Grant: Master's Award: Canada Graduate Research Scholarships - Master's program
University of British Columbia
$27,000.00
Deep Learning Prediction of Positive Breast Biopsy Outcomes Using BC Cancer Data
EF 2025-26
Résumé
Breast screening programs face a major challenge due to the high number of biopsies that yield negative results creating unnecessary patient anxiety, and using significant healthcare resources. This project aims to develop and evaluate multi-modal deep learning (DL) models to predict positive breast biopsy outcomes using mammographic, clinical, and genomic data collected by the BC Cancer Breast Screening Program. Previous approaches focus on long-term outcomes and recurrence rather than positive biopsies. Classical statistical models have shown limited performance, and few have integrated genomic and clinical data. This study will apply convolutional neural networks for mammography images and additional branches for clinical and genomic data, combing them through feature-level fusion, and using fully connected layers to estimate the probability of a positive biopsy. Key predictive features will be identified and interpreted, and clinical impact will be evaluated in terms of reducing unnecessary biopsies while maintaining high sensitivity for cancer detection. Imaging data can vary across different populations due to imaging protocols or clinical workflows, so applying this study specific to BC is valuable and can support generalizing the model to other populations. By applying a BC-specific approach, this research addresses a real clinical challenge, aiming to reduce patient anxiety, improve resource allocation, and support more personalized and effective breast screenings. The research question asks: Can a BC-specific multi-modal DL model identify features which predict positive breast biopsies, allowing for fewer unnecessary biopsies while maintaining clinically required sensitivity?
Programme
Master's Award: Canada Graduate Research Scholarships - Master's program
Bénéficiaire
Emplacement
British Columbia, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database
Tel que publié
Id
573155
Name
Ali Muskaan D
Region
British Columbia
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
Deep Learning Prediction of Positive Breast Biopsy Outcomes Using BC Cancer Data
Programtype2
Studentship Programs
Orgnameinp2
University of British Columbia
Cihramount2
$27,000
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Research Scholarships - Master's program
Keyworddelim
Bc Cancer; Breast Cancer; Clinical Data; Convolutional Neural Network (Cnn); Deep Learning; Genomic Data; Mammography; Multi-Modal Neural Network; Positive Biopsy Prediction; Predictive Modelling
Abstract
Breast screening programs face a major challenge due to the high number of biopsies that yield negative results creating unnecessary patient anxiety, and using significant healthcare resources. This project aims to develop and evaluate multi-modal deep learning (DL) models to predict positive breast biopsy outcomes using mammographic, clinical, and genomic data collected by the BC Cancer Breast Screening Program. Previous approaches focus on long-term outcomes and recurrence rather than positive biopsies. Classical statistical models have shown limited performance, and few have integrated genomic and clinical data. This study will apply convolutional neural networks for mammography images and additional branches for clinical and genomic data, combing them through feature-level fusion, and using fully connected layers to estimate the probability of a positive biopsy. Key predictive features will be identified and interpreted, and clinical impact will be evaluated in terms of reducing unnecessary biopsies while maintaining high sensitivity for cancer detection. Imaging data can vary across different populations due to imaging protocols or clinical workflows, so applying this study specific to BC is valuable and can support generalizing the model to other populations. By applying a BC-specific approach, this research addresses a real clinical challenge, aiming to reduce patient anxiety, improve resource allocation, and support more personalized and effective breast screenings. The research question asks: Can a BC-specific multi-modal DL model identify features which predict positive breast biopsies, allowing for fewer unnecessary biopsies while maintaining clinically required sensitivity?
Competitiondate
202512
Approvedterm2
1 yr 0 mth
Pinamesdelim
Ali, Muskaan D
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $27,000~ $0~