VGG19+CNN: Deep Learning-Based Lung Cancer Classification with Meta-Heuristic Feature Selection Methodology

Abstract

Lung illnesses are lung-affecting illnesses that harm the respiratory mechanism. Lung cancer is one of the major causes of death in humans internationally. Advance diagnosis could optimise survivability amongst humans. This remains feasible to systematise or reinforce the radiologist for cancer prognosis. PET and CT scanned images can be used for lung cancer detection. On the whole, the CT scan exhibits importance on the whole and functions as a comprehensive operation in former cancer prognosis. Thus, to subdue specific faults in choosing the feature and optimise classification, this study employs a new revolutionary algorithm called the Accelerated Wrapper-based Binary Artificial Bee Colony algorithm (AWBABCA) for effectual feature selection and VGG19+CNN for classifying cancer phases. The morphological features will be extracted out of the pre-processed image; next, the feature or nodule related to the lung that possesses a significant impact on incurring cancer will be chosen, and for this intention, herein AWBABCA has been employed. The chosen features will be utilised for cancer classification, facilitating a great level of strength and precision. Using the lung dataset to do an experimental evaluation shows that the proposed classifier got the best accuracy, precision, recall, and f1-score.