Lung Anatomy Reconstruction with Deep Learning

Reconstructing lung models to better understand aerosol deposition
Thematic area: Health, Projects
Financing: IFAB call for projects
Enabling Technology: Advanced Modeling and Simulation, Artificial Intelligence, Digital Twin, Image recognition

Most lung models currently available to study aerosol deposition in the lungs are not patient-specific and do not conform to their actual lung volume/lung function.

Deposition models could be vastly improved by adapting currently available deep learning-based reconstruction algorithms to generate better reconstruction of the lung as a whole, especially by integrating the upper airways and the trachea-bronchial tree and by fitting the generated model to the actual shape of the lung lobes, which can be acquired from scans.

This anatomical reconstruction will form the basis for multiscale deposition simulation that enables better design of the therapeutic aerosol.


Reconstruct the entire human trachea-bronchial tree from deep learning analysis of patient CT scans integrated with generative algorithms for the distal airway portion.

The initial challenge

Although 90% of AI/ML-assisted interpretation of CT scans is for diagnostic purposes, only a very limited amount of software includes algorithms capable of generating a complete tracheobronchial tree by reconstructing the portion below the resolution of the CT scan. Of these, only a small part is constructing the missing portion of the tree based on the postmortem statistical distribution, i.e., generating physiologically realistic trees. Even a smaller subset is open source and freely available.

The solution

Once the tracheo-bronchial tree generation tool is tested and made available, Chiesi will incorporate it into a larger project to create a digital twin of the human lung.

The ultimate goal of such a tool, from the company’s perspective, is to have a high-fidelity predictive tool for therapeutic aerosol deposition to assist in the design and optimization of new pharmaceutical products.


The main outcome will be the creation of a prototype instrument for tracheo-bronchial tree reconstruction from patient CT scans.

Enabling remote access to the prototype instrument will have great impact in several contexts: medical (enabling real-time diagnostics after CT scans), safety and environment (providing lung morphologies to study toxic or carcinogenic aerosol deposition), pharmaceuticals (paving the way for physiologically based pharmacological and pharmacokinetic studies).

Keeping the prototype online and available will ensure further development of the algorithms and related testing on real high-throughput use cases.


For further information, please contact:

Sustainable Development Goals


Share on:

You may also be interested in:

Stay updated on the latest IFAB events and projects.Subscribe to our monthly newsletter