Leveraging Deep Learning Models for Enhanced Detection of Early-Stage Lung Cancer in CT Scan Images

Patients afflicted with lung cancer have a high mortality rate, making it a deadly condition. It is possible to save patients' lives by detecting this disease early and correctly determining the stage of lung cancer. Medical professionals face challenges when it comes to accurately and quickly diagnosing lung cancer using various image processing, biomarker-based, and machine automation methodologies. This research makes use of CT scan images extracted from lung images acquired at a hospital. Traditional approaches rely on the patient's permission before providing CT scan images to diagnose lung cancer. Lung Cancer Cell Detection using Fusion Neural Network (LCCDC-FNN) is a new approach to early and accurate diagnosis that is proposed in this research article. Deep neural networks are employed to extract the features from the CT scan images. To save patients from this deadly disease, accurate feature extraction is crucial for early detection of malignant cells. This research aims to improve diagnostic accuracy by using an advanced 3D-convolution neural network (3D-CNN) in conjunction with U-Net and Mask R-CNN. Differentiating between benign and malignant tumors is also made possible by the proposed method. The results validate the feasibility of the suggested hybrid deep learning (DL) method for early detection of lung cancer when assessed using conventional statistical methods. The empirical results obtained from the proposed mechanism demonstrate that the proposed improved model provides U-Net, R-CNN and 3D-CNN classifications with 83.4%, 70% and 80.3% accuracy and LCCDC-FNN has accuracy of 98.99%.