computational radiology

Radiology could provide a non invasive way to address different diseases and treatments. Providing automatic tools for the analysis of radiology data could help on screening, diagnosis or even prognosis of radiologists during their clinical practice. Our team collaborates with the best clinicians to develop relavant methods for processing of radiology data. Currently, we have active collaborations with clinicians in APHP hospitals, Gustave Roussy and Henry Dunant Hospital towards this direction including Hagnodice, Prism, Noesia and Archimedes Unit projects.

Towards cancer treatment

• In Martinot et al. 2023 we introduce a new class of Gamma index Passing Rate (GPR)-based loss functions for training of deep learning methods towards radiotherapy dose simulation.

• In Laousy et. al 2021 we propose a deep reinforcement learning method for accurate localization of the third lumbar area (L3) CT slice towards sarcopenia prediction.

• In Sun et al. 2018 we proposed one of the first radiomics signatures for CD8 cells and we indicated a promising way to predict the immune phenotype of tumours and to infer clinical outcomes for patients with cancer who had been treated with anti-PD-1 and PD-L1

Towards kidney transplantation monitoring

• In Milecki et al, 2023 ropose a framework that generates medical prompts using automatic textual data augmentations from LLMs. Our goal is to learn meaningful manifolds of renal transplant DCE MRI, interesting for the prognosis of the transplant or patient status.

• In Milecki et al. 2022 we propose a sequential architecture based on transformer encoders to predict the renal function 2-years post-transplantation, focusing on representation learning and missing data.

Towards lung interstitial diseases

• In Chassagnon et al. 2021 we propose an automatic deep learning-based disease quantification using an ensemble of architectures, and a data-driven consensus for the staging and outcome prediction of the COVID-19 patients fusing imaging biomarkers with clinical and biological attributes.

• In Chassagnon et al. 2021 we develop a deep learning–based method to depict worsening of ILD based on lung shrinkage detection from elastic registration of chest CT scans in patients with systemic sclerosis.