Cancellous Bone

All patients in our hospital underwent three-dimensional reconstructive cervical CT (PHILIPS, Brilliance, slice thickness 1.5 mm, distance 1.5 mm, tube voltage 120 kV) within 1 week before surgery. We use the picture archiving and communication system (PACS) measurement of the C2-C7 HU value. HU values were measured using CT scans according to a previously described method [16 (link)]. The average HU values of each vertebral body were based on the axial plane inferior only to the superior endplate, middle of the vertebral body and axial plane superior only to the inferior endplate. The HU value was measured by placing the largest elliptical region of interest (ROI) at the mid-vertebral body, and the ROI was chosen to include as much trabecular bone as possible and to avoid cortical bone and heterogeneous areas, such as cortical bone margins, osteophytes and osteosclerosis. The average of HU values measured from the three ROIs was regarded as the HU for the individual vertebral (Fig. 1).Fig. 1

Midsagittal (A) and axial CT images demonstrating the measurement of vertebral HU value on the axial plane inferior only to the superior endplate (B), middle of the vertebral (C) and axial plane superior only to the inferior endplate (D) (the first letter C stands for cervical vertebra, which consists of seven segments from top to bottom, denoted by C1-C7, the same as T1)

As determined by the individual situation, an autogenous tricortical bone graft of appropriate size was harvested from the ipsilateral iliac crest. Cancellous bone was harvested with the smallest osteotome possible. A longitudinal dorsal incision was made lateral to the extensor hallucis longus tendon with an interface between the extensor hallucis longus tendon and the dorsalis pedis artery, both of which were retracted correspondingly. The soft tissue was distracted by a lamina spreader to expose the talonavicular and navicular-cuneiform joints. The talonavicular and navicular-cuneiform joints were distracted using a Hintermann distractor over separate K-wires. The articular surfaces were debrided in situ with cartilage shovels to the subchondral bone. A K-wire was used to drill the subchondral sclerotic bone plate into a favaginous condition for fusion. Then bite off the excess osteophyte from the lateral 4-corners. The plantar ligament and plantar soft tissue of the navicular are loosened with a sharp knife, leaving only the insertion point of the posterior tibial tendon. The whole debridement process created a relative space around the navicular bone. Subsequently, a periosteal detacher was pressed against the lateral bony protrusion of the navicular bone to rotate the bone outwards to the original top location. Once the reduction was deemed satisfactory by direct visualization, two to three crossing K-wires were used for temporary fixation. After the demonstration of the corrected medial longitudinal arch on the C-arm, the lateral half of the navicular bone (including the talonavicular and navicular-cuneiform joints involved in the necrotic lesion) was excised using an osteotome to form a broad dorsal trapezoid laterally and a rectangular longitudinal bone bed. And the modified tricortical iliac bone block was inserted into the space between the talus and the cuneiforms. Finally, two hollow lag screws and a dorsal LCP were used to arthrodese the talonavicular-cuneiform joints. A transverse Herbert screw was used (where needed) to fix the bone block and medial navicular bone. The wound was closed after packing the previously acquired cancellous bone to smooth the defect gaps.
Postoperatively, a protective non-weight bearing short-leg plaster cast was applied for 6 weeks, after which weight-bearing was gradually allowed as tolerated.

The mechanobiological model requires hexahedral elements aligned within the growth plate stacked in several layers to define transition zones and enable progressive growth simulations for each layer of the growth plate. We developed the GP-Tool to automatically create hexahedral meshes based on the subject-specific femoral geometry. The STL-files obtained from the segmented femurs were used as input for the GP-Tool. A visual overview of the steps to create a mesh with the GP-Tool is shown in Figure 1.
Using MATLAB’s (MathWorks, Natick, MA, United States) principal component analysis (“pca”-function) the main orientation of the growth plate was determined. All parts of the femur were rotated so that growth plates’ main orientation was parallel to the XY plane. The growth plate itself was removed and the part above the growth plate of the proximal trabecular bone was positioned on the part which is below the growth plate. Smoothing of the intersection region was performed with MeshLab (Cignoni et al., 2008 (link)). These steps were necessary to create a continuous mesh to add perfectly aligned hexahedral elements in the growth plate later. Sculpt tool of Coreform Cubit (Coreform, Utah, United States) was used to create a hexahedral mesh with an element size of approximately 1.5 mm. A mesh convergence study was conducted based on three femurs to ensure that the results are not influenced by the number of elements in the mesh (see Supplementary Material). The part above the growth plate was moved to its original position and ten layers with equal height were added and presented the growth plate. Finally, the mesh was optimized to no longer include elements with negative Jacobians. This procedure was performed for each femur of all participants (N = 50) and resulted in meshes of approximately 150.000 nodes and 140.000 elements varying due to different femur sizes.