Mylène Jansen

380 Chapter 18 imaging protocol were included (10 from each trial/original indication TKA or HTO, respectively). KJD treatment was performed using an external fixation frame, fixed to the joint laterally and medially using 4 pairs of bone pins. During surgery the joint was distracted to a distance of 2 mm, which was gradually extended by 1 mm per day over the next 3 days, reaching 5 mm of total distraction. This was confirmed radiographically, after which patients were discharged. Full weight-bearing on the treated knee was allowed and encouraged, using crutches if necessary. After 6 weeks, patients returned to the hospital, where the frame and pins were removed, without further imposed rehabilitation protocol. The original RCTs and the extended imaging protocol were granted ethical approval by the medical ethical review committee of the University Medical Center Utrecht (protocol numbers 10/359/E, 11/072 and 11/482/E). All patients gave written informed consent. CT analyses Patients underwent CT scanning with a reconstructed slice thickness of 0.45–0.5 mm, at baseline (pre-treatment) and 1 and 2 years after treatment. All CT scans were made at the UMC Utrecht using the same CT scanner and settings. All scans were performed with 120 kVp and exposure 87–232 mA. The field of view was 512x512 pixels and pixel spacing varied between 0.27x0.27 mm and 0.98x0.98 mm. The CT dose index (CTDI vol ) was 3.9–10 mGy and dose length product 174–495 mGy*cm. Stradview v6.0 (University of Cambridge Department of Engineering, Cambridge, UK, in- house developed software freely available at https://mi.eng.cam.ac.uk/Main/StradView) was used for semi-automatic segmentation of the tibia and femur. Cortical bone thickness (mm, referring to the subchondral bone plate as well as cortical bone in non-articular regions) was determined using an automated optimized Gaussian model fit algorithm able to measure bone thickness in the sub-millimeter range, unconstrained by the point spread function limit of the CT imaging system. 15 Trabecular bone density (Hounsfield units, HU) was also measured as part of this optimized solution, from the inner cortical bone edge inwards to 12 mm beneath the mesh surface (outer bone surface). This is not the same as bone mineral density, as no dedicated phantom was scanned for calibration, but gives a reasonable approximation. A 3D isosurface was generated for the 2 bones separately through semi-automatic segmentation. This software and technique have been explained in detail previously. 16,17 Osteophytes were excluded from the segmentation (see example of segmentation in Figure 1).