Fast lung segmentation and registration to assess local ventilation from computerized tomography images of patients with acute respiratory distress syndrome

The project will tackle the lack of fast and reliable tools to personalize the artificial ventilation settings for patients with acute respiratory distress syndrome (ARDS). These patients need artificial ventilation to survive while the cause of the disease is treated, but inappropriate settings can increase mortality, and clinicians lack tools to assess the response of the patient's lungs to artificial ventilation. Lung aeration maps and biomarkers useful for the clinicians, such as alveolar recruitment and hyperinflation, can be extracted from pairs (e.g., inspiratory/expiratory) of 3D chest computed tomography (CT) images, provided accurate lung segmentation and registration methods are available, but conventional algorithms fail due to the lack of contrast typical of ARDS. Yet, thanks to our partnerships with physicians and the resulting access to unique annotated databases, we have recently shown that deep learning methods can successfully perform these tasks. Our aim today is to make our models available to intensive care units, which requires to make them more robust, and to guarantee their high performance, whatever CT scanner is used. To achieve this, the project aims to improve the generalizability of the models by both optimizing the learning strategy (domain adaptation) and diversifying the database. Prior knowledge of the lung shape will be introduced via a parametric model based on spectral shape analysis, and a new method will be developed for generating numerous ultra-realistic image pairs with ground-truth lung shape and deformations. These hybrid images will combine thorax morphologies and lung masks from easy-to-segment images of patients without ARDS, on the one hand, and lung lesions, heterogeneous deformations and density changes learned from patients with ARDS, on the other. The models will be implemented in a graphical interface for use in clinical trials aiming to propose and validate image-assisted ventilation strategies and, ultimately, reduce patient mortality.

Funding requested from ED EEA