Combining MRI markers with sleep EEG features to track brain health and detect pathology
Description: Age has been considered for a long time one of the dominant driving factors of disease progression in multiple sclerosis (MS). Independently of disease duration, older age raises the risk of MS progression and disability accumulation. At the same time, conventional magnetic resonance imaging (MRI) biomarkers, including white matter lesion volume, weakly correlate with disease activity. Recently, advanced MS lesional biomarkers, such as cortical and paramagnetic rim lesions, have emerged as playing an important role in MS progression, but also do not adequately explain disability accumulation. Brain age estimation uses machine learning (ML) to predict the individual age from MRI and it is a potential marker for brain health and integrity. A combination of brain age and advanced lesional biomarkers could provide important insights into disease progression and disability accumulation in MS patients. This may be particularly helpful in expanding our understanding of the causes of cognitive deficits in MS, which are highly prevalent but poorly understood.In this project, I will retrospectively analyze two unique datasets that include high-resolution MRI scans and cognitive scores of 165 MS patients (with 7 tesla (7T) and 3T scans) and ~1500 MS patients (with 3T scans), respectively, over two to three time-points. I propose three complementary work packages (WP) that, by leveraging structural MRI and novel ML approaches, aim at detecting relevant MS prognostic biomarkers and estimating disease progression. Brain age will be computed per each patient at baseline and at successive time-points. In addition, relevant MS lesional biomarkers will be detected semi-automatically with deep learning. The association between these findings and disability measures will be quantified analyzing cross-sectional data. Finally, I will develop a prediction model that, combining these potential prognostic biomarkers, estimates disease progression on longitudinal data.This project has strong potential for real-world impact. Identifying MRI biomarkers relevant to brain health would enhance our general knowledge of brain aging. Furthermore, an automated machine learning method able to accurately predict multiple sclerosis progression would enable clinicians to provide a more accurate prognosis and might be useful as a modifiable outcome measure in clinical trials.
Amount: CHF 122,300
Funding type: fellowship
Funder scheme: Postdoc.Mobility
Start date: 2022-06-01
End date: 2024-05-31
Investigator: La Rosa, Francesco
Institution: Institution abroad - United States of America
Subfield: Pathology and Forensic Medicine
Field: Medicine
Domain: Health Sciences
Funded outputs count: 9
Funder award ID: 206833
OpenAlex ID: G4720912349