Explainable deep learning-based multiclass classification of foot radiographs into normal, plantar fasciitis, and flatfoot

Key Points

• Classification accuracy reached 85%, differentiating between normal, plantar fasciitis, and flatfoot.
• Deep learning algorithms effectively analyzed foot radiographs, showing promise for clinical use.
• The approach utilized a dataset of labeled radiographs, enabling the model to learn distinct features of each classification.
• Results may enable more accurate diagnoses in orthopedics, though external validation is needed.