A Deep Learning Approach to Improve Pulmonary Cancer Diagnosis Using CNN

This is the authors' abstract. We don't add key points for this paper.

It has been proposed that physicians use machine learning-based lung cancer prediction algorithms to treat indeterminate pulmonary nodules that are discovered by accident or during screening. These methods might improve decisionmaking, decrease variability in nodule classification, and decrease the proportion of benign nodules that require unnecessary monitoring or investigation. In this work, we summarise the principal lung cancer prediction methods that have been put out so far and discuss some of the advantages and disadvantages of each. We discuss some of the challenges in creating and verifying such methods and outline the path to clinical acceptance. To ensure the best possible quality of input data for the deep learning models, the preprocessing step involves anatomical area segmentation, noise reduction, and image enhancement. The process of feature extraction involves using convolutional neural networks (CNNs) to extract intricate patterns and features from the lung pictures, which are then utilized to diagnose benign and malignant tumors. Typically, a deep neural network or an ensemble of models is used as the classification network to process this information and determine the final diagnosis. A faster and better outcome for patients may result from the recommended pulmonary cancer detection system's ability to assist medical professionals in accurately identifying malignant tumors in lung imaging. Deep learning is projected to be used in lung cancer detection, which will eventually result in earlier and more effective treatment, hence reducing the burden of this fatal condition.