Consistent orientation of the knee in the CT scanner is key for accurate tibial component displacement detection

CT-based load-induced displacement measurements can assist in diagnosing tibial component loosening following total knee arthroplasty (TKA).This involves acquiring two CT-scans of the knee under valgus and varus loads and measuring relative implant-to-tibia displacement through segmentation and registration. However, metal artefacts from the implant vary with its orientation in the CT scanner, affecting measurement accuracy. This study investigated how implant tilt angle and direction impact displacement measurement accuracy.

A phantom containing a TKA tibial component and a PVC-reference object was scanned at different tilt angles. A neutral scan was acquired with the implant stem aligned along the gantry's central axis (z-axis). The phantom was rotated sidewards (from -40 to + 40° in 10° increments about the CT's y-axis) and forwards/backwards (from -30° to 30° in 10° increments about the CT's x-axis). Image analysis included segmentation and registration between CT scans with different tilts. Three metrics were used: 1) intensity similarity around the implant based on PVC-reference registration (unaffected by metal artefacts), 2) intensity similarity based on implant registration (affected by artefacts), and 3) relative displacement differences between registered implant and PVC-reference, quantified as mean target registration error (mTRE).

Intensity similarity decreased with increasing implant tilt, especially when tilt angles crossed the central gantry axis. Implant registration increased intensity similarity but caused positioning errors over 0.5 mm (mTRE) for tilt differences larger than 20°.

Both implant tilt angle and direction influence metal artefact severity, but maintaining consistent alignment helps preserve segmentation quality and registration accuracy, and reliable displacement measurements.