IMAGIKA SYSTEM FOR MEASURING TOTAL KNEE ARTHROPLASTY WEAR

Purpose of the study: The accuracy, reproducibility and concordance of wear measurements made with the Imagika system were tested on knee prostheses. Material and methods: Anteroposterior radiographic images of implants with tibial inserts measuring 9, 7, and 11 mm thick were obtained, the tibial base had a 28 mm bead for calibration. The ap images were digitalized and variations in incidence were controlled with a plumb line. Combining the tilt positions from −10° to +10° and rotation from −5° to +5°, a total of 132 images were obtained. Four groups were defined according to tilt and rotation (±5°, ±3°). The images were read by to observers. Reproducibility and agreement were assessed for the overall series and for each of the four groups. Two images were read 40 times by the same observer using variable digital quality (100–300dpi) to determine measurement accuracy and error. Results were compared with the manufacturer’s data sheet. Results: Measurement agreement was poor in the four groups. Reproducibility was excellent at the 1% threshold for the overall series and for groups 3 and 4 for both observers. The accuracy improved from 0.6 to 0.5 mm by improving digitalization from 100 to 300 dpi with an error to the order of 0.05. Guiding the points used by the software for measurements, the accuracy could be improved to 0.25mm. The difference between the lateral plateau and the medial plateau appeared for the lesser tilt and for rotation greater than 2°. The thickness displayed was close only for the groups with little tilt. Discussion: The reproducibility and accuracy of the Imagika system can be brought below the 0.5 mm threshold. For views without tilt, the difference compared with the displayed values varied from 0.1 to 0.3mm. The agreement remains poor if the position of the measurement points is not predefined by the system. Conclusion: Offering an excellent reproducibility and good accuracy, the Imagika system enables monitoring TKA wear if the incidence of the radiographic images is perfectly controlled. For a more quantitative approach, the calibration method would have to be improved.