A Butterfly Malar Rash Detection Model for Early Systemic Lupus Erythematosus Diagnosis

As estimated by the Lupus Foundation of America, at least 5 million people worldwide including 1.5 million Americans (approx.) have some form of Lupus. Among the different forms of Lupus, Systemic Lupus Erythematosus (SLE) is most prevalent. Diagnostic procedures for SLE present challenges due to its’ elusive etiology and diverse symptomatology. However, among the symptoms of SLE, the Butterfly Malar Rash (BMR), manifesting as a butterfly-shaped rash across the cheekbone and nasal bridge, is a prominent indicator. The challenge in detecting BMR is due to their similarity with other rashes from rosacea, and fifth disease. In this paper, we develop a CNN-based model that can differentiate between BMR and other rashes in facial images. We also conducted a comparative study with other BMR detection models and observed that our proposed model outperforms the existing models with an accuracy of 93.2% on the dataset used in this work.