We present an analysis of manual and computer-assisted preoperative pedicle screw placement planning. Preoperative planning of 256 pedicle screws was performed manually twice by two experienced spine surgeons (M1 and M2) and automatically once by a computer-assisted method (C) on three-dimensional computed tomography images of 17 patients with thoracic spinal deformities. Statistical analysis was performed to obtain the intraobserver and interobserver variability for the pedicle screw size (i.e. diameter and length) and insertion trajectory (i.e. pedicle crossing point, sagittal and axial inclination, and normalized screw fastening strength). In our previous study, we showed that the differences among both manual plannings (M1 and M2) and computer-assisted planning (C) are comparable to the differences between manual plannings, except for the pedicle screw inclination in the sagittal plane. In this study, however, we obtained also the intraobserver variability for both manual plannings (M1 and M2), which revealed that larger differences occurred again for the sagittal screw inclination, especially in the case of manual planning M2 with average differences of up to 18.3°. On the other hand, the interobserver variability analysis revealed that the intraobserver variability for each pedicle screw parameter was, in terms of magnitude, comparable to the interobserver variability among both manual and computer-assisted plannings. The results indicate that computer-assisted pedicle screw placement planning is not only more reproducible and faster than, but also as reliable as manual planning.

Summary

Optimum position of pedicle screws can be determined preoperatively by CT based planning. We conducted a comparative study in order to analyse manually determined pedicle screw plans and those that were obtained automatically by a computer software and found an agreement in plans between both methods, yet an increase in fastening strengths was observed for automatically obtained plans.

Pelvic incidence is as a key factor for sagittal balance regulation that describes the anatomical configuration of the pelvis. The sagittal alignment of the pelvis is usually evaluated in two-dimensional (2D) sagittal radiographs in standing position by pelvic parameters of sacral slope, pelvic tilt and pelvic incidence (PI). However, the angle of PI remains constant for an arbitrary subject position and orientation, and can be therefore compared among subjects in standing, sitting or supine position. Such properties also enable the measurement of PI in three-dimensional (3D) images, commonly acquired in supine position. The purpose of this study is to analyse the sagittal alignment of the pelvis in terms of PI in 3D computed tomography (CT) images.

A computerised method based on image processing techniques was developed to determine the anatomical references, required to measure PI, i.e. the centre of the left femoral head, the centre of the right femoral, the centre of the sacral endplate, and the inclination of the sacral endplate. First, three initialisation points were manually selected in 3D at the approximate location of the left femoral head, right femoral head and L5 vertebral body. The computerised method then determined the exact centres of the femoral heads in 3D from the spheres that best fit to the 3D edges of the femoral heads. The exact centre of the sacral endplate in 3D was determined by locating the sacral endplate below the L5 vertebral body and finding the midpoint of the lines between the anterior and posterior edge, and between the left and right edge of the endplate. The exact inclination of the sacral endplate in 3D was determined from the plane that best fit to the endplate. Multiplanar 3D image reformation was applied to obtain the superposition of the femoral heads in the sagittal view, so that the hip axis was observed as a straight not inclined line and all anatomical structures were completely in line with the hip axis. Finally, PI was automatically measured as the angle between the line orthogonal to the inclination of the sacral endplate and the line connecting the centre of the sacral endplate with the hip axis.

The method was applied to axially reconstructed CT scans of 426 subjects (age 0–89 years, pixel size 0.4–1.0 mm, slice thickness 3.0–4.0 mm). Thirteen subjects were excluded due to lumbar spine trauma and presence of the sixth lumbar segment. For the remaining subjects, the computerised measurements were visually assessed for errors, which occurred due to low CT image quality, low image intensity of bone structures, or other factors affecting the determination of the anatomical references. The erroneous or ambiguous results were detected for 43 subjects, which were excluded from further analysis. For the final cohort of 370 subjects, statistical analysis was performed for the obtained PI. The resulting mean PI ± standard deviation was equal to 46.6 ± 9.2 degrees for males (N = 189, age 39.7 ± 20.3 years), 47.6 ± 10.7 degrees for females (N = 181, age 43.4 ± 19.9 years), and 47.1 ± 10.0 degrees for both genders (N = 370, age 41.5 ± 20.1 years). Correlation analysis yielded relatively low but statistically significant correlation between PI and age, with the correlation coefficient r = 0.20 (p < 0.005) for males, r = 0.32 (p < 0.0001) for females, and r = 0.27 (p < 0.0001) for both genders. No statistically significant differences (p = 0.357) were found between PI for male and female subjects.

This is the first study that evaluates the sagittal alignment of the pelvis in terms of PI completely in 3D. Studies that measured PI manually from 2D sagittal radiographs reported normative PI in adult population of 52 ± 10 degrees, 53 ± 8 degrees and 51 ± 9 degrees for 25 normal subjects aged 21–40, 41–60, and over 60 years, respectively [3], and 52 ± 5 degrees for a cohort of 160 normal subjects [4]. The PI of 47 ± 10 degrees obtained in our study is lower than the reported normative values, which indicates that radiographic measurements may overestimate the actual PI. Radiographic measurements are biased by the projective nature of X-ray image acquisition, as it is usually impossible to obtain the superposition of the two femoral heads. The midpoint of the line connecting the centres of femoral heads in 2D is therefore considered to be the reference point on the hip axis, moreover, the inclination of the sacral endplate in the sagittal plane is biased by its architecture and inclination in the coronal plane. On the other hand, the measurements in the present study were obtained by applying a computerized method to CT images that determined the exact anatomical references in 3D. Perfect sagittal views were generated by multiplanar reformation, which aligned the centres of the femoral heads in 3D. The measurement of PI was therefore not biased by acquisition projection or structure orientation, as all anatomical structures were completely in line with the hip axis. Moreover, the range of the PI obtained in every study (standard deviation of around 10 degrees) indicates that the span of PI is relatively large. It can be therefore concluded that an increased or decreased PI may not necessary relate to a spino-pelvic pathology.