Purpose of the study: The issue of patellar kinematics remains a difficult problem for patellar resurfacing during conventional or computer-assisted knee surgery, yet adequate knowledge is required for appropriate orientation of the patellar cut and insert positioning. The purpose of this study was to develop a non-invasive tool for in vivo kinematic analysis of the patellar tract and to compare results with the gold-standard invasive method.

Material and methods: A special experimental set-up designed for this study enabled experimental simulation of load-bearing flexion-extension cycles of the knee joint. Range of motion from 0 to 102° was imposed with a computer-controlled motor. The analysis was conduced on 14 complete lower limb cadaver specimens. Patellar kinematics was analyzed for each knee simultaneously with two systems: a non-invasive method using a low-dose stereoradiographic scan linked to a 3D reconstruction software; and the reference system using tripodes implanted on the patella and radio-opaque spherical markers. Six degrees of freedom were considered: three translations and three rotations. Sequential kinematic recordings were made by calculating the position of a patellar landmark in relation to a femoral landmark.

Results: The mean difference between the results obtained with the two systems was less than 1 mm for anteroposterior and vertical translations, greater for mediolateral translations. It was less than 2° for patellar flexion-extension, to the order of the motion itself for abduction-adduction, and to the order of 5° for horizontal tilt.

Discussion and conclusion: The non-invasive technique proposed here appears to be reliable for patellar translations and flexion, but need further improvement for tilt and adduction-abduction. This is particularly true for the 45° to 90° range of motion because of the difficult problem of determining the contours of the patella. Further developments for this tool are under way.