Global motion invariant signatures for fast and accurate motion tracking in a digital image-based elasto-tomography system

Abstract

A method for converting digital images of an actuated breast into an accurate description of breast surface motion from a digital image-based elasto-tomography system comprises the steps of artificially placing a high density of fiducial markers on the breast surface, whereby the fiducial markers have different qualities and are placed in different proportions according to their quality; utilizing motion invariant properties of the fiducial markers to form a global motion invariant signature; tracking the markers on the actuated breast surface from image to image in each digital camera using the global motion invariant signature; and using the cameras calibration to measure the breast surface motion.

1 . A method for generating a high resolution feature registration and motion tracking system with minimal computation wherein said method comprising the steps of:
 a) placing a plurality of fiducial markers on a surface of a breast, each of the markers having a characteristic according to a class of characteristics, the class having a plurality of subclasses wherein the markers in each subclass have a common characteristic, the subclasses each having a unique number of markers wherein the unique number of markers in one subclass is different than the number of markers in any other subclass;   b) tracking the motion of the markers of a first subclass having the fewest number of markers on the surface from a first image to a second image;   c) partitioning the surface based upon the first subclass of markers;   d) tracking the motion of the markers of a second subclass with the next fewest number of markers within each partition; and   c) partitioning the surface based upon the second subclass of markers.   
   
   
       2 . The method of  claim 1  wherein the class of characteristics is colour. 
   
   
       3 . The method of  claim 2  wherein each subclass has fiducial markers in a unique colour selected from the group consisting essentially of red, green, blue and black. 
   
   
       4 . The method of  claim 2  wherein the fiducial markers are applied in glitter paint. 
   
   
       5 . The method of  claim 1  wherein the class of characteristics is shape and each subclass has markers with a common shape that is unique to the subclass. 
   
   
       6 . A method for generating a high resolution feature registration and motion tracking system with minimal computation in connection with a digital image-based elasto-tomography system, said method comprising the steps of:
 (a) placing a plurality of fiducial markers on a tissue surface;   (b) actuating the tissue surface;   (c) imaging the tissue surface with an array of digital cameras;   (d) choosing motion invariant properties of the fiducial markers to form a global motion invariant signature;   (e) tracking the markers on the actuated tissue surface from image to image in each digital camera using the global motion invariant signature; and   (d) using the tracked motion in each camera and the camera calibration to measure tissue surface motion.   
   
   
       7 . The method of  claim 6 , each of the fiducial markers having a characteristic according to a class of characteristics, the class having a plurality of subclasses wherein the markers in each subclass have a common characteristic. 
   
   
       8 . The method of  claim 7  wherein the motion invariant properties of each subclass of fiducial markers is the closest distances to fiducial markers of other subclasses. 
   
   
       9 . The method of  claim 7  wherein the motion invariant properties of each subclass of fiducial markers is the relative proportions of fiducial markers of other subclasses within a predefined neighbourhood. 
   
   
       10 . The method of  claim 9  further comprising the step of determining the three-dimensional coordinates of the fiducial markers according to a spatial calibration of the cameras, wherein the predefined neighbourhood is all the markers within a chosen distance of each marker in a particular class as measured using the three-dimensional coordinates. 
   
   
       11 . The method of  claim 7  wherein the motion invariant properties of each subclass of fiducial markers is the closest distances to fiducial markers of other subclasses and the proportions present of fiducial markers of other subclasses within a predefined neighbourhood. 
   
   
       12 . The method of  claim 7  wherein the subclass of fiducial markers that are in the smallest proportion are used to form the global motion invariant signature. 
   
   
       13 . A method for generating a high resolution feature registration and motion tracking system with minimal computation in connection with a digital image-based elasto-tomography system said method comprising the steps of:
 (a) placing a plurality of fiducial markers on a tissue surface;   (b) imaging the tissue surface with an array of spatially calibrated digital cameras;   (c) choosing camera invariant properties of the fiducial markers and forming camera angle invariant signatures;   (d) identifying common markers in the images of a non-actuated tissue between all cameras in the array using the camera angle invariant signatures;   (e) tracking the markers on the actuated tissue surface from image to image in each digital camera using a global motion invariant signature;   (f) using the tracked motion in each camera and the camera calibration to measure tissue surface motion.   
   
   
       14 . The method of  claim 13  wherein the camera angle invariant properties are affine invariant or projective invariant. 
   
   
       15 . The method of  claim 14  further comprising the steps of classifying the fiducial markers according to a chosen characteristic and grouping markings having common characteristics into subclasses; choosing one of the subclasses; determining the two or more closest points in non-chosen subclasses to each marker in the chosen subclass; and forming affine and projective invariant ratios of triangular areas between the markers. 
   
   
       16 . The method of  claim 15  wherein the classes of characteristics are selected from the group consisting essentially of colours and shapes. 
   
   
       17 . The method as claimed in  claim 15  wherein class of characteristics is color and the fiducial markers comprise glitter paint in colours selected from the group consisting essentially of red, blue, green and black. 
   
   
       18 . A digital image-based elasto-tomography apparatus, comprising:
 an array of spatially calibrated cameras;   a vibration unit situated proximate to the array of cameras; and   a computer system in electrical communication with the cameras and configured for computing the surface motion of an object actuated by the vibration unit and within a field of view of the cameras.   
   
   
       19 . The digital image-based elasto-tomography apparatus of  claim 18 , wherein the computer computes the surface motion of the actuated object with a global motion invariant signature. 
   
   
       20 . The digital image-based elasto-tomography apparatus of  claim 18 , further comprising a patient support proximate to the array of cameras and the vibration unit. 
   
   
       21 . The digital image-based elasto-tomography apparatus of  claim 20 , the patient support being a table. 
   
   
       22 . The digital image-based elasto-tomography apparatus of  claim 18 , the object being artificially marked with a plurality of fiducial markers for the computer system to track in computing the surface motion.