Canada Spends Logo

Explorez notre analyse du budget 2026

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

University of Ottawa

$27,000.00

Improving Equity in Medical AI Through Clustering and Data Preprocessing for Brain Tumor Detection

EF 2025-26

Résumé
Early diagnosis in oncology depends on reliable AI systems, yet the development of neural networks for medical imaging remains limited within Canadian research (Darcel et al., 2023). High-quality, diverse datasets are essential for training accurate models, but medical data are often incomplete or unbalanced, which reduces generalizability and can introduce bias (Liu et al., 2023). These limitations may cause models to learn patterns based on data gaps rather than true clinical features, leading to uneven performance across patient groups (Palaniappan, Lin, Vogel, & Lim, 2024). Preprocessing can address the limitations of incomplete or unbalanced datasets by clustering relevant medical information into more representative groups. Therefore, the project examines how clustering can strengthen data quality before model training by improving feature organization, representation, and bias mitigation (Liu et al., 2023). Achieving this requires datasets with sufficient volume, demographic variation, and clinically meaningful variables, which will be provided by the ReMIND dataset. To evaluate the best preprocessing strategy, the study will compare k-means, DBSCAN, and agglomerative clustering, three techniques that capture different structures in complex medical data (Morissette & Chartier, 2013; Ester, Kriegel, Sander, & Xu, 1996; Oti & Olusola, 2024). By improving how medical features are clustered during preprocessing, this project aims to support the development of equitable neural network models for brain tumor detection and contribute to future synthetic data generation that can strengthen small or incomplete datasets.
Programme
Master's Award: Canada Graduate Research Scholarships - Master's program
Bénéficiaire
Emplacement
Ontario, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database

Tel que publié

Id
572719
Name
Espinoza Giovanni
Region
Ontario
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Improving Equity in Medical AI Through Clustering and Data Preprocessing for Brain Tumor Detection
Programtype2
Studentship Programs
Orgnameinp2
University of Ottawa
Cihramount2
$27,000
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Research Scholarships - Master's program
Keyworddelim
Artificial Intelligence; Bias Mitigation; Brain Tumor Detection; Clinical Decision Support; Clustering Techniques; Data Preprocessing; Health Equity; Machine Learning; Medical Imaging Data; Neural Networks
Abstract
Early diagnosis in oncology depends on reliable AI systems, yet the development of neural networks for medical imaging remains limited within Canadian research (Darcel et al., 2023). High-quality, diverse datasets are essential for training accurate models, but medical data are often incomplete or unbalanced, which reduces generalizability and can introduce bias (Liu et al., 2023). These limitations may cause models to learn patterns based on data gaps rather than true clinical features, leading to uneven performance across patient groups (Palaniappan, Lin, Vogel, & Lim, 2024). Preprocessing can address the limitations of incomplete or unbalanced datasets by clustering relevant medical information into more representative groups. Therefore, the project examines how clustering can strengthen data quality before model training by improving feature organization, representation, and bias mitigation (Liu et al., 2023). Achieving this requires datasets with sufficient volume, demographic variation, and clinically meaningful variables, which will be provided by the ReMIND dataset. To evaluate the best preprocessing strategy, the study will compare k-means, DBSCAN, and agglomerative clustering, three techniques that capture different structures in complex medical data (Morissette & Chartier, 2013; Ester, Kriegel, Sander, & Xu, 1996; Oti & Olusola, 2024). By improving how medical features are clustered during preprocessing, this project aims to support the development of equitable neural network models for brain tumor detection and contribute to future synthetic data generation that can strengthen small or incomplete datasets.
Competitiondate
202512
Approvedterm2
1 yr 0 mth
Pinamesdelim
Espinoza, Giovanni
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $27,000~ $0~