Utilization of Neural Network in the Diagnosis of Pes Planus and Pes Cavus with a Smartphone Camera

Category: Midfoot/Forefoot Introduction/Purpose: Pes Planus (flatfoot) and Pes Cavus (high arch) are common foot deformities requiring clinical assessment and radiographic imaging for diagnosis and subsequent management. While effective, traditional diagnostic methods pose limitations such as cost, radiation exposure, and accessibility, particularly in underserved areas. With the advent of artificial intelligence and its integration into mobile cameras, we aim to develop a tool that detects and classifies such deformities using photos taken with smartphone cameras. Methods: An algorithm that integrated a deep learning, convolutional neural network (CNN) into a smartphone camera was utilized to detect Pes planus and Pes cavus deformities. The study was conducted at a tertiary hospital with participants recruited from two orthopaedic foot and ankle clinics. The CNN model was trained and tested using photographs of the medial aspect of participants’ feet. Participants included subjects with standard foot alignment, pes planus, or pes cavus determined by an expert clinician using the Foot Posture Index (FPI). The model’s performance was compared to clinical assessment and radiographic measurements, including the lateral tarsal-first metatarsal angle (LTMA) and calcaneal inclination angle (CIA). Results: The CNN model demonstrated high accuracy in diagnosing Pes planus and Pes cavus, with an optimized area under the curve of 0.897 for Pes Planus and 0.898 for Pes cavus. It showed a specificity and sensitivity of 84% and 87% for Pes Planus detection, respectively; for Pes cavus, 97% and 70%, respectively. The model’s prediction correlated moderately and significantly with radiographic LTMA measurements (r=-0.65, P< 0.001), indicating the model’s excellent reliability in assessing food arch deformity. Conclusion: Our smartphone-based CNN model is highly accurate as a decision-support tool, and it is reliable and accessible for predicting Pes planus and Pes cavus deformities. This tool may be very useful in underserved healthcare settings and for patients with limited access to expert clinical assessment. Future efforts should enhance this technology for pediatric use and improve severity differentiation, paving the way for more personalized and inclusive healthcare solutions.