Detecting pediatric foot deformities using plantar pressure measurements: A semi-supervised approach

This paper presents a semi-supervised data-driven approach to identify pediatric foot deformities using plantar pressure measurements. Essentially, the developed merges desirable features of kernel principal components analysis as feature extractor and Kantorovich Distance-driven monitoring scheme for detecting deformities. For extending flexibility proposed scheme, density estimation based nonparametric decision threshold is adopted. The method assessed via publically available data containing three types footsteps (i.e., normal, flat, cavus). detection results show that proved promising results, thus, outperforming commonly applied schemes.