An Improved Multiscale Faster R-CNN Based Lung Cancer Detection System

Lung cancer is a leading cause of cancer-related deaths, making early and accurate detection crucial. This project develops an automated detection system using Convolutional Neural Networks (CNN) to improve the identification of malignant nodules in lung CT scans. The process involves collecting a large dataset from public repositories, followed by preprocessing steps like normalization and augmentation to enhance input quality. The CNN architecture is designed to capture complex features of lung nodules, with training on expertly annotated data. Techniques such as dropout and batch normalization are used to prevent overfitting. The model's performance is evaluated using metrics like accuracy and ROC curve analysis, showing significant improvement over traditional methods. The findings suggest that CNNs offer a promising approach to early and accurate lung cancer detection, with potential for clinical application and future integration with other diagnostic tools. At the stage of diagnosis our major aim is to reach the higher level accuracy and efficiency. The improved model, based on lung cancer increases the performance, by combining the properties of various network structures. The proposed model achieved 96.2 % ACC, 94.7% Spec, 93.5% Recall, 98% Prec, and 98.5 % F1-Score. In addition to helping medical professionals with patient care and monitoring, the enhanced model may more accurately detect lung cancer, lower the possibility of incorrect and delayed diagnoses resulting from physician personal reasons, and enhance the overall evaluation and treatment process.