Medical Chest X-ray Image Enhancement Based on CLAHE and Wiener Filter for Deep Learning Data Preprocessing

In medical imaging, an X-ray image generated using a flat panel detector (digital) typically has poor image quality, affecting the capability of successful medical diagnosis based on the images. The image enhancement process intends to provide better interpretability of the information contained in the images. The main problems considered for medical images include poor quality and low contrast. Therefore, the general objectives of image enhancement include contrast improvement and noise reduction. This study proposes an upgraded X-ray image enhancement hybrid algorithm that utilizes and consists of the Contrast Limited Adaptive Histogram Equalization (CLAHE) method combined with the Wiener filter. Based on the performance metrics results, including MSE, PSNR, and Entropy, as compared to the existing CLAHE method only, the proposed methodology has a lower MSE signifying lower error, a higher PSNR representing a lower amount of distortion, and higher information entropy which indicates higher obtained information. Furthermore, the implementation of the proposed approach is applied to 6000 X-ray images before deep learning classification modeling, which significantly improved from 50% to 78% validation accuracy. Therefore, the proposed method improves the image enhancement methodology and can substantially assist in diagnosing diseases.