A Wavelet-CNN-Based Framework for Automated Diagnosis of Rheumatoid Arthritis

Rheumatoid arthritis (RA) is an autoimmune disease that causes chronic inflammation in the joints, leading to pain, swelling and stiffness which significantly affects patients’ quality of life. Early detection and diagnosis are crucial for successful clinical intervention and management to prevent disease progression. In this paper, we introduce a novel diagnostic framework that combines Wavelet Transform and Convolutional Neural Networks (Wavelet CNN) to improve feature extraction and classification performance in RA detection from hand radiographs. The proposed method first decompose input x-ray images into multi-resolution frequency components using the Discrete Wavelet Transform (DWT). This will emphasize critical pathological features such as joint space narrowing and bone erosions. These decomposed representations are then fed into a custom CNN architecture designed to learn both low-level texture patterns and high-level semantic features. The proposed model was evaluated on a labeled RA X-ray dataset and demonstrated superior performance compared to baseline CNNs, achieving an accuracy of 98%, sensitivity of 99%, and specificity of 99%. This approach leverages the strengths of both spatial and frequency domain analysis, offering a robust and interpretable solution for automated RA diagnosis and potential clinical decision support.