Method of real-time tracking of moving/flexible surfaces

1. A vision system for real-time tracking of moving and flexible surfaces, comprising:

a) a stereo endoscope for acquiring images of a region of a moving flexible surface, and a light source for illuminating the region of moving flexible surface;

b) a vision processor connected to said stereo endoscope and said light source, said visional processor programmed with instructions for analyzing images from said stereo endoscope and computing, in real time, a current location of the region of moving flexible surface based on a current image of the region of moving flexible surface and a previous location of the region of moving flexible surface;

c) a computer controller connected to said vision processor, said computer controller including a visual display for displaying images of a region of the moving flexible surface;

d) said computer controller being programmed to pass one or more designated positions on said moving flexible surface to said vision processor, and wherein the vision processor is programmed to continuously track the one or more designated surgical positions on the moving flexible surface to provide updated positions of said one or more designated positions.

2. The system according to claim 1 wherein said vision processor is programmed for computing rigid and stretching transformations by the steps of

a) synchronously acquiring left and right image frames from left and right camera members of a stereo camera, of the region of moving tissue at a time K;

b) computing, from said left and right image frames 3D point cloud data in a coordinate frame of said stereo camera at said time K;

c) fitting a parametric surface to said 3D point cloud at said time K;

d) extracting a location of at least three features of interest in said left image frame at said time K in 2D pixel coordinates, extracting a location of each at least three feature of interest in said right image frame at said time K in 2D pixel coordinates, storing said 2D pixel coordinates of said locations of said features of interest in said left and right image frames at said time K in a vision processor memory storage;

e) computing, using said stored 2D pixel coordinates of said locations of said features of interest in said left and right image frames at said time K, a 3D location of said features of interest at said time K in a coordinate frame of said camera and storing said 3D locations in said vision processor memory storage;

f) mapping said 3D location of said features of interest onto said parametric surface at time K and storing the parametric surface with mapped 3D locations in the vision processor memory storage;

g) retrieving from the vision processor memory storage 3D locations of said features of interest from previous left and right image frames synchronously taken at a time K−1;

h) computing said rigid transform using said retrieved 3D locations of the features of interest at time K−1 and the 3D locations of the features at time K;

i) retrieving from the vision processor memory storage a parametric surface with mapped 3D locations at time K−1, applying said rigid transformation to said retrieved parametric surface with mapped 3D locations at time K−1 to obtain an intermediate parametric surface with mapped 3D locations;

j) matching the 3D locations of features of interest on the intermediate parametric surface at time K−1 to the same corresponding 3D locations on the stored parametric surface at time K and from said correspondences computing said stretching transformation; and

k) using said rigid and stretching transformations to calculate at time K a location of the region of moving tissue based on a location of the same region of moving tissue on said moving tissue at time K−1 stored in said vision processor memory storage.

3. An image-guided surgical system for performing procedures on deformable and/or moving sections of tissue, comprising:

a) a stereo endoscope for acquiring images of a region of moving tissue and a surgical tool tip being used in said region of tissue, and a light source for illuminating the region of moving tissue and said surgical tool;

b) a vision processor connected to said stereo endoscope and said light source, said visional processor programmed with instructions for analyzing images from said stereo endoscope and computing, in real time, a current location of the region of moving tissue based on a current image of the region of moving tissue and a previous location of the region of moving tissue;

c) a computer controller connected to said vision processor, said computer controller including a visual display for displaying images of a region of tissue being operated on in which the region of moving tissue is located;

d) a robot controller connected to said computer controller;

e) at least one surgical robot connected to said robot controller;

f) a surgical tool releasibly grippable by said robot, said surgical tool including said surgical tool tip;

g) said computer controller being programmed to pass one or more designated surgical positions being acted on by the surgical tool tip on the region of moving tissue to said vision processor, and wherein an operator initiates a surgical procedure by instructing the computer controller to command the robot to activate the surgical tool tip to proceed to a first of said one or more designated surgical positions, and wherein the vision processor is programmed to continuously track the one or more designated surgical positions on the moving tissue to provide updated positions of said one or more designated positions, and communicate these updated positions to said robot controller for tracking and to said computer controller for display to the operator, and wherein said computer controller is programmed to command said surgical robot to proceed to each designated position and perform a surgical procedure in a preselected order from the first designated position point to a final designated position.

4. The system according to claim 3 wherein said vision processor is programmed for computing rigid and stretching transformations to calculate at a current time K a location of any region on said moving tissue based on a location of the same region on said moving surface at previous time K−1.

5. The system according to claim 4 wherein said vision processor is programmed for computing rigid and stretching transformations by the steps of

a) synchronously acquiring left and right image frames from left and right camera members of a stereo camera, of the region of moving tissue at a time K;

b) computing, from said left and right image frames 3D point cloud data in a coordinate frame of said stereo camera at said time K;

c) fitting a parametric surface to said 3D point cloud at said time K;

d) extracting a location of at least three features of interest in said left image frame at said time K in 2D pixel coordinates, extracting a location of each at least three feature of interest in said right image frame at said time K in 2D pixel coordinates, storing said 2D pixel coordinates of said locations of said features of interest in said left and right image frames at said time K in a vision processor memory storage;

e) computing, using said stored 2D pixel coordinates of said locations of said features of interest in said left and right image frames at said time K, a 3D location of said features of interest at said time K in a coordinate frame of said camera and storing said 3D locations in said vision processor memory storage;

f) mapping said 3D location of said features of interest onto said parametric surface at time K and storing the parametric surface with mapped 3D locations in the vision processor memory storage;

g) retrieving from the vision processor memory storage 3D locations of said features of interest from previous left and right image frames synchronously taken at a time K−1;

h) computing said rigid transform using said retrieved 3D locations of the features of interest at time K−1 and the 3D locations of the features at time K;

i) retrieving from the vision processor memory storage a parametric surface with mapped 3D locations at time K−1, applying said rigid transformation to said retrieved parametric surface with mapped 3D locations at time K−1 to obtain an intermediate parametric surface with mapped 3D locations;

j) matching the 3D locations of features of interest on the intermediate parametric surface at time K−1 to the same corresponding 3D locations on the stored parametric surface at time K and from said correspondences computing said stretching transformation; and

k) using said rigid and stretching transformations to calculate at time K a location of the region of moving tissue based on a location of the same region of moving tissue on said moving tissue at time K−1 stored in said vision processor memory storage.

6. The system according to claim 3 wherein said one or more designated surgical positions are suture points.

7. The system according to claim 6 wherein said suture points are selected by an operator designating selected suture points on said left or right images.

8. The system according to claim 3 wherein said tissue is a deformable lumen.

9. The system according to claim 3 including a user operated input specification device for the operator to highlight said or more designated surgical positions being acted on by the surgical tool tip in the region of tissue being operated on.

10. The system according to claim 9 wherein said user operated input specification device is a software based cursor designation system configured such that an operator designates said one or more designated surgical positions on one or more of the displayed images of the region of tissue being operated on.

11. The system according to claim 9 wherein said user operated input specification device is a physical designation system configured such that an operator manually designates said one or more designated surgical positions directly on the region of tissue being operated on, and wherein said computer controller transmits coordinates of said one or more designated surgical positions to said surgical robot based on locations of said one or more designated surgical positions in a field of view of said stereo endoscope.