COVID-19 infection manifests in its severe form as acute alveolo-interstitial and vascular pneumonitis. However, long-term outcome remains unknown. A progression to fibrosing pneumonia could affect 10 to 30% of survivors of severe forms, making it a public health problem through secondary disability. The project concerns the analysis of CT images of patients followed at the Avicenne APHP hospital (Bobigny) in collaboration with two expert image treatment/modelling teams for an evaluation of vascular remodelling and mechanical simulation of the regional lung compliance. These parameters obtained at 2-4 months of COVID-19 infection will be correlated with clinical, CT, and functional data at 6 months and 1 year follow-up. This project should make it possible to understand the COVID-19 infection manifests as an acute alveolo-interstitial and vascular pneumonitis in its severe form. However, long-term outcome remains unknown. It has been hypothesized that a progression to fibrosing pneumonia could affect 10 to 30% of survivors of severe forms, making it a public health problem through secondary disability. But there is no parameter which could currently allow to predict such an outcome. The SILICOVILUNG project is based on the CT images analysis of patients followed at the Avicenne APHP hospital (Bobigny, France) after severe COVID pneumonitis in collaboration with two expert image treatment/modelling teams [ Artemis Telecom Sud-Paris (Evry, France) and LMS INRIA Ecole Polytechnique (Palaiseau, France)] for an evaluation of vascular remodeling and mechanical simulation of the regional lung compliance using a poromechanical model. These parameters obtained at 2-4 months of COVID-19 infection will be correlated with clinical, CT, and functional data at 6 months and 1-year follow-up. This project should make it possible to understand the mechanisms of regional physiology in the evolution of COVID-19 pneumonias in their severe forms and to anticipate the development or not of sequelae, particularly fibrosing.

"The SILICOVILUNG project is based on the retrospective analysis of pulmonary CT images of patients followed after severe COVID pneumonitis at the Avicenne APHP hospital (Bobigny, France). CT series will be retrieved from the data bank of the radiology and pneumology departments (Avicenne APHP hospital Bobigny, France).

Records ( CTscan and clinical data) will be retrospectively included if they fulfilled the following criteria: patients > 18y with initial severe pneumonitis ( COVID-19 PCR+; oxygen therapy > 3L/min; at least 25% lesions on CT scan; no mechanical ventilation; not treated for pulmonary embolism). Records from patients with a history of COPD and/or ILD will be excluded. Records from thirty patients will be selected on data obtained at 2-4 months after the initial COVID pneumonitis episode: twenty patients with significant lung lesions observed on the CT at 2-4 months, ten patients with DLCO <65% associated with faint pulmonary lesions on the CT at 2-4 months.

Pulmonary CT images will be treated by mathematical analysis and modeling in collaboration with two expert image treatment/modelling teams [ Artemis, Samovar lab Telecom Sud-Paris (Evry, France) and LMS INRIA Ecole Polytechnique (Palaiseau, France)] A 3D evaluation of the pulmonary small arteries remodeling/pruning will be done on multislice CT using a method recently developed in collaboration with the Artemis Samovar team.

A 4D evaluation of the local pulmonary compliances will be done using a recently developed model of the lungs at the breathing time scale and the organ space based on a general poromechanical formulation compatible with large strains and thermodynamics.

The vascular modeling/pruning observed patterns on the 2-4-month follow-up pulmonary CT will be identified, described and quantified using an order of remodeling. The poromechanical observed patterns on the 2-4-month follow-up pulmonary CT models will be identified, described and quantified using an order of porosity and stiffness in various regions of the lung.

These parameters obtained at 2-4 months of COVID-19 infection will be correlated with clinical, CT, and functional (DLCO) data at 6 months and 1-year of follow-up The primary outcome will be the identification of pulmonary vascular pruning and poromechanical patterns predictive for evolutive sequalae post COVID19.

The secondary outcome will be the identification and order of magnitude of pulmonary arterial pruning/vascular remodeling and DLCO impairment at 2-4 months and 1 year post initial COVID pneumonitis "

Inclusion Criteria:

1. Patients who are informed and have not objected to participating in the research.
2. Patients with initial severe pulmonary disease [COVID + PCR, O² supplementation > 3l/min O²; > 25% CT lesions visible on the CT scan at the time of hospitalization, no pulmonary embolism and no mechanical ventilation].
3. Fifty files will be selected according to the types of short-term sequelae observed during the first check-up at 2-4 months: 30 on an aspect of alveolar and interstitial CT scan lesions associated or not with respiratory functional impairment; 20 others will be selected on the existence of significant functional impairment (DLCO <65% of the theoretical value) without or with few CT abnormalities (situations where sequelae of pulmonary vascularization would potentially predominate).

Exclusion Criteria:

1. Patients under 18 years old.
2. Patients with a history of. of interstitial lung disease or COPD.
3. Patients under guardianship.
4. Patients receiving state medical aid.

Retrospective analysis of radiological and clinical data from patients with postCOVID follow-up.