Reliability of measuring morphology of the paediatric foot using the Artec Eva hand held scanner

The growth and development of the paediatric foot throughout childhood is poorly understood. To inform theory that underpins clinical practice, there is a clear need to revisit our understanding of how the foot develops. Hand-held 3-D scanners provide portability and allow researchers to collect data about foot development in the children's natural environment. However, there are methodological challenges to consider: scanning the plantar surface in a static weight-bearing position, the children's ability to remain static for the duration of the scanning and software capabilities. The aim of this study was to determine the reliability of using a hand-held scanner to capture children's foot shape and size. For this study, 15 children aged two years (Group 1: n=5), five years (Group 2: n=5) and seven years (Group 3: n=5) were recruited. Children stood barefoot in a comfortable bipedal stance, on a Perspex platform of 550mm height. Their feet were scanned three times, including the plantar surface through the platform, using the Artec Eva (Artec Group, Luxembourg, Luxembourg) hand held scanner. Post-processing of the scans was performed in Artec Studio 12 (Artec Group, Luxembourg, Luxembourg). Data processing and statistical analysis of 3D data were performed in Matlab R2018a (The Mathworks, Natuck, USA), while linear measures were calculated in Foot3D (INESCOP, Elda, Spain). To assess reliability, root mean square error (RMSE) of 11 linear measurements, mesh deviations (Euclidean distances) of the 3D coordinates of corresponding vertices (after rigid registration of the meshes) and RMSE for shape-index (SI) and curvedness (CU) were calculated. Results showed good reliability for eight linear measures with an average RMSE of 1.14mm across groups and all measures (RMSE range: 0.19mm - 3.73mm). Three measures exceeded a RMSE of 2mm, two of which were from Group 1. Mesh deviation results showed good reliability in the older children (Group 2: deviations under 0.5mm: 73.03%, under 1mm: 94.12%, Group 3: deviations under 0.5mm: 68.82%, under 1mm: 96.20%), but not in the youngest group (deviations under 0.5mm: 53.19%, under 1mm: 85.83%). The heat maps of mesh deviations across the foot surface, indicate increasing mesh deviations in the toe and ankle area from Group 3 to Group 2, while Group 1 also had higher than 1mm deviations on the lateral and dorsal surface of the foot. Root mean square error for curvedness and shape-index for the 3 scans of the same foot decreased with increasing age, but in general indicated good reliability. The results of this study demonstrated that the hand-held scanner was reliable for capturing children's 3D foot shape, however there were methodological issues in the youngest group. In Group 1, the mesh deviation results demonstrated lower reliability in four distinct areas (toes, lateral and dorsal surface and ankle). The higher mesh deviations were a result of these children being unable to stand still for the duration of the scan and having a more variable stance on the platform between scans. The fact that the RMSE of two linear measures exceeded 2mm in the youngest group also supported this proposal. Future studies employing hand held 3D scanners should consider these results and handle 3D scanning data of two years old children with caution