generally has local frame components (v, w, 0), and can be written as:

[0014] $\begin{array}{cc}{\overrightarrow{X}}^{\prime }=R\overrightarrow{x}-\overrightarrow{n}\left(\overrightarrow{n}·R\overrightarrow{x}\right)=\left(I-{\mathrm{nn}}^t\right)R\overrightarrow{x},\mathrm{or}{\overrightarrow{X}}^{\prime }=\mathrm{PR}\overrightarrow{x}\mathrm{where}P\equiv \left(I-{\mathrm{nn}}^t\right)& \left(3\right)\end{array}$
From the above, three independent ratios of the lengths constructed from the vectors defined in equation (1) can be used to determine the three degrees of freedom associated with the rotation matrix R, and can be defined as:
l₁={right arrow over (X)}′₁^T{right arrow over (X)}′₁={right arrow over (X)}₁^T R^T P R {right arrow over (X)}₁
l₂={right arrow over (X)}′₂^T{right arrow over (X)}′₂={right arrow over (X)}₂^T R^T P R {right arrow over (X)}₂
l₃={right arrow over (X)}′₃^T{right arrow over (X′)}₃={right arrow over (X)}₃^T R^T P R {right arrow over (X)}₃
l₄={right arrow over (X)}′₄^T{right arrow over (X)}′₄={right arrow over (X)}₄^T R^T P R {right arrow over (X)}₄  (4)
From the above, three independent ratios of the lengths constructed from the vectors defined in equation (4) can be used to determine the three degrees of freedom associated with the rotation matrix R. From these ratios of distances, a cost function can be constructed for defining an error associated with a given rotation matrix, as defined below:

[0015] $\begin{array}{cc}C={\left(\frac{l_2}{l_2}-\frac{{\overrightarrow{x}}_1^{T}Q{\overrightarrow{x}}_1}{{\overrightarrow{x}}_2^{T}Q{\overrightarrow{x}}_2}\right)}^2+{\left(\frac{l_2}{l_3}-\frac{{\overrightarrow{x}}_2^{T}Q{\overrightarrow{x}}_2}{{\overrightarrow{x}}_3^{T}Q{\overrightarrow{x}}_3}\right)}^2+{\left(\frac{l_3}{l_4}-\frac{{\overrightarrow{x}}_3^{T}Q{\overrightarrow{x}}_3}{{\overrightarrow{x}}_4^{T}Q{\overrightarrow{x}}_4}\right)}^2& \left(5\right)\end{array}$

[0016] where Q=R^T P R

With the quantities l₁, l₂, l₃ and l₄ determined from the localization system display, a rotation matrix R can be estimated by minimizing the cost function C in equation (5).

[0017] The cost function provides a method for evaluating rotation matrices that will project or transform a vector between the local frame and the global navigation frame, to determine a rotation matrix that will provide the most accurate projection or transformation of a vector between the local coordinate frame and the global coordinate reference frame of the navigation system. Specifically, the cost function C evaluates each rotation matrix R to find the best R which minimizes C using an algorithm. Since R may be written as a rotation by an angle about an axis, R is completely determined by the angle parameters ({acute over (ø)}, θ, β). The algorithm performs a search in ({acute over (ø)}, θ, β) space to find the best choice of a rotation matrix point R that minimizes the cost function. The algorithm may optimally perform the search by defining a coarse polar grid that divides {acute over (ø)}, θ and β into a predetermined number of intervals or radial slices. In the preferred embodiment, the interval for angles {acute over (ø)} and β is preferably a size of 2π/20 radians, and the interval for θ is preferably a size of π/10 radians. This establishes a semi-spherical region divided into a coarse polar grid, in which C at each point is evaluated to find the point at which C is least. A second polar grid may also be established that comprises interval sizes less that that of the first polar grid, for further evaluation of points to again find the point at which C is least. The point of which the value of C is least is selected and identified as Γ. About this point, a 4×4×4 fine grid may be established as shown in FIG. 4A with the center of the grid 32 encompassing the point Γ. The resolution of this fine grid in the preferred embodiment is preferably a interval of π/80, and is illustrated in FIG. 4A. It should be noted that the fine grid recited above is only an example, and may alternatively comprise any number of grid intervals or various resolution sizes. The algorithm evaluates each point to again find the point at which the cost function is least, and may subsequently establish a second fine grid and third fine grid as shown in FIG. 4B about the point 34 at which C is least to iteratively determine a rotation matrix that yields a desired resolution or minimum C value. At the n-th step, the resolution is reached in a geometrical progression to yield an accurate resolution in as few steps as possible. In the preferred embodiment, this practice achieves the desired resolution in about three or four steps. From the cost function equation, the algorithm can determine a point at which the rotation matrix minimizes the cost function, and this rotation matrix can be used for transformation between the localization display and the global frame of the navigation system. A corresponding vector in the global or fluoroscopy image system or navigation system reference frame can then be determined from a vector in the localization system. This method provides a means for determining and establishing a scale invariant transformation or “registration” of the coordinate systems of the localization system and the navigation system. Thus, an input means such as a joystick can be used to input a displacement vector based on or mapped to the localization system display, which can then be transformed to navigation system coordinates and provide for control of the navigation system to deploy the medical device to a desired location within the body.

[0018] The imaging system, localization system, and navigation system of the system are defined relative to one or more spatial coordinate systems, which must be “registered” to one another within the computer. Fluoroscopic or MRI images form the imaging system are viewed relative to a “tissue” or body coordinate system. When the device is localized relative to the coordinate system of a navigation system, then this coordinate system must be registered to the image coordinate system to properly transform directions defined on the localization display to image coordinates. In this manner, the localization system can be “virtually” registered with the navigation system to enable direct navigation control via the localization system display, while also enabling images of the catheter to be viewed in both the localization and imaging system displays.

[0019] Registration as described here can be done in part as a calibration process between known points or anatomical features, of which are non-planar and have known three dimensional coordinates relative to the navigation system reference frame as can be determined for instance by fluoro-localization. Anatomical landmarks or markers fixed to the subject can be used to register a local reference frame, and to serve as references for localization. In some situations it may be useful for the physician to facilitate registration, for example by accessing anatomical landmarks within the subject with a localized medical device that is seen both on a fluoroscopic imaging system and in the localization system display.

[0020] The physician interface preferably includes an input device, that allows the physician to input at least a desired destination or a direction. This input means preferably allows the physician to input the desired destination or direction using the display 32 of the imaging system 30. The physician must control the interventional robotic system using a safe and efficient user interface. In the “telemetric” mode, the physician uses hand controls such as a joy stick to move the catheter while observing a real time image of the catheter and tissue. In the “automatic” mode, the physician specifies an end-point or process, and the computer automatically causes the medical device to move. For example, the physician may point and click on an anatomical point on a three dimensional in the tissue image, and the computer would then advance and steer the catheter to the specified point. Alternatively the physician could define a path by a touching and dragging a line or curve on a display screen of a localization system, or touch an anatomical location on the localization display as a target destination for the medical device. In every case, the physician interface allows the doctor to interact with the tissue image to specify device movements, and to observe the subsequent movements of the medical device.

[0021] The navigation system provides for manipulating the distal end of the medical device through the body. In one preferred embodiment the navigation system is a magnetic navigation system. The navigation system could alternatively be a mechanically driven system utilizing gears, motors, or cables and wires to shape the distal portion of the medical device, and electrostrictively actuated system, a hydraulic actuation system, or others known to those skilled in the art of actuation. In the case of a magnetic navigation system, the system orients the distal end of the medical device in a selected direction in the operating region through the interaction of externally applied magnetic fields with magnetic materials associated with the medical device inside the operating region, using at least one external source magnet outside the subject's body. The various alternative actuation schemes described here are provided for the purposes of non-limiting illustrative examples only, and a person skilled in the art can choose to construct a system based on the teachings detailed herein and any of a variety of actuation methods as may be convenient for a particular application.

[0022] The navigation system could also include an advancer acting on the medical device adjacent the proximal end of the medical device for selectively advancing and retracting the medical device.

[0023] The navigation system may comprise an input device such as a computer mouse, a joystick, a hand-held localized stylus, a screen pen or keyboard, or other user input means known to those skilled in the art for receiving at least a destination for the distal end of the device input by the user using the localization system, or a spatial direction input specified by the user, or other device positioning information such as a device deflection direction. A controller, responsive to the destination or direction input provided by the user, operates the navigation system to orient the distal end of the medical device in the proper orientation to reach the input destination or in the appropriate direction, and as the distal end of the medical device moves to its proper orientation, an advancer can be operated either by a user or automatically by the navigation system to advance the distal end to the destination input or direction specified by the user.

[0024] It is important to note that cost functions other than the one described here can be used to implement the determination of an optimum best-fit coordinate transformation. Likewise, the successive grid subdivision method described here is for non-limiting illustrative purposes only, and other optimization methods such as gradient-based methods, simplex methods, methods based on neural networks or genetic algorithms, or any of a variety of methods known to those proficient in such mathematical methods can be used to implement the determination of a suitable transformation.

[0025] The system includes a computer to manipulate the digital images; add device localization information to the images; interpret physician commands and translate these to suitable device deflection commands; translate information between all relevant coordinate reference frames; could coordinate auxiliary data such as ECG signals used to gate the images; controls the advancer consistent with physician commands and coordinated with steering commands; supply data to the user interface monitors, and receive commands from the user interface. In the “automatic” mode, the computer can drive the catheter to a physician specified end point.

[0026] The interventional robotics system of this invention allows the physician to automatically direct the tip of a medical device to points, or along a path or direction, within body lumens or cavities of a subject. The physician interacts with a user interface that sends physician commands to a control computer, and presents the physician with images of tissue in the operating region, including an image of the device. The control computer integrates and registers real time images, pre-operative images, and local images, and commands and coordinates the actions of a device advancer and device tip steering mechanism. The physician can operate remotely from the subject to reduce his exposure to radiation. Exquisite device manipulation skills are not required, and the physician can concentrate his attention on navigation commands and the delivery of therapy.

[0027] Once a localization system display is available to view the reat-time location of a device and a navigation system with input from an imaginf system is available, the control algorithm proceeds to: (i) receive inputs of coordinates of a minimum number of points having known coordinates relative to the navigation system from the localization system display, (ii) determine a rotation matrix with the lowest cost function, for transformation of a vector in the localization display reference frame to that of the navigation reference frame, and (iii) perform a mapping of user inputs to control the navigation system in guiding the medical device within the body. This allows navigation to be performed to a significant extent from the localization system display alone, which reduces the frequent x-ray irradiation that occurs during the use of fluoroscopy imaging for navigation purposes.