A Study on Lung Cancer Detection Using Deep Learning

Abstract

Lung cancer is among the leading causes of mortality worldwide, and early detection is critical to improving patient survival. Recent advances in artificial intelligence particularly deep learning-based image analysis have opened new opportunities for earlier diagnosis. This study aims to develop a deep learning model to classify four categories of lung computed tomography (CT) images (malignant and benign) and to implement it in a mobile application. We developed a hybrid convolutional architecture combining MobileNetV2 and ResNet50 and trained it on the open-access “Lung Cancer 4 Types” dataset from Kaggle. Model performance was evaluated using standard metrics, including overall accuracy, precision, recall, and F1-score. The proposed model achieved an overall accuracy of 85.4%, with per-class F1-scores ranging from 80.0% to 94.0%, indicating effective discrimination among lung cancer categories. Finally, the trained model was converted to TensorFlow Lite and integrated into an Android application, thereby bringing deep learning solutions closer to clinical practice.