A Transfer Learning for Intelligent Prediction of Lung Cancer Detection
Tabreer T. Al–Shouka, Khattab M. Ali Alheeti
Abstract
Cancer is a widespread and potentially lethal illness. There are many different types of cancer. The most common type of cancer is lung cancer. High mortality rates from lung cancer are major public health concerns. Therefore, early detection of cancer reduces the possibility of dying from the disease. A CT scan is used to determine the best course of treatment. Screening with CT has been shown to reduce mortality from lung cancer by detecting the disease at its earliest stages when it manifests as pulmonary nodules. Artificial Intelligence (AI) systems, which employ transfer-learning models, have shown great potential in enhancing the precision and speed of lung cancer diagnosis and treatment. Transfer learning enables adapting pre-existing models to new tasks, facilitating more efficient and effective medical data analysis. Additionally, these systems can assist clinicians in identifying the specific type and stage of lung cancer, which can guide treatment decisions. Our study explores the use of CNN transfer learning with RESNET, MobileNetV2, Xception, and VGG16 models for medical image analysis tasks; res-net achieved the highest testing accuracy in our research, at 0.94, with a testing loss of0.16, which has demonstrated the potential to enhance the accuracy and efficiency of healthcare AI systems.