Twenty consecutive AIS patients (all female) who underwent posterior correction and fusion surgeries with a segmental PS construct between November 2008 and October 2009 were included in this study. All the patients had a major thoracic curve (Lenke type 1: 15 patients; type 2: 5 patients). The mean age at the time of surgery was 15.9
years (range, 12–23
years). The mean preoperative Cobb angle of the main thoracic curve was 58
13° (range, 41-81°), and the mean preoperative thoracic kyphosis (T5-12) was 18.9
7.5° (range, 4.1-28.8°) on standing radiographs. The 3D computer simulation was conducted for all the patients.
The simulated corrections of scoliosis for each patient were performed on segmented 3D models of the whole scoliotic spine, created using 3D image processing software (Mimics; Materialise NV, Belgium) and based on a 1-mm-thick preoperative CT scan slice. Each segmented 3D model of the whole spine was created as follows: First, a 3D surface-reconstruction model of the whole spine was created from the CT data (Image 1). Then, the L5 vertebra was extracted from Image 1 using the range of interest (ROI) edit function of the software, to create a segmented L5 vertebra. The same procedure was repeated for each vertebra from L5 to T1 to create a segmented 3D model of the whole spine (Figure ). In the segmented 3D spine model, each vertebra could be manipulated independently. The simulated correction was then performed in two different ways: 1) complete coronal correction (C correction), and 2) complete coronal correction with complete axial derotation of vertebra (C
D correction), as described below.
Figure 1 Segmented 3D spine model.A: Frontal view. B: Lateral view. A segmented 3D model of the whole spine, in which is each vertebra could be manipulated independently, was created using 3D image-processing software (Mimics; Materialise NV, Belgium), and based (more ...)
Complete coronal correction (C correction) (Figure ): First, the mid-sagittal plane of the whole spine was defined as the plane that included three anatomical landmark points: one at the center of the posterior vertebral wall of C7, and two at the centers of the anterior and posterior margins of the upper sacral endplate. The correction was started from the L5 vertebra and performed by rotating the L5 vertebra around the axis consisting of the intersecting points between the mid-sagittal plane and the anterior and posterior margin of the upper endplate of the lower vertebra (S1), thus constraining the correction to the coronal plane (Figure A). The correction was continued until the upper endplate of the L5 vertebra became horizontal (Figure B). During the correction, the vertebrae cranial to L5 were rotated together with the L5 vertebra, maintaining their original relative position. The same correction procedures were continued to C7, to align the whole spine on the mid-sagittal plane.
Figure 2 Simulated coronal correction using the 3D segmented spine model.A: Before correction of the L5 vertebra. B: After correction of the L5 vertebra. The coronal correction was performed by rotating the vertebra around an axis consisting of intersecting points (more ...)
Complete coronal correction with complete axial derotation of the vertebrae (C
D correction) (Figure ): First, a “reference plane” defined as the plane including three points, one at the central notch of the lamina and two at the center of the upper and lower endplates on the posterior vertebral wall (Figure A), was determined for each vertebra from T1 to L5. Then, an “intervertebral axis,” defined as a line intersecting the neighboring reference planes was determined (Figure B). The simulated correction was performed by rotating each vertebra around the intervertebral axis until the neighboring vertebral reference planes became matched on the same plane (Figure C and D). The simulated correction was started at L5 and continued to T1. During the correction, the vertebrae cranial to the vertebra being corrected were rotated together with it, thus maintaining their original relative position. These corrections resulted in the complete coronal correction and axial derotation of the vertebrae (Figure D).
Figure 3 Simulated coronal correction+derotation of vertebrae using the 3D segmented spine model. A “reference plane” defined as a plane including three points: one at the central notch of the lamina and two at the upper and lower (more ...)
To evaluate the relationship between thoracic kyphosis and vertebral derotation, the thoracic kyphosis angle (T5-12), the radius of the thoracic curvature, and the vertebral rotation angle at the apex were measured before and after the two different simulated corrections for each patient. The thoracic kyphosis angle and radius of thoracic curvature were measured on the mid-sagittal plane. The radius of curvature at a given point is the radius of a circle that mathematically best fits the spinal curve at that point. The radius of curvature was measured at each adjacent segment from T1-T12, and then a value for the whole thoracic spine was determined as the mean of the values for each segment. The vertebral rotation angle at the apex was measured using Aaro’s method [29
] against a reference point set at the pelvis. The clinical relevance of these simulated corrections was evaluated by comparing the values obtained from the simulations with those measured on the postoperative CT taken for each patient.
Surgical procedures included the segmental placement of the PS, placement of the first rod on the concave side of the curve, rod rotation maneuver for the sagittal and coronal corrections, in-situ contouring for coronal correction, direct vertebral derotation for axial correction, and placement of the second rod, as described previously [18
This study was approved by the medical ethics committee of Keio University Hospital (2009-203-2).