A Novel Deep Transfer Learning-Based Approach for Automated Pes Planus Diagnosis Using X-ray Image

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Orthopedics

Background:

• Pes planus (flatfoot) is a common foot deformity characterized by the loss of the medial longitudinal arch.
• Diagnosis typically involves weight-bearing X-rays, but automated methods are lacking.
• Artificial intelligence (AI) shows promise for medical image analysis.

Purpose of the Study:

• To develop and validate a novel deep learning model for automated pes planus diagnosis from X-ray images.
• To address the gap in AI-based diagnostic tools for flatfoot.
• To establish a new automated system for pes planus detection.

Main Methods:

• A new dataset of weight-bearing X-ray images was collected and annotated by radiologists.
• Image augmentation and pyramidal patching (21 images per original) were employed.
• A deep learning framework using MobileNetV2 for feature extraction, iterative ReliefF for feature selection, and SVM for classification was utilized.

Main Results:

• The proposed model achieved a high accuracy of 95.14% with 10-fold cross-validation.
• Key features were identified and selected for efficient classification.
• The model demonstrated robust performance in diagnosing pes planus.

Conclusions:

• A novel deep learning model provides accurate automated diagnosis of pes planus using X-ray images.
• The developed framework shows potential as a valuable auxiliary tool in clinical settings.
• This study represents the first literature contribution to automated pes planus diagnosis via AI and X-rays.