Using data compiled from CT scans, researchers at the Technical University of Munich (TUM) have developed a computational model of the lung that can be used to predict and prevent injuries caused by mechanical ventilation—a tool, the researchers hope, that can improve survival rates for patients with COVID-19 as well as other conditions that require the use of ventilators.

“The real crux of the problem is that when we’re treating a patient using mechanical ventilation, up until now, there hasn’t been any way to detect overdistention of the lung tissue,” says Wolfgang Wall, PhD, professor for computational mechanics at TUM. From the main bronchial tubes through to the tiniest structures in the lungs, there are more than 20 levels of branching. Currently, there’s no method for measuring what’s happening in the smallest, microlevel branches of the lung during artificial respiration.”

Although some medical texts still—inaccurately—portray the lung’s air sacs (alveoli) as similar to grapevines and bunches of grapes, lung tissue actually has a more sponge-like consistency. And it’s through this fine-walled tissue where the exchange between the air and the bloodstream occurs. Breathing comprises an extremely complex mechanical interaction between the different types of tissue, the liquid film on the tissue and the flow of air.

For several years, TUM researchers have been working to develop ever-more sophisticated models to simulate the behavior of lung tissue and airflow. Together with improved methods of micromechanical testing on lung tissue samples, their research has resulted in the creation of a computational lung model.

This model is the basis of a computer program which can calculate the local strains which would be placed on the lung’s microlevel tissues by different ventilator settings. Having these data at hand, medical staff and doctors can adjust the ventilator settings accordingly to provide a protective ventilation.

The current clinical standards guiding treatment with mechanical ventilation use a patient’s body weight to determine optimal ventilator pressure settings. However, the program developed by Wall and his team models the actual lung based on data compiled from a CT lung scan. It even considers the condition of individual areas of the lung that have already been damaged by the disease or previous injuries.

By measuring the changes in pressure and volume that occur during an inhalation and exhalation cycle, the digital lung model calculates the individual mechanical characteristics of the patient’s lungs. The result: a digital “twin” model of the patient’s lungs. It is so precise, that it can accurately predict which ventilator settings will cause damage to the patient’s lungs.

Parallel to continuing his working group’s research together with clinical partners, Wall and three former colleagues founded the company “Ebenbuild” to bring their research into clinical practice as quickly as possible.

A key step in realizing this goal was automating the generation of lung models using artificial intelligence (AI). Wall and his team have harnessed the computing power of AI to developed a digital tool that can “map” a patient’s lungs—and which can even be used for early detection of COVID-19 infections.

“More than 80% of COVID-19 deaths are the result of acute lung failure,” says Wall. “And with long-term mechanical ventilation, the survival rate for our most critically ill patients drops to only 50%,” he adds. “The goal of our work is that in the future, at each ventilation site a digital lung model helps to optimize the ventilation to the patient’s needs so that we can significantly increase the chance of survival.”

Read more from TUM.

Featured image: Utilizing data compiled from a CT lung scan, the software uses artificial intelligence to calculate the condition and health of a patient’s lungs. In this image, damage caused by a COVID-19 infection is marked in orange. Courtesy, J. Richter / TUM