Efficientnet-B3-Based Automated Lung Cancer Detection System for CT Scans

Lung cancer is a significant global health issue, responsible for around 7.6 million deaths annually, according to the World Health Organization (WHO). Early diagnosis is essential for enhancing treatment success and survival rates. However, accurately diagnosing and staging lung cancer through imaging techniques like Magnetic Resonance Imaging (MRI), X-rays, and Computed Tomography (CT) images remains a considerable challenge for medical professionals. This study presents an automated lung cancer detection system that utilizes Deep Learning (DL) methods, specifically a convolutional neural network (CNN) built on the EfficientNetB3 architecture, to enhance detection accuracy in CT scans. A comprehensive dataset of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$C T$</tex> scan images was sourced from the Kaggle repository, and various preprocessing techniques were applied to prepare the data for model training. The EfficientNet-B3 model was trained end-to-end to automatically extract key features and identify cancerous nodules. Our model achieved an accuracy of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 3. 3 3 \%}$</tex> in classifying lung nodules and determining their malignancy levels. To make this system userfriendly, we developed a web application using the Streamlit framework, providing an interactive interface for real-time diagnostics. The results of this study establishes that the presented CNN-based model significantly improves the detection precision of cancerous lung CT scans while providing ease of use through a user-friendly user interface. This automated system could support clinicians in making early diagnoses and planning treatments, potentially leading to better patient outcomes.