Mimics research 21

We selected a 31-year-old healthy male volunteer with a height of 175 cm and a weight of 70 kg. The right knee joint was scanned by a 1.5T MRI machine (ESSENZA, Siemens) from sagittal, coronal and transverse directions, and a T2 proton-weighted image was obtained with a layer thickness of 1.5 mm, spacing of 0 mm, a matrix of 192 × 320, a field of view size180mm. The MRI image data were saved as DICOM files. All DICOM files were imported into Mimics Research 21.0 software (Materialise, Belgium) for the initial 3D reconstruction of each structure of the knee joint. Nineteen solid structures including bone, cartilage, ligament, and meniscus were created and saved as STL format files.

The Digital Imaging and Communications in Medicine (DICOM) data of the affected hip 3D-T1WI were imported into the Mimics Research 21.0 software (Materialise, Belgium). The iliac crest, pubic bone, ischium, femoral head, and superior femoral bone, as well as the per femoral head and acetabular cartilage, were modeled by using magnetic lasso, regional growth, mask editing, and 3D model calculation. Import the 3D model into 3-material research 13.0 software (Belgium materialise) for mesh optimization to generate surface network. The optimized surface mesh is used to generate tetrahedral mesh, which is imported into FEA module, and the three-dimensional finite element model is output to ANSYS software in CDB format. In ANSYS 19.0 software, and the proximal femur was subjected to grid densification. A total of 186,351 units were generated, including 116,849 units of cortical bone and 69,502 units of cancellous bone (Fig. 1). Referring to the previous literature, the grid is assigned with different material properties to the imported model.Fig. 1

Three-dimensional model of the hip

Following preparation, all specimens underwent radiological assessment for bone mineral density (BMD) and morphology. The Horizon DXA system (Hologic, Inc., Marlborough, MA, USA) was used to assess the bone mineral density (g/cm²) of the proximal femur. QCT image sequences were acquired with the SOMATOM Force 128-slice dual-source CT scanner (Siemens Healthcare GmbH, Erlangen, Germany) with the following parameters for tube A and tube B: voltage: 120 and 150 kV, intensity: 270 and 540 mAs, respectively. QCT image sequences were reconstructed at a slice thickness of 0.6 mm and position increment of 0.4 mm using the Qr69 kernel and the ‘Bone’ window.
Images were segmented and postprocessed for preoperative planning and creation of drill guides (see Experimental Setup) using Mimics Research 21.0 and 3-Matic Research 13.0 (both Materialise NV, Leuven, Belgium). Cortical width (mm) was measured at 12 sites (6 on medial and 6 on lateral side) along the axis of the implant in the coronal plane.

Following the reconstruction, a phase‐contrast MRI angiogram (PC‐MRA) was created by voxel‐wise multiplication of the phase‐contrast magnitude images with the absolute velocity, and subsequently averaged over all cardiac frames.21 PC‐MRA data were used for segmentation of the aorta anatomy in Mimics Research 21.0 (Materialise, Leuven, Belgium).22 Segmentations were conducted for all 4D flow MRI sequences separately.
For 2D analysis, 4D flow MRI scans were resliced to the slice position of the 2D flow MRI scan in GTFlow (Gyrotools). ROIs of the AAo and DAo were drawn in the 2D flow MRI scan as well as in the resliced 4D flow MRI images. All ROIs were drawn in the images at the time of peak flow. To match the highest temporal resolution of 40 cardiac frames, all velocity and flow measurements of lower temporal resolution were interpolated by a shape‐preserving piecewise cubic interpolation.
For 3D analysis, WSS at the peak velocity time frame was calculated in MATLAB, as previously described.23 For voxel‐wise comparisons of velocity and WSS, datasets were registered by a rigid transformation followed by nearest neighbor interpolation. Additionally, directional differences in velocity and WSS vectors were assessed by comparing angular difference distributions.22, 24

The extracted Dicom file was imported into Mimics Research 21.0 (Materialise Inc., Belgium), and the bone boundary was extracted through threshold segmentation. Each segment of the vertebra was acquired with filling and splitting and saved as a different Mask. The reconstructed vertebras were saved in STL format. The model surface was smoothed, meshed, and converted to a STEP solid model with Geomagic Studio (3D Systems Corporation, Rock Hill, South Carolina, United States). The C1-S1 segment model in STEP format was imported into Solidworks 2019 (Dassault Systèmes SolidWorks Corporation, Vélizy-Villacoublay, France) to simulate surgical correction according to the patient’s postoperative radiographs. Subsequently, cortical and cancellous bone were divided, and the intervertebral disc and facet joint capsule were established, where the intervertebral disc contained the annulus fibrosus and nucleus pulposus. The proportion of nucleus pulposus was between 30% and 50%. The thickness of cortical bone was 1 mm.