Fig 1.
Stages in the generation of 3D-PS outputs.
Left panel: projection of white light strips on the body surface, allowing reconstruction of skin surface topography through computer algorithms. Centre panel: raw body image extracted by computer algorithms. Right panel: electronic tape measure applied, following automatic extraction of key body landmarks.
Table 1.
Feasibility in performing 3D photonic scans in young children in a field setting.
Table 2.
Age and size by scan outcome in those who underwent a scan.
Fig 2.
Age-associated reductions in the percentages of children refusing to be scanned, or producing scans that were not technically successful.
The rate of refusal fell steadily from 6% at 5 years to zero at 11 years, whereas the percentage of failed scans decreased with age, but remained ~25% even in the older age groups.
Table 3.
Reasons for scan failure in the whole sample and by ethnicity.
Fig 3.
Scan characterised by technical failure, due to inability to extract chest girth data because of atypical posture.
However, the raw scan was of good quality, indicating that the problem lies in the software.
Fig 4.
Scans characterised by technical failure, due to inability to locate leg girths (left hand panel) and movement of the arms during the scanning process (right hand panel).
Fig 5.
Scans characterised by technical failure, due to artefacts from inappropriate clothing (left hand panel) and movement of the scanning truck (right hand panel).
Table 4.
Mean, variability and precision of shape outputs by 3D-PS and manual methods.
Fig 6.
Plots of 3D-PS values (y-axis) against manual values (x-axis) for each of the eight outcomes, superimposed on the line of identity.
The plots illustrate a general tendency for 3D-PS values to exceed manual values, with this over-estimation varying between outcomes. The magnitude of 3D-PS bias is described in detail in Table 5 and supporting information file S1 Table.
Table 5.
Correlations between 3D-PS and manual outcomes for the whole sample and for each ethnic group.
Table 6.
Biases between 3D-PS and manual outcomes.