Grant: Project Grant - PA: Neurosciences, Mental Health and Addiction
University of Calgary
$100,000.00
Prediction of multiple sclerosis disability worsening scores using multi-stream deep learning
EF 2019-20
Résumé
Multiple sclerosis (MS) is a complex neurological disease that disables mostly young Canadians. While therapies are available for many who have a relapsing remitting clinical form (RRMS), >50% of them worsen to a secondary progressive form (SPMS) within 10-15 years of disease onset despite treatment, from which point the disease is hardly manageable. A major current limitation is our inability to identify who worsens rapidly that needs immediate aggressive therapies, and when, as disease course is never the same between RRMS patients. In this project, we will establish a novel deep machine learning method to predict the 5-year risk of disability score worsening in RRMS individuals. This will take a unique approach that simultaneously evaluate patient outcomes from brain pathology, clinical measures, and brain magnetic resonance imaging (MRI). Deep learning algorithms have shown enormous fascinating data recently in image analysis and paralleled human ability in disease prediction. Based on standard brain MRI, we have now achieved an ~85% accuracy using deep learning to separate RRMS from SPMS. Yet deep learning is better using pathology-directed image feature maps than standard MRI, and far superior using both. We have done pioneering work in assessing the pattern (texture) of MRI scans and are the first to show that MRI texture features detect specific elements of brain MS pathology. Now we have made new advancement in deep learning that integrates image texture, brain MRI, and clinical variables including sex and age that are known to impact MS patient outcomes, and built cutting-edge technologies to uncover how deep learning works. Built on these exciting results, this project will provide a novel objective method to identify the high and low risks of sustained worsening or SPMS onset for RRMS individual so that each will receive the best personalized care. This will be crucial for people with RRMS, their families, the health care system, and the Society.
Programme
Project Grant - PA: Neurosciences, Mental Health and Addiction
Bénéficiaire
Emplacement
Alberta, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database
Tel que publié
Id
414149
Name
Zhang Yunyan
Region
Alberta
Country
Canada
Orgtype
Institution Paid; Research Institution
Instname2
Neurosciences, Mental Health and Addiction
Theme2
Biomedical
Prcname2
Medical Physics & Imaging
Projecttitle
Prediction of multiple sclerosis disability worsening scores using multi-stream deep learning
Programtype2
Operating Grants
Orgnameinp2
University of Calgary
Cihramount2
$100,000
Cihrequipment2
$0
Programname2
Project Grant - PA: Neurosciences, Mental Health and Addiction
Keyworddelim
Brain Pathology; Clinical Variables; Deep Machine Learning; Disability Worsening Score; Magnetic Resonance Imaging; Multiple Sclerosis; Personalized Treatment; Relapsing Remitting Multiple Sclerosis; Secondary Progressive Multiple Sclerosis; Texture Analysis
Abstract
Multiple sclerosis (MS) is a complex neurological disease that disables mostly young Canadians. While therapies are available for many who have a relapsing remitting clinical form (RRMS), >50% of them worsen to a secondary progressive form (SPMS) within 10-15 years of disease onset despite treatment, from which point the disease is hardly manageable. A major current limitation is our inability to identify who worsens rapidly that needs immediate aggressive therapies, and when, as disease course is never the same between RRMS patients. In this project, we will establish a novel deep machine learning method to predict the 5-year risk of disability score worsening in RRMS individuals. This will take a unique approach that simultaneously evaluate patient outcomes from brain pathology, clinical measures, and brain magnetic resonance imaging (MRI). Deep learning algorithms have shown enormous fascinating data recently in image analysis and paralleled human ability in disease prediction. Based on standard brain MRI, we have now achieved an ~85% accuracy using deep learning to separate RRMS from SPMS. Yet deep learning is better using pathology-directed image feature maps than standard MRI, and far superior using both. We have done pioneering work in assessing the pattern (texture) of MRI scans and are the first to show that MRI texture features detect specific elements of brain MS pathology. Now we have made new advancement in deep learning that integrates image texture, brain MRI, and clinical variables including sex and age that are known to impact MS patient outcomes, and built cutting-edge technologies to uncover how deep learning works. Built on these exciting results, this project will provide a novel objective method to identify the high and low risks of sustained worsening or SPMS onset for RRMS individual so that each will receive the best personalized care. This will be crucial for people with RRMS, their families, the health care system, and the Society.
Competitiondate
201909
Approvedterm2
1 yr 0 mth
Pinamesdelim
Zhang, Yunyan
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $100,000~ $0~