The “Lung Anatomy Reconstruction with Deep Learning” project, funded by Fondazione IFAB, is published today in an international scientific journal. An initiative that involved Chiesi Farmaceutici, Quantyca, E4 Computer Engineering, and Spoke 6 ICSC, with the goal of reconstructing the entire human tracheobronchial tree from patient CT scan analysis, integrating generative algorithms for the distal portion of the airways — the part that no scan can directly “see.”
Why it matters. Today, 90% of AI/ML applications on lung CT scans are for diagnostic purposes. Only a tiny fraction include algorithms capable of generating a complete bronchial tree, reconstructing the portion below CT resolution in a physiologically realistic way. Even fewer solutions are open and available to the scientific community. The IFAB project fits exactly into this space.
From research to industry. The value of this work does not end with publication: Chiesi will integrate the tool into a broader project to build a digital twin of the human lung, with the aim of obtaining a high-fidelity predictive tool for therapeutic aerosol deposition, to support the design and optimization of new pharmaceutical products.
An impact that goes beyond pharma. Making the prototype remotely available means opening up applications in three areas: real-time post-CT diagnostics, studying the deposition of toxic or carcinogenic aerosols for safety and environmental purposes, and physiologically based pharmacokinetic studies grounded in the patient’s real anatomy.
The role of IFAB. As the funding body, IFAB selected the project within its investment lines in research, innovation, and technology transfer, financially supporting its development. A case that confirms the Foundation’s mission: identifying and accelerating high-impact initiatives for the region and the production system, bringing them all the way to international scientific validation.
The paper is published in the International Journal for Numerical Methods in Biomedical Engineering (Wiley) and is available at the link.
🔗 Discover the project: Lung Anatomy Reconstruction with Deep Learning
