What are the Opportunities, Challenges, and Best Practices Regarding Computer-Aided Diagnoses in Tumour Detection for Lung Cancer?

Abstract

Background: Lung cancer is the leading cause of cancer-related mortality globally and in Canada. Early detection is associated with better health outcomes but traditional screening methods, such as low dose computed tomography, are limited by human error and increasing workload demands. Computer-aided diagnosis (CAD) systems, powered by deep learning (DL) and convolutional neural networks (CNNs), which are subfields within artificial intelligence (AI), have emerged as tools for enhancing lung cancer screening and diagnosis. Objectives: The objective of this paper is to explore the role of CAD systems in lung cancer detection, assessing their performance, with a focus on CNN-based approaches and to examine the impact of AI on Canadian radiologists, medical students, and healthcare implementation, addressing challenges and best practices for integration. Methods: A literature review was conducted using Ovid MEDLINE to analyze studies from 2019 to 2025 and clinicaltrials.gov to analyze Canadian clinical trials. The first objective focused on CAD systems and DL applications in lung cancer imaging, while the second examined AI’s influence on radiology professionals and implementation considerations. Results: CAD systems demonstrate high sensitivity in lung nodule detection, segmentation, and classification, reducing false positives and improving diagnostic precision. CNN-based models 3 such as ResNet and DenseNet enhance feature extraction, while transfer learning addresses challenges posed by limited medical imaging datasets. However, there are still many concerns regarding data bias, model interpretability, patient privacy, and liability. Medical students express anxiety about AI’s impact on radiology careers, highlighting the need for greater education and leadership within the field. The Canadian Association of Radiologists has issued guidelines to regulate AI implementation, focusing on data privacy, liability, and deployment. Conclusion: AI has the potential to revolutionize lung cancer diagnostics, complementing radiologists and improving early detection. However, challenges related to data quality, ethical considerations, and professional adaptation must be addressed through interdisciplinary collaboration and policy development. Future efforts should prioritize model explainability, improved datasets, and education for medical professionals.