An Artificial Intelligence Techniques in Lung Cancer 18F- Fluorodeoxyglucose Radiotracer PET/CT Imaging

Lung cancer is one of the most prevalent and deadly diseases in the world; thus, early identification is essential for increased survival rates. PET/CT scans with an 18 F-FDG radiotracer are crucial tools in the diagnosis of cancer, as they combine two imaging modalities to provide a more complete view than each alone. PET/CT scans don’t always provide the whole picture, though, because tumors are complicated and image interpretations can differ. AI can help with this. AI increases accuracy and lessens the workload for physicians by automating diagnosis and classification. These advanced algorithms, trained to match expert knowledge, not only detect key symptoms but also enhance image quality, leading to more precise diagnoses. This study examines the advancements in applying AI- algorithms to 18 F-FDG-PET/CT imaging for lung cancer, to enhance clinical outcomes. We conducted a comprehensive search on several databases to get the most recent works, using Springer, Scopus, PubMed, Web of Science, IEEE, and Google Scholar in the last five years (2019-2024). Identifying studies that met the criteria for inclusion that focused on AI-driven models applied to PET/CT data in lung cancers. The findings highlight the varying performance of AI models in lung cancer analysis using 18 F-FDG-PET/CT imaging, with deep learning models like Convolutional Neural Networks (CNNs) and Residual Neural Network (ResNet) achieving higher accuracy (The area under the ROC curve AUC 98-99%) compared to traditional machine learning approaches (AUC 63%). Challenges such as small sample sizes, lack of external validation, and computational complexity hinder broader clinical adoption. Furthermore, imaging quality, data consistency, and model interpretability limit effectiveness. The conclusion how AI has the potential to revolutionize lung cancer diagnosis and treatment. However, it also underscores the importance of continued research to tackle challenges like ensuring the models work well across different settings, making their decisions easier to understand, and adapting them for practical, real-world use.