Enhancing Content-Based Image Retrieval for Lung Cancer Diagnosis: Leveraging Texture Analysis and Ensemble Models

In image retrieval systems, many search engines rely on text-based methods, extracting information from descriptions and tags to retrieve relevant images. However, this conventional approach often overlooks the optical or semantic information present in photographs, leading to errors in the retrieved results. To address this issue, it is essential to consider content-based information such as intensities, edges, and pixel values when designing image retrieval systems, especially in domains like medicine. This study focuses on developing a Content-Based Image Retrieval (CBIR) system for the domain of medicine, with a particular emphasis on lung cancer detection. Early identification and treatment of lung cancer are critical for patient outcomes, making suitable Computer-Aided Diagnosis (CAD) systems essential for assisting clinicians. The proposed ensemble model incorporates various feature extraction methods, including image processing techniques and wavelet transform, to harness the valuable information from images. These features are then fed into a deep learning model to assess their effectiveness in image retrieval tasks. By combining machine learning and image processing techniques, this work presents a comprehensive CBIR system capable of efficiently retrieving relevant medical images. Such advancements in image retrieval technology hold significant promise for aiding medical professionals in diagnosing and treating diseases like lung cancer promptly.