Four-Dimensional CT Analysis Using Sequential 3D-3D Registration

A method to reconstruct the motion of a moving bone with respect to the fixed bone from 4D CT data, facilitating the description of detailed joint kinematics. Once the soft of the whole bone has been created, and your reference points have been selected, most of the reconstruction processes will be completed automatically. With the target joint of the subject inside the CT gantry of the 4D CT image, ask the subject to move the joint during the 10-second scan time, while obtaining a series of volume data.

For surface semi-automatic segmentation of the 3D CT data, load the CT DICOM data into the appropriate software, and click Edit New Label Field to open the label field. Identify which threshold CT attenuation value is appropriate to extract cortical bone from the source bone, and select materials with CT attenuation values above the threshold. Check the label for bone cortex selection and use an appropriate editing tool to manually modify the demarkation.

To generate the surface data from the labeled bone cortex position data, click Generate Surface and use the slider to smooth the extent. Then click Apply. To save the surface data in STL format, click File and Export Data As, and select STL binary Little Endian.

To perform an automatic segmentation of the 4D CT volume data, use the DICOM reading module in the programming software to extract the geometric data showing the CT attenuation values above the threshold from all 51 frames of the 4D CT data. Then use a batch processing script to reconstruct all of the surface data of the point cloud with higher CT attenuation values than the threshold for all of the 4D CT frames in the image processing software. To perform surface registration from static 3D CT to the first frame of the 4D CT, use the selecting phase function to trim the bones in a static 3D CT into partial segment data that are included in all of the frames of the 4D CT movie data.

The 4D CT surface data are only partial segments that will be included in each volume image, because the surface registration requires that one surface data point is included in another surface. Use the Pick Points function to select three landmarks in the fixed and moving bones that can be easily identified from the trimmed 3D CT surface and the surface data of the first frame of 4D CT in the 3D mesh editing software. Then match the partial fixed and moving bones roughly on the first frame of the 4D CT surface data according to the picked landmarks, and use the iterative closest-point algorithm to perform surface registration in the open-source software.

To perform a sequential surface registration, use the iterative closest-point module in the open-source software to first match the partial surfaces of the fixed bone in the first 4D CT frame onto the surface data of the second frame. Then, reconstruct moving bone motion with respect to the fixed bone. When the rotation parameters have been measured, define the coordinate systems of the fixed and moving bones.

The example data of knee kinematics shows that the Varus angle of the tibia gradually decreases as the tibia is extended. The tibial external rotation increases at the end of the extension, corresponding with the home movement of the knee in previous reports. Graphs of the error for translation and rotation show that the error is tolerable for femur lengths longer than 9%of the whole length, and tibia lengths longer than 7%of the whole length, as the CT slice thickness is 0.5 milimeters, and exceeds the error size.

In addition, the calculation of patellar kinematics demonstrates that the lateral tilt of the patella corresponds to the knee flexion angle. The surface reconstruction of the bone cortex should be visually checked and may require manual segmentation. Recently, many studies of automatic segmentations have used Using these techniques, reconstruction can performed automatically.