Artificial Intelligence in Medicine: Enhancing Pneumonia Detection Using Wavelet Transform

This research introduces a deep learning technique for classifying lung images using wavelet transform-based feature extraction in TensorFlow. The accurate and automated analysis of lung images is vital for diagnosing and treating lung diseases. In this study, we present a new approach that combines wavelet transform-based feature extraction with a convolutional neural network (CNN) to achieve precise lung image classification. The proposed approach involves several steps. First, the lung images undergo preprocessing to eliminate noise and enhance contrast. Next, the images are decomposed into various frequency sub-bands using wavelet transform. The resulting wavelet coefficients serve as features for the classification process. Additionally, we utilize our custom CNN architecture as a classifier in TensorFlow to categorize lung images as either normal or pneumonia. To assess the effectiveness of the proposed method, we utilized a dataset containing 5216 lung images. Experimental results demonstrate that the proposed approach achieves an impressive classification accuracy of 96.9% for lung images. Furthermore, our method outperforms other state-of-the-art techniques in the field of lung image classification.