Pedobarographic statistical parametric mapping (pSPM): A pixel-level approach to foot pressure image analysis

Traditional pedobarographic analyses conduct statistical tests on single pressure values extracted from discrete anatomical regions, a process which yields a low-resolution view of the continuous foot–ground interaction and which can involve substantial user interaction for region definition. Using image processing techniques derived from a cerebral imaging methodology called ‘statistical parametric mapping’ (SPM), we describe a fully automatic method that requires no anatomical assumptions or region definitions and that generates high-resolution continuous statistical maps across the entire plantar foot surface. Here, we demonstrate both pedobarographic SPM (pSPM) and its robustness to arbitrary foot postures by producing statistical maps for a sample of nine healthy young adults walking: normally, with everted feet, and with inverted feet. After spatially smoothing pedobarographic images, within-subjects (WS) and between-subjects (BS) registration were performed using an optimal rigid body transformation and an optimum affine transformation, respectively. Statistical tests were performed over all 742 foot pixels of the 270 registered images using a linear mass-univariate model and the resulting SPMs were compared qualitatively with results obtained using a traditional ten-region technique. SPMs were found to provide a qualitatively improved view of pedobarographic changes, but the more important finding was that regional pedobarographic statistics can misrepresent the trends of their constituent pixels and thus potentially lead to misinterpretations of foot function. Since pSPM is fully non-interactive, is robust to arbitrary foot posture, and provides rapid and easily interpretable results, it appears to be a suitable alternative to regionalization for routine pedobarographic analyses in both laboratory and clinic.