A Model for Lung Nodule Detection using a Hybrid Approach by Combining YOLOv۵ and ResNet۱۰۱ Pretrained Artificial Intelligence Models

Lung cancer is among the most common and deadly cancers worldwide, and its timely diagnosis can significantly improve survival rates. Artificial intelligence technologies assist in the early detection of lung cancer by analyzing data and recognizing patterns in medical images. These technologies enhance diagnostic accuracy and benefit both patients and healthcare systems by reducing treatment costs. In this research, our goal is to develop an automated system for identifying and detecting cancerous regions in lung CT scan images using deep learning models. Transfer learning techniques in neural networks are employed to achieve more accurate simulation and identification of cancerous areas. In this regard, two primary models, YOLOv۵ and ResNet۱۰۱, have been utilized. The results of this study demonstrate that combining these two models can improve the accuracy of lung cancer detection systems and serve as an effective tool for early diagnosis of the disease. The best results were obtained using the Adam optimizer for the ResNet model and the Momentum and SGD optimizers for the YOLO model, achieving a precision of ۸۳.۷۲% and a recall of ۹۷.۷۲%.