Lung cancer screening uses low-dose chest scans to find cancer early, but these scans can also reveal other diseases before symptoms appear. Interpreting large numbers of scans is challenging due to the limited number of radiologists worldwide. We introduce TANGERINE, an efficient and easy-to-use, open-source AI model that can analyse three-dimensional chest scans efficiently and accurately. The model learns patterns from thousands of scans and can then be adapted to detect lung diseases using only small amounts of new data. TANGERINE performs as well as, or better than, more complex systems while using less computing power. Its open-source design provides a foundation for future tools that could make advanced scan analysis more accessible, supporting earlier diagnosis and improved lung health worldwide.
