Hybrid Deep Learning Based GRU Model for Classifying the Lung Cancer from CT Scan Images

Lung cancer is a potentially fatal condition, posing significant challenges for early detection and treatment within the healthcare domain. Despite extensive efforts, the etiology and cure of cancer remain elusive. However, early detection offers hope for effective treatment. This study explores the application of image processing techniques, including noise reduction, feature extraction, and identification of cancerous regions within the lung, augmented by patient medical history data. Leveraging machine learning and image processing, this research presents a methodology for precise lung cancer categorization and prognosis. While computed tomography (CT) scans are a cornerstone of medical imaging, diagnosing cancer solely through CT scans remains challenging even for seasoned medical professionals. The emergence of computer-assisted diagnostics has revolutionized cancer detection and diagnosis. This study utilizes lung images from the Lung Image Database Consortium (LIDC-IDRI) and evaluates various image preprocessing filters such as median, Gaussian, Wiener, Otsu, and rough body area filters. Subsequently, feature extraction employs the Karhunen-Loeve (KL) methodology, followed by lung tumor classification using a hybrid model comprising a One-Dimensional Convolutional Neural Network (1D-CNN) and a Gated Recurrent Unit (GRU). Experimental findings demonstrate that the proposed model achieves a sensitivity of 99.14%, specificity of 90.00%, F -measure of 95.24%, and accuracy of 95%.