The devil is in the displacement Inducible displacement CT for implant loosening detection

Inducible displacement computed tomography (CT) is an emerging technique for diagnosing aseptic loosening of tibial components after total knee arthroplasty. The current workflow comprises three steps: (1) application of a loading device that induces valgus and varus moments to the knee, (2) acquisition of CT scans under both loading conditions, and (3) image analysis to quantify relative displacement between implant and bone. However, variability throughout this pipeline, including differences in device application, metal artefacts during CT acquisition, and elastic deformation of the tibia under load, may influence measurements.
This dissertation evaluates these sources of variation and proposes optimizations to reduce their impact on implant displacement measurements. The first part focuses on acquisition and demonstrates that, with training, the loading device can be applied reproducibly by different operators. Implant orientation within the CT scanner was shown to influence metal artefact severity and measurement accuracy. Although CT protocol optimizations improved visual image quality, their effect on quantitative outcomes was limited.
The second part addresses image analysis and evaluates current practices in inducible displacement CT. Valgus and varus loading induced elastic tibial deformation, resulting in overestimation of implant displacement. Restricting analysis to the proximal tibia reduced this effect and improved differentiation between fixed and loose implants. Additional compensation using a probabilistic model further enhanced diagnostic accuracy in a patient cohort, although refinement remains necessary. Finally, a scoping review revealed heterogeneity across workflow steps in the literature and emphasized the need for transparent reporting and validation to define its clinical value.