Lung Cancer Detection and Classification using Transfer Learning with Pre-trained VGG19 Convolutional Neural Networks

It is possible to diagnose and identify a variety of illnesses with the use of high-quality figures obtained from medical equipment. However, obtaining and storing these kinds of photos may be quite costly, and diagnosing a patient might take a long time. Artificial intelligence (AI) solutions for automatic diagnosis may play a major role in addressing the challenges of time and cost. The categorization of medical images may be effectively solved by pre-trained deep learning models. Lung cancer continues to be a widespread and serious type of cancer on a global scale. Timely and precise identification is crucial for successful treatment and enhanced patient results. This study presents a novel approach to the identification and categorization of lung cancer by using transfer learning via a VGG19 Convolutional Neural Network (CNN) that has been trained beforehand. Our work harnesses the power of transfer learning by leveraging the VGG 19 architecture, provides the basis for lung cancer diagnosis, having been pre-trained on a large dataset. Our carefully curated dataset encompasses over 14600 diverse lung images, covering benign, malignant, and normal cases, providing the model with a rich source of information. The model demonstrated outstanding performance metrics, achieving a training accuracy of 99.68%, 98.11% for validation accuracy and 97.93% for test accuracy. Such high accuracy holds great promise for expediting and improving the diagnostic capabilities of healthcare professionals. By leveraging Transfer Learning from VGG19, our model demonstrates adaptability to diverse medical imaging tasks, making it a valuable tool for real-world clinical applications.