Masking approaches to analyse plantar pressure data of new and confident walking infants

Due to its easy and straightforward use, regional analysis with the "standard" mask is the most common approach for quantifying plantar pressures in infancy. Such a mask, however, identifies foot regions based on typical foot proportions and pressure gradients. Alternatively, the use of a customised mask retaining infants' feet proportions has not been explored.

Does a customised mask scaled on infants' feet improve processing of pressure data collected during walking development compared with a standard mask?

Thirteen infants walked across an EMED xl platform. Steps were grouped applying eight foot-regions standard and customised masks. To evaluate masks' performance, peak pressure (PP) and contact area (CA) were extracted from each region, and mask. Intra-individual coefficients of variation were then calculated for each variable, and compared between masks using a Mann-Whitney U test (p < 0.05). Unsuccessful masks application was reported, expressed as percentage of data loss.

For CA variation, significant differences were found in all the regions but the lateral toes in new (Z = -0.184, p = 0.8540) and confident walking (Z = -1.562, p = 0.118). For PP variation, a significant difference was found in confident walking within the lateral midfoot (Z = -2.598, p = 0.009). With the standard mask, 22-27 % of data was lost in new and confident walking respectively, compared to 1.6-0 % with the customised. As a result, the customised mask characterised the more variable steps, demonstrating higher variation compared to the standard mask.

Identifying foot regions using a mask based on infants' feet proportions yielded an improved performance compared to the standard mask. With the customised mask, we retained almost all the steps and characterised the variability of the data, thereby providing an appropriate approach for infants' pressure data processing. Application of the customised mask could therefore be beneficial in future studies analysing highly variable data sets.