FILM: Functional Imaging and Modelling of the Lungs

Heart ventricles, pulmonary arterial tree and emboli segmented out from a thoracic CT scan

 

Coordinator

Maciej ORKISZ <maciej.orkisz [at] creatis.insa-lyon.fr>

webpage: https://www.creatis.insa-lyon.fr/site/en/users/orkisz

About

This project stems from the pulmonary part of the COVID-19 transversal project.

We are investigating image processing techniques for the analysis of lung parenchyma and vascularization in three-dimensional (3D) thoracic computed tomography (CT) images. This imaging modality provides information on the air-to-tissue ratio within the lung parenchyma. Changes between CT scans acquired under different respiratory conditions (e.g., end-inhale and end-exhale) can be used to assess surrogate measures of ventilation, while analysis of pulmonary vascular tree structures can help understand perfusion disorders. 

Methods

We use deep learning techniques to delineate the lungs and vascular trees despite contrast changes due to pathological conditions.

We incorporate a priori shape models to increase the robustness of our methods.

We also leverage image-registration methods to align lung anatomical structures between different scans and calculate surrogate 3D ventilation maps.

Back to Myriad page