An Efficient Framework for Lung Cancer Prediction Using LBP Extraction and Cascaded CNN

Lung cancer, which affects both men and women, is currently the leading cause of cancer related fatalities globally. Lung cancer is mostly caused by smoking. It is thought that 90% to 95% of cases of lung cancer begin in the epithelial cells lining the bigger and smaller airways (bronchi and bronchioles), while the illness can grow elsewhere in the lung. Medical responsibilities make a better treatment decision for patient diagnosis and treatment to find a lung cancer at early stage using the classification models. This study’s main goal is to identify lung cancer by employing Cascaded Convolutional Neural Networks (CNN) with Local Binary Pattern and Wiener Filter. The lung image dataset is taken as the input. The lung images have been pre processed by Wiener filter. After, the processed images are segmented using the Otsu Binary Threshold method. The features of the cancer affected images are to be extracted using the Local Binary Pattern approach. This work implements in a python platform for developing a classification of lung images using the Cascaded CNN classifier. The proposed Cascaded CNN technique provides more accuracy than other classification models for predicting lung cancer. The proposed classifier shows the highest specificity 95% of and sensitivity of 94%.