Deep Learning in Early Osteoarthritis Detection via Biomarkers Deep Learning for Differential Diagnosis in Equine Osteoarthritis: Exploring Synovial Fluid Biomarkers

Abstract

In the past, radiographic techniques have been used to diagnose osteoarthritis (OA), a common joint disease leading to lameness in racehorses. However, these methods typically only reveal the disease once irreversible damage has occurred. In this thesis, the focus shifts from imaging to biological biomarkers for earlier detection. Using synovial fluid biomarkers, this work explores the potential for deep machine learning models to accurately classify early stages of OA, thereby enabling timely interventions to prevent disease progression. In addition, a user-friendly web appli cation was developed to assist practitioners in making real-time diagnoses. Based on preliminary results, deep learning approaches, particularly those involving neu ral networks, can effectively differentiate between stages of OA, offering a promising tool for diagnosis and treatment. These findings suggest significant potential for improving diagnostic accuracy and, consequently, treatment outcomes in veterinary medicine.