Classification of Footprints for Correctives in Orthopedics

Foot disorders, a commonly overlooked issue, are prevalent in developed societies. These disorders can have a significant impact on a person’s quality of life and can even be debilitating, regardless of age. Non-invasive pedobarographic examinations, which evaluate the pressure fields of the plantar surface of the foot and a supporting surface, allow for analysis of a patient’s gait and posture using 2D footprints or scans. This data can then be used by orthopaedists to create customized shoes or insoles as correctives. However, the lack of standardized protocols and guidelines and the scarcity of evidence-based information can lead to a subjective evaluation and selection of correctives by the orthopaedist. This study proposes an objective and quantifiable method for the classification of footprints using computer vision paradigms, in order to create more appropriate correctives. The results show that the proposed machine learning models are able to correctly identify the required one of three correctives with an accuracy of 70% for RGB scans and 49% for blueprints. Based on the current results, future work should focus not only on the classification of suitable correctives, but also on the determination of the corrective’s position to ensure the best possible patient outcomes.