Objectives: There were five objectives: (1) To evaluate the accuracy of the Femoral Head Detection Technique for the TKR CT-less application. (2) To assess improvements if a pelvic reference is used in addition to a femoral reference. A pelvic reference takes into account potential pelvic displacements during head center acquisition. (3) To compare two types of algorithms ("Pivot" and "Conical"). (4) To define the optimal number of static positions (positions that the user will have to manipulate the shank to) necessary for satisfactory Femoral Head Detection. (5) To determine an appropriate threshold level that the algorithm uses to evaluate the validity of the computation and beyond which the computation will be rejected as a safety measure. Design/Methods: This study was undertaken using a prototype of the TKR CT-less SNM module including a modified tool calibration algorithm for femoral head detection. It was modified to take only conical (instead of conical and axial) values. Each position of the femoral and pelvic reference cases was recorded in order to study the difference between the different head detection procedures. The recorded data was later used to compare the different algorithms and their accuracy when using a pelvic reference, and to find the optimal number of static positions. Tests were performed by 4 different users on 7 legs obtained from 4 Ml cadavers. Results were compared to a measurement of the actual location of the femoral head ("nominal center"). The following methodology was applied: (1) Fix the active reference to the pelvis; (2) Put the femoral reference on the distal part of the anterior cortex; (3) Perform a kinematic procedure with 14 different static position while recording tracking data; (4) Rename the output file; (5) Repeat steps 1 to 4 another four times; (6) After 20 simulations (five times for four persons) the femoral head was dislocated, and 20 points were digitized uniformly on the spherical surface of the femoral head with a pointer to measure the head center as a basis to compare with the results of calibration algorithm; 7) Using the recorded data* calculations corresponding to all other scenarios were performed off-line to evaluate the effects of altering various parameters. A procedure was rejected when maximum values for a parameter were exceeded. The following analyses were performed: (a) Verify the accuracy of the nominal center measurements by verifying how well the 20 points fit a sphere as assumed for the hip center detection algorithm; (b) The difference in accuracy between the pivot algorithm and the conical algorithm; (c) The difference in accuracy between 7 and 14 static position on the accuracy of the procedure; (d) The different in accuracy with and without pelvic reference (fixed or mobile); (e) The difference in accuracy between two threshold parameter settings.
展开▼
