Fiche de BOTTANI Simona

Titre : Machine learning for neuroimaging using a very large scale clinical datawarehouse
Date : 1 avril 2021
Heure: 14:00
Résumé : Deep learning (DL) models have been widely used for the computer-aided diagnosis (CAD) of neurodegenerative diseases but they have been mostly validated using research data sets. The validation on large clinical data sets is necessary to understand the performance of these tools in a real setting. This PhD work consisted in applying DL algorithms to data from the clinical data warehouse (CDW) of the AP-HP to validate CAD of neurodegenerative diseases. We developed, thanks to the manual annotation, an automatic approach for the quality control of T1-brain MRI from the CDW. Then we proposed to homogenize such large clinical data set by converting images acquired after the injection of gadolinium into non-contrast-enhanced images using DL models. Lastly, we assessed whether DL algorithms could detect dementia in a CDW using T1w brain MRI. We compared the ability of DL algorithms to detect dementia patients in a research data set and in the AP-HP CDW set.
Lieu : ISC - Institut des Systèmes Complexes, Rue Nationale, Paris