Combining CNNs with U-Net for Improved Lung Tumor Detection and Analysis

This research explores the application of deep learning techniques for lung cancer segmentation and classification. By combining U-Net with the VGG-19 architecture, the proposed approach aims to address the challenges associated with deeper networks, complex feature learning, and robustness to noise and overfitting. The Res-U-Net architecture is employed to identify lung regions in CT images, enhancing segmentation accuracy and efficiency. Residual connections help mitigate the vanishing gradient problem, enabling the model to learn more effectively and improve tumor detection. Various parameters were evaluated to assess the classification method’s effectiveness. Python was used for algorithm development and execution, while Google Colab provided the necessary computational resources and collaborative environment. This research contributes to the advancement of lung cancer diagnosis by providing a robust and accurate segmentation and classification framework. The proposed approach leverages the strengths of U-Net and Res-U-Net to overcome challenges and improve the performance of lung cancer detection.