FEDERATED-PET: AI at the service of lung cancer patients

The FEDERATED-PET project puts Artificial Intelligence at the service of medicine to improve the life expectancy of patients with lung cancer.

The objective of the project is to predict the response to immunotherapy of patients followed for lung cancer and to personalize the therapeutic strategy to improve the quality and life expectancy of patients.

The “FEDERATED-PET” Artificial Intelligence research project is led by the Prof. Olivier HUMBERT(Center for the Fight Against Cancer Antoine Lacassagne/University Côte d’Azur/3IA Côte d’Azur), in partnership with Marco LORENZIresearcher at the Inria center of the Université Côte d’Azur in the Epione team and holder of a 3IA Côte d’Azur chair.

The project started in November 2022 and will last 3 years.

The project involves 8 French hospitals, 4 internationally renowned research centers and is the 2022 winner of the Translational Research Program in Oncology (PRT-K) of the National Cancer Institute (INCa) – Directorate General for Healthcare ( DGOS).

Immunotherapy has become the standard first-line treatment for patients with metastatic lung cancer.

Although immunotherapy allows a very good tumor response with prolonged remissions hitherto not observed with other chemotherapies and targeted therapies, it only benefits approximately one in two patients.

There is therefore very active research to identify “biomarkers” to predict the effectiveness of immunotherapy and better target patients eligible for this treatment.

The ultimate goal is to better individualize the therapeutic strategy to improve the life expectancy of responder patients, while limiting toxicities and loss of time for non-responder patients.

New artificial intelligence algorithms enable “in-depth” analysis of medical images, to extract characteristics not visible to the human eye but potentially very informative on the future effectiveness of treatments. The difficulty is that these algorithms need to be trained on a large number of images to “learn” a combination of parameters capable of predicting the response to treatment.

However, since these medical images are private and confidential, it is currently difficult to build large secure databases outside of hospitals.

The “FEDERATED-PET” project aims to respond to this constraint on medical data thanks to an innovative technology developed by Inria: the Fed-BioMed software platform.

This is an innovative IT infrastructure allowing several hospitals to communicate with each other and train artificial intelligence models in a collaborative and secure way on heterogeneous data. This training is therefore done on the biomedical data of several hospitals, without the need to bring them out of the computer network of the hospital which produced them.

This ensures the confidentiality and security of this very sensitive data. This technology is called “Federated Learning”. The “FEDERATED-PET” project is the first large-scale federated learning initiative in France.

The “FEDERATED-PET” project is an ambitious project for the “federated” analysis of 1000 18FDG (18Fluoro-deoxy-glucose) PET (positron emission tomography) images of patients presenting with lung cancer and eligible for immunotherapy.

PET imaging provides a lot of non-invasive information on the tumor but also on the patient’s immune system. These 1000 PET images will be distributed in eight different hospitals.

The objective of the project will be twofold:

Develop a communication infrastructure between hospitals to collectively “train” the Artificial Intelligence algorithms, without ever leaving the medical data of the hospitals where it is produced.

Develop a new Artificial Intelligence tool using an algorithm capable, from the analysis of 1000 PET images of patients with metastatic bronchopulmonary cancer, to predict the future effectiveness of immunotherapy for, in the long term, to propose a personalization of the therapeutic strategy, adapted to the immuno-reactivity profile of each patient.

The project started in November 2022 and will last 3 years. It involves 8 French hospitals and 4 internationally renowned research centers.


HealthTech Côte d’Azur news

We wish to thank the writer of this short article for this awesome material

FEDERATED-PET: AI at the service of lung cancer patients


We have our social media profiles here as well as additional related pages here.https://yaroos.com/related-pages/