Application of Artificial Neural Networks in Predicting Malignant Lung Nodules on Chest CT Scans

Objective: To explore a simple method for improving the diagnostic accuracy of malignant lung nodules based on imaging features of lung nodules. Methods: A retrospective analysis was conducted on the imaging data of 114 patients who underwent lung nodule surgery in the Thoracic Surgery Department of the First People’s Hospital of Huzhou from June to September 2024. Imaging features of lung nodules were summarized and trained using a BP neural network. Results: Training with the BP neural network increased the diagnostic accuracy for distinguishing between benign and malignant lung nodules based on imaging features from 84.2% (manual assessment) to 94.1%. Conclusion: Training with the BP neural network significantly improves the diagnostic accuracy of lung nodule malignancy based solely on imaging features.