Automated detection of plantar pressure ROIs based on multiple frame data

Semi-automatic or manual method were usually used to define the region of interest (ROI) of dynamic plantar pressure for foot's biomechanical research. However, the existing methods are time-consuming, complicated, or unrepeatable. So, automatic method to extract ROIs is needed. Moreover, spatial adjacency of ROIs in the forefoot put forward the challenge for automatic methods. The objective of this study is to find out an automatic method for defining the ROIs using dynamic plantar pressure information of multiple frames. Twenty subjects were asked to participate in tests. The plantar pressure data and 3D kinematical data were collected. The maximal inter-frame difference during stand phase was calculated. Then the locations of ROIs were determined by using maximal inter-frame difference and priori knowledge about spatial relation of anatomical location of ROIs. Recognition accuracy by using our method was compared with by using Motion Analysis. The results indicated that using inter-frame difference was able to produce recognition accuracy 100%, 100%, 90%, 100%, 90%, 90% for halluces and metatarsus I to V, respectively, which had highly agreement with the result of 3D motion capture system. Furthermore, our method is simple and discreet, which could be employed in wider applications.