Promising results of an non-invasive measurement of knee implant loosening using a loading device, CT-scans and 3D image analysis

After total knee arthroplasty up to 13% requires revision surgery to address loosening. No current diagnostic modalities have a sensitivity or specificity higher than 70–80% to detect loosening, leading to 20–30% of patients undergoing unnecessary, risky and expensive revision surgery. A reliable imaging modality is required to diagnose loosening. This study presents a new and non-invasive method and evaluates its reproducibility and reliability in a cadaveric study. Ten cadaveric specimens were implanted with a loosely fitted tibial components and CT scanned under load towards valgus and varus using a loading device. Advanced three-dimensional imaging software was used to quantify displacement. Subsequently, the implants were fixed to the bone and scanned to determine the differences between the fixed and the loose state. Reproducibility errors were quantified using a frozen specimen in which displacement was absent. Reproducibility errors, expressed as mean target registration error, screw-axis rotation and maximum total point motion were 0.073 mm (SD 0.033), 0.129 degrees (SD 0.039) and 0.116 mm (SD 0.031), respectively. In the loose condition, all displacements and rotation changes were larger than the reported reproducibility errors. Comparing the mean target registration error, screw axis rotation and maximum total point motion in the loose condition to the fixed condition resulted in mean differences of 0.463 mm (SD 0.279; p = 0.001), 1.769 degrees (SD 0.868; p