Automated Lung Cancer Detection Using CNN From CT Scan Images: A Deep Learning Approach

Abstract

Lung cancer remains one of the leading causes of cancer-related mortality worldwide, largely because it is diagnosed at advanced stages when treatment options are limited. Conventional diagnosis relies on radiologists manually reviewing CT scans, a process that is time-taking and prone to error. This paper presents a lung cancer detection framework using a Convolutional Neural Network (CNN) to classify chest CT images as cancerous or normal without relying on hand-crafted features. The pipeline covers dataset acquisition from Kaggle, image preprocessing (resizing, normalization, noise reduction, and augmentation), automatic CNN-based feature extraction, and binary classification. The proposed CNN achieved 98.73% accuracy, 99% precision, 99% recall, an F-measure of 98.52%, and an error rate of 1.27%, outperforming four classical machine learning baselines — Decision Tree, Logistic Regression, Random Forest, and Naive Bayes — by a substantial margin.