1996 SPIE Medical Imaging Symposium, Newport Beach, California, 10-15 February 1996.
A. Purpose: Rapid and complex motions in human joints are believed to contribute to the development of osteoarthrosis and bone mineral erosion. Traditional motion analysis techniques are incapable of accurately measuring joint kinematics. Movement of muscle and skin obscures internal joint motion when surface mounted markers are observed. To study these motions we are developing a biplane digital radiography system to record internal joint movements during running impact or rapid limb articulation.
B. Methods: A specialized gantry has been developed to permit two radiographic views, 45 to 90 degrees apart, in the horizontal plane. Presently, images are recorded from conventional image intensifiers (30 cm, three field intensifiers with high current power supplies) using synchronized square wave pulses of radiation (150 kW, 120 kVp high frequency generator) lasting about .3 seconds and exposing the object to about 75 mR of radiation. During this pulse, digital images have been recorded using a CCD device operating at 250 frames per second (fps) and acquiring digital images with 500 x 240 pixels (interline transfer CCD and 40 MHz frame grabber). Correction of geometric distortion is done using a calibration phantom consisting of 567 markers arranged in a square array. Internal markers in phantoms or subjects are located and tracked using a pattern recognition algorithm. Marker position is identified to subpixel accuracy using a centroid calculation. Marker coordinates from each camera are then transferred to a motion analysis system (EVa, Motion Analysis Corp., Santa Rosa, CA) which tracks and aligns the coordinates to achieve a 3D estimate of kinetic motion. A graphic display of motion observed from any angle may then be replayed at slow speed.
C. Results: Accuracy of the method was evaluated with a 5 cm acrylic cube containing three 1.6 mm tantalum markers in a right isosceles triangle with 3 cm legs. Elastic suspension produced rapid, complex motion. Relative position of markers was determined to an accuracy of .02 mm with a precision of .1 mm in dynamic tests. The angle between marker pairs was determined to an accuracy of .11 degrees with a precision of .22 degrees in dynamic tests. Functionality of the system has been demonstrated using normal and ACL deficient canines running on a treadmill. Differences in pre-post ACL transection are clearly demonstrated.
D. New Work: The combination of resolution and frame rate in one plane of this system exceeds the performance achieved in other laboratories doing kinetic x-ray imaging. The 3D registration achieved with biplane imaging further distinguishes the capabilities of this system. Experimental work in our laboratory has demonstrated the ability to record digital images at 1000 fps with a .1 msec gate and measure velocities of 27.6 m/s with a 1% precision. We are now developing digital CCD recording systems which would acquire 512 x 512 frames at 1000 fps and investigating methods to produce pulsed x-ray emission at 1000 pulses per second with a .1 msec pulse width.
E. Conclusions: We believe that high speed biplane digital radiography provides an effective method to dynamically measure a wide range of in-vivo joint motions. New algorithms to track natural radiologic patterns in the joint image will eliminate the present need to implant small metal markers. Initial studies of normal and abnormal human subjects are scheduled for this fall.