Publication: Lung cancer detection system using deep learning based on various data augmentation techniques
Loading...
Date
2023-08
Authors
Sioh, Zhe Heng
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Lung cancer is a significant global health concern, requiring timely detection and accurate diagnosis for better patient outcomes. Deep learning, using innovative
techniques like image analysis and machine learning, shows promise in developing lung cancer detection systems. These systems can identify abnormal features in medical images, potentially aiding early-stage diagnosis and improving patient survival rates. The project aims to create a system that detects lung cancer from CT scan images, categorizing tumours as normal, benign, or malignant through CNNs. The proposed framework employs techniques like SMOTE, class weight approach, and image augmentation to enhance model generalization and accuracy. The approach with the most favourable validation accuracy will be chosen for the final system. Besides, the LeNet-5 models using different techniques were compared, with the class weight approach showing the best performance. It achieved remarkable results in detecting lung cancer, with accuracy, precision, recall, and F1-score reaching 99.39%, 99.19%, 99.19%, and 99.30%, respectively. The system is implemented in a GUI application, allowing users to upload CT scan images for diagnosis, displaying the lung cancer classification and confidence percentage. In summary, this system offers significant assistance in early detection and improved accuracy of lung cancer diagnosis