Table 1.
Inclusion and exclusion criteria, demographic and anthropometric data for all three studies.
Fig 1.
Data collection procedure for all 3 studies (left).
EOS imaging (middle) and optical 3D surface scanning (right) was always performed; markers were only placed in Studies 2 and 3, and EOS posture measurement and replication was only performed for study 3.
Fig 2.
Inter- and intrarater reliability (IIR) study: Interclass reliability between clinicians and students, inter- and intrarater reliability among clinicians and among students, and validation with manual annotation in PACS by a board-certified spine surgeon.
Fig 3.
Lateral outline (red) for radiographic image (left) and lateral projection of optical back surface scan (right).
Fig 4.
Example of smoothed SPL and ISL for sagittal (left image) and coronal (right image) planes with indication of the positions of C7, T12, and L5 (black).
Table 2.
Median RMSE and SD between the smoothed ISL lines from the 3 repetitions for all 6 clinicians and all 10 students.
Table 3.
Correlation [95% confidence interval] of Cobb angles of coronal, kyphotic, and lordotic curvatures calculated from smoothed ISL lines drawn by clinicians and students and manually annotated in PACS by a board-certified spine surgeon.
Table 4.
Median differences (IQR) between clinicians and students for Cobb angles of coronal, kyphotic, and lordotic curvatures calculated from the smoothed ISL lines and median differences between Cobb angles calculated using PCdare and manually annotated by a board-certified spine surgeon in PACS.
Fig 5.
Difference between coronal Cobb angles (left), kyphotic Cobb angles (middle), and lordotic Cobb angles (right) calculated from smoothed ISL lines drawn by clinicians and students (interclass) and manually annotated by a board-certified spine surgeon in PACS (validation).
Table 5.
Median absolute difference (MAD) and IQR between the coronal Cobb angles from the 3 repetitions from a single rater for all 6 clinicians and all 10 students.
Fig 6.
Alignment errors: RMSE (left image) and SD (right image) between lateral outline of the optical back surface scan and lateral outline of the radiographic image for all patients in the datasets of Studies 1,2, and 3. The dashed line indicates that an outlier was visually truncated (original value: 36.3 mm).
Table 6.
Median Pearson correlation coefficient (PCC) and interquartile range (IQR) for all 3 datasets from Studies 1,2, and 3. Each PCC is separately calculated for the sagittal and coronal plane, and for the original smoothed line markings of SPL and ISL and the smoothed line markings after applying a Procrustes transformation to the ISL.
Fig 7.
Correlation between smoothed SPL and ISL lines for the sagittal plane (top left) and the coronal plane (top right); Shape correlation: after applying a Procrustes transformation, for the sagittal plane (bottom left) and the coronal plane (bottom right).
The dashed lines indicate that outliers were visually truncated (min value: −0.94).