Currently, to our knowledge, there is little data on the combination of tools based on a similar concept to understand and evaluate ILDs. It is expected that this portfolio of multi-tool software implemented in radiology departments, applied to routine thoracic TDM, will provide additional qualitative and quantitative information in real time that will be of great help for diagnosis, prognosis prediction, and treatment decision-making in ILDs.

Thoracic CT scanning has revolutionized the definition of interstitial lung diseases (ILDs), some of which inexorably progress to pulmonary fibrosis (e.g., progressive pulmonary fibrosis or PPF), leading to early death or lung transplantation. Over the past decade, various treatments have shown effectiveness in slowing this fibrotic progression, but it is still not possible to define which patients might personally benefit from these treatments and when to prescribe them. Two major questions remain:

Why do some patients develop fibrosis despite seemingly appropriate treatment? What are the mechanisms driving this fibrotic progression? Hence, there is a great need to define biomarkers to answer these questions, particularly in the early phase. For more than 5 years, within a consortium including Avicenne Hospital APHP 93000 Bobigny, INSERM Unit 1272 Sorbonne Paris North University, and two partner laboratories (Mines Telecom and Ecole Polytechnique-INRIA, both belonging to the Institut Polytechnique), we have been developing the applications of artificial intelligence (AI) to lung imaging, extracting static and dynamic data from thoracic CT scans to aid in the diagnosis and follow-up of patients without additional examinations beyond standard care. Our project's objective is to identify patients at risk of progressive and irreversible fibrosis and those who could respond to antifibrotic treatments, by developing the identification of qualitative and quantitative biomarkers from the numerical modeling of routine thoracic CT scans.

Our program, which has just been funded in 2023 by the National Research Agency (ANR 2023 MLQ-CT), aims to:

Develop a portfolio of software tools, whose use should be facilitated in the hospital sector based on research prototypes already built and tested in our consortium for several years.

Apply them to a set of interstitial lung diseases (ILDs) known to be at risk of fibrotic progression.

Transfer these tools to the radiology department of Avicenne Hospital APHP. Conduct real-time experimentation between two pulmonology departments, one at Avicenne Hospital APHP and the other at Caen University Hospital, and the radiology department of Avicenne Hospital APHP, to validate the feasibility of using such biomarkers.

Inclusion criteria:

• Informed patients who have consented to participate in the research.
• Retrospective data from patients followed for ILDs who underwent two thoracic TDM scans in inspiration and expiration (IE-TDM) at least one year apart, meeting or not meeting the criteria for progressive fibrosis [presence of two of the following criteria within one year of follow-up: 1/clinical worsening, 2/radiological evidence of disease progression between the two IE-TDM scans, 3/decline in FVC ≥5% or absolute decrease in DLCO (corrected for Hb) > 10%].
• 300 records, based on retrospective data, will constitute the initial AVICENNE database (500 records will be selected at Avicenne Hospital APHP so that 300 meet the quality criteria for inspiration/expiration thoracic TDM scans)

Exclusion criteria:

• Patients under 18 years of age.
• Patients under guardianship/curatorship.
• Patients under AME (State Medical Assistance).

Retrospective data from patients monitored for ILDs who underwent two thoracic TDM scans in inspiration and expiration (IE-TDM) at least one year apart, whether meeting the criteria for progressive fibrosis or not.