Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

The University of British Columbia

$83,000.00

Deployable Machine Learning with Large and Multi-scale Ultrasound Data for Interpretable Decision Making

FY 2024-25

Summary
The main goal of this research program is to push the boundaries of existing state-of-the-art machine learning solutions for ultrasound imaging and develop a generalizable learning ecosystem from large-scale data. In recent years, ultrasound imaging has enjoyed exponential growth in user numbers due to advances in miniaturization and the development of portable point-of-care ultrasound systems. However, addressing the challenge of limited supply of highly trained ultrasound operators is crucial to expand access to underprivileged and rural areas. Machine learning integration with ultrasound scanners has shown significant promise to facilitate data acquisition and interpretation. However, existing machine learning-based approaches are limited in scope as they are based on small and highly curated labeled ultrasound data. This makes their generalization to ultrasound operators with less experience and to under-served populations with unknown data distribution uncertain. Scaling up the approach to larger datasets is particularly hindered by limited access to experts for data curation and labeling. This program will address the following fundamental questions: 1.How can a generalizable machine learning framework be developed using large datasets with a high prevalence of anomalous and heterogeneous samples, and how can it be safely deployed in the presence of out-of-distribution data given variability in ultrasound operator experience? This will involve developing ultrasound-specific machine learning advances in self-supervised learning to model data distribution, detect out-of-distribution data samples, and, in combination with supervised learning, identify samples with noisy labels. Next, semi-supervised learning will be used to generate weak labels for noisy data to improve learning from such data. Finally, new methods for automatic uncertainty estimation will be developed to improve robustness and trust in the output of machine learning models in the presence of out-of-distribution data. 2.How can an explainable machine learning framework be designed to facilitate user trust and interpretation of ultrasound data? This will involve developing explainable machine learning systems that account for key properties of ultrasound imaging, including its inherently dynamic nature and prevalence of anomalous data. Moreover, physics-based simulations from microscopic tissue samples will be used in tandem with machine learning to integrate data from various ultrasound systems across different spatial resolutions, and to improve ultrasound interpretation through enhanced tissue characterization. This program will train 12 HQP (highly qualified personnel), including 4 PhD students, 3 MSc students, and 5 undergraduates, in ultrasound imaging and machine learning. The skills developed will be transferable to other disciplines in engineering and computer science that emphasize risk mitigation in the design of automation systems for critical decision making.
Program
Discovery Grants Program - Individual
Location
British Columbia, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
RGPIN-2024-03930
Name nom
Abolmaesumi, Purang
Department département
Electrical and Computer Engineering
Organization id
2
Institution établissement
The University of British Columbia
Province en
British Columbia
Province fr
Colombie-Britannique
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
83000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1510
Committee name en
Electrical and Computer Engineering
Committee name fr
Génie électrique et informatique
Area of application code
1102
Area of application group en
Health, education and social services
Area of application group fr
Santé publique, éducation et services sociaux
Area of application en
Biomedical engineering
Area of application fr
Génie biomédical
Research subject code
2510
Research subject group en
Electrical and electronic engineering
Research subject group fr
Génie électrique et électronique
Research subject en
Adaptive, learning and evolutionary systems
Research subject fr
Systèmes d'apprentissage adaptatifs et évolutifs
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Deployable Machine Learning with Large and Multi-scale Ultrasound Data for Interpretable Decision Making
Keywords
medical imaging; machine learning; anomalous data; data integration; biomedical engineering; interpretable systems; explainable systems; trustworthy systems; physics-based simulation; large-scale and multi-scale data
Application summary
The main goal of this research program is to push the boundaries of existing state-of-the-art machine learning solutions for ultrasound imaging and develop a generalizable learning ecosystem from large-scale data. In recent years, ultrasound imaging has enjoyed exponential growth in user numbers due to advances in miniaturization and the development of portable point-of-care ultrasound systems. However, addressing the challenge of limited supply of highly trained ultrasound operators is crucial to expand access to underprivileged and rural areas. Machine learning integration with ultrasound scanners has shown significant promise to facilitate data acquisition and interpretation. However, existing machine learning-based approaches are limited in scope as they are based on small and highly curated labeled ultrasound data. This makes their generalization to ultrasound operators with less experience and to under-served populations with unknown data distribution uncertain. Scaling up the approach to larger datasets is particularly hindered by limited access to experts for data curation and labeling. This program will address the following fundamental questions: 1.How can a generalizable machine learning framework be developed using large datasets with a high prevalence of anomalous and heterogeneous samples, and how can it be safely deployed in the presence of out-of-distribution data given variability in ultrasound operator experience? This will involve developing ultrasound-specific machine learning advances in self-supervised learning to model data distribution, detect out-of-distribution data samples, and, in combination with supervised learning, identify samples with noisy labels. Next, semi-supervised learning will be used to generate weak labels for noisy data to improve learning from such data. Finally, new methods for automatic uncertainty estimation will be developed to improve robustness and trust in the output of machine learning models in the presence of out-of-distribution data. 2.How can an explainable machine learning framework be designed to facilitate user trust and interpretation of ultrasound data? This will involve developing explainable machine learning systems that account for key properties of ultrasound imaging, including its inherently dynamic nature and prevalence of anomalous data. Moreover, physics-based simulations from microscopic tissue samples will be used in tandem with machine learning to integrate data from various ultrasound systems across different spatial resolutions, and to improve ultrasound interpretation through enhanced tissue characterization. This program will train 12 HQP (highly qualified personnel), including 4 PhD students, 3 MSc students, and 5 undergraduates, in ultrasound imaging and machine learning. The skills developed will be transferable to other disciplines in engineering and computer science that emphasize risk mitigation in the design of automation systems for critical decision making.
The University of British Columbia - $83,000 Grant (2… | Canada Spends