Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Master's Award: Canada Graduate Research Scholarships - Master's program

McGill University

$27,000.00

Federated Machine Learning for Lung Cancer Auto-Segmentation

EF 2025-26

Résumé
Lung cancer remains the leading cause of cancer-related mortality worldwide, and timely treatment can significantly improve patient prognosis. A source of treatment delay is the identification and contouring of tumours in 4DCT scans, an often lengthy process, especially for smalle centres with limited staff and resources. This project aims to expedite this segmentation process by developping, validating, and potentially implementing an artificial intelligence model for lung cancer auto-segmentation using 4DCT scan data. Using the nnU-Net framework, the model can adapt to new medical imaging datasets, improving robustness even when faced with the challenge of tumour heterogeneity. Much of this has already been done in some way, however this project aims to improve model generalizability by implementing a federated learning system that enables multiple Quebec healthcare institutions to collaboratvely train a shared model all while maintaining patient data privacy. This approach will not only provide a performative model, but also allow smaller centres in the province to have access to advanced AI tools for which they may lack sufficient data to train in-house. Model performance will be evaluated quantitatively using standard segmentation metric and qualitatively using oncologist review. We will also assess workflow effects by measuring reductions in contouring time. If the model proves sufficiently performative, it can be integrated into the clinical pipeline as a proof-of-concept for safe and effective use of federated AI tools in the Quebec healthcare system. This research seeks to reduce treatment delays by tackling inefficiencies and contour accuracy during the segmentation phase of the clinical pipeline at centres across Quebec.
Programme
Master's Award: Canada Graduate Research Scholarships - Master's program
Bénéficiaire
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database

Tel que publié

Id
573248
Name
Coull-Neveu Ben
Region
Québec
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Federated Machine Learning for Lung Cancer Auto-Segmentation
Programtype2
Studentship Programs
Orgnameinp2
McGill University
Cihramount2
$27,000
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Research Scholarships - Master's program
Keyworddelim
Artificial Intelligence; Auto-Segmentation; Lung Cancer; Nnu-Net
Abstract
Lung cancer remains the leading cause of cancer-related mortality worldwide, and timely treatment can significantly improve patient prognosis. A source of treatment delay is the identification and contouring of tumours in 4DCT scans, an often lengthy process, especially for smalle centres with limited staff and resources. This project aims to expedite this segmentation process by developping, validating, and potentially implementing an artificial intelligence model for lung cancer auto-segmentation using 4DCT scan data. Using the nnU-Net framework, the model can adapt to new medical imaging datasets, improving robustness even when faced with the challenge of tumour heterogeneity. Much of this has already been done in some way, however this project aims to improve model generalizability by implementing a federated learning system that enables multiple Quebec healthcare institutions to collaboratvely train a shared model all while maintaining patient data privacy. This approach will not only provide a performative model, but also allow smaller centres in the province to have access to advanced AI tools for which they may lack sufficient data to train in-house. Model performance will be evaluated quantitatively using standard segmentation metric and qualitatively using oncologist review. We will also assess workflow effects by measuring reductions in contouring time. If the model proves sufficiently performative, it can be integrated into the clinical pipeline as a proof-of-concept for safe and effective use of federated AI tools in the Quebec healthcare system. This research seeks to reduce treatment delays by tackling inefficiencies and contour accuracy during the segmentation phase of the clinical pipeline at centres across Quebec.
Competitiondate
202512
Approvedterm2
1 yr 0 mth
Pinamesdelim
Coull-Neveu, Ben
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $27,000~ $0~