EfficeintYoNet: An Attention Based Lung Cancer Detection Through Hybrid CNN Approach

Lung cancer is a major cause of cancer deaths worldwide and is brought on by imbalanced cell development. Lung cancer is brought on by uneven cell proliferation, which accounts for a significant portion of cancer-related fatalities worldwide. Early detection is crucial for improving survival rates. This study uses a hybrid model that combines YOLOv8, an object detection model, with EfficientNetB7, a feature extraction and classification model, to identify lung cancer from Computed Tomography (CT) scans. We start by detecting pulmonary nodules with a tuned YOLOv8 model and then use EfficientNetB7 to extract features and classify the severity of these nodules. YOLOv8’s outputs act as a self-attention mechanism, enhancing the focus of the classification model. Using the IQ-OTHNCCD Lung CT dataset, we split the data into 80% for training and 20% for testing. Also, a preprocessing pipeline is created for a swift training process. The model achieved 99.88% training accuracy and 97.72% testing accuracy, demonstrating its effectiveness. Using this model lung cancer can be detected at an early stage, which will help save many lives worldwide.