Lung cancer detection: A systematic literature study

Lung cancer is the primary cause of cancer-related deaths. Lung cancer presents with symptoms only in its advanced stages. Machine Learning and Deep Learning can be used to detect lung cancer early. This study aims to find the state-of-the-art approach to detecting lung cancer. The topic at hand has both potential and challenges, which are highlighted by the diversity of datasets, model architectures, and methodological approaches. Interestingly, the incorporation of image data turns out to be crucial for practical uses, highlighting the shortcomings of models trained in the absence of such data. It becomes clear how important dataset size is, with larger datasets potentially providing benefits in terms of model robustness. While certain models such as CNN VGG-19, LCP-CNN, and FPSOCNN perform admirably, they also highlight subtle issues, like the requirement for refinement in order to properly categorize nodules and the expense of computing. Future research directions are informed by the identification of these strengths and limits, which highlight the necessity for customized optimizations and the taking into account of real-world constraints.