R-LSTM-CNN Framework Based Lung Cancer Detection and Classification from Chest CT Images
A. Gopinath, P. Gowthaman, Mohammad Shabbir Alam, Jhakeshwar Prasad, S Senthurya, Vasudha Vasudha
Abstract
Lung cancer is one of the top causes of death worldwide, accounting for almost five million lives lost annually. Computed Tomography (CT) scan findings can be used to help diagnose lung conditions. The fundamental objective of this work is to detect malignant nodules in the lungs and classify the severity of lung cancer based on an input lung image. In this proposed approach to use state-of-the-art Deep learning techniques to precisely locate cancerous lung nodules. Preprocessing, segmentation, feature extraction, and model training are the first four steps of the proposed method. The major goal of this preprocessing is to extract the lung area from the CT scan image, along with any other ROIs that may be present. A huge digital image can be broken down into more manageable chunks using a process called segmentation. Intensity and gradient orientation histograms, as well as Gabor and entropy filters, are used in feature extraction. After leveraging information gain to choose useful features, the models are trained via R-LSTM-CNN. The proposed method outperforms the two most popular competing methods, CNN and LSTM.