Artificial Intelligence Algorithms for the Detection of Pathologies Related to Lung Cancer through Image Analysis using Convolutional Neural Networks and Data Augmentation: a systematic mapping of the literature
Systematic mapping of 96 articles (2015–2026) on AI/deep learning applications in lung cancer detection via medical image analysis Convolutional Neural Networks (CNNs) combined with transfer learning and data augmentation emerge as the most effective techniques for improving diagnostic accuracy AI and DL models demonstrate high sensitivity and specificity, offering a viable alternative for early lung cancer diagnosis Key barriers to clinical adoption include lack of standardized datasets, model
Analysis
TL;DR
- Systematic mapping of 96 articles (2015–2026) on AI/deep learning applications in lung cancer detection via medical image analysis
- Convolutional Neural Networks (CNNs) combined with transfer learning and data augmentation emerge as the most effective techniques for improving diagnostic accuracy
- AI and DL models demonstrate high sensitivity and specificity, offering a viable alternative for early lung cancer diagnosis
- Key barriers to clinical adoption include lack of standardized datasets, model explainability gaps, patient privacy concerns, and unresolved ethical/social implications
- The study concludes that while AI/DL holds significant promise for lung cancer care, further research and regulatory frameworks are essential before safe deployment
Why It Matters
This systematic mapping provides AI practitioners and healthcare researchers with a consolidated overview of the current state of deep learning in radiology, highlighting both the proven effectiveness of CNN-based approaches and the critical gaps that must be addressed for real-world clinical deployment. For industry stakeholders, it underscores that technical performance alone is insufficient—regulatory, ethical, and standardization challenges remain the primary bottlenecks to adoption.
Technical Details
- Scope: 96 peer-reviewed articles sourced from PubMed, IEEEXplore, Scopus, and Web of Science, covering publications from 2015 to 2026
- Core Architecture: Convolutional Neural Networks (CNNs) are the dominant model class, with transfer learning identified as a key strategy for overcoming limited labeled medical data
- Data Augmentation: Highlighted as a critical technique for improving model generalization and accuracy, particularly given the scarcity and imbalance of medical imaging datasets
- Performance Metrics: Studies consistently report high sensitivity and specificity for AI-assisted lung cancer detection, though exact benchmark figures vary across papers
- Clinical Context: Focus on radiology image analysis (CT scans, X-rays) for early-stage pathology detection, aiming to reduce reliance on expert radiologists and accelerate diagnosis
Industry Insight
- Regulatory readiness is the next frontier: Organizations investing in medical AI should prioritize building explainability frameworks and standardized evaluation pipelines, as these are identified as the top barriers to clinical adoption
- Data standardization will be a competitive differentiator: The lack of standardized datasets across institutions suggests that companies developing shared, well-annotated medical imaging benchmarks will gain significant influence over the field
- Ethical and privacy infrastructure is non-negotiable: Any commercial deployment of lung cancer AI tools must embed patient privacy safeguards and ethical review processes from the outset, as these are flagged as essential for responsible and safe application
Disclaimer: The above content is generated by AI and is for reference only.