Grant: Master's Award: Canada Graduate Scholarships
University of British Columbia
$17,500.00
"Utilization of Machine Learning, Deep Learning, and Radiomics for the classification of sub-cm lung nodules in early cancer diagnosis"
FY 2019-20
Summary
"Low Dose Computed Tomography (LDCT) screening programs are well known for detecting small lung nodules in at risk lung cancer patients. Predicting the future behavior of small nodules has proven challenging even for expert clinicians. As such, nodules in this category often have a 'wait and see' plan assigned to them. Computer assisted diagnosis (CADx) tools are in high demand as a cost-effective diagnostic approach for predicting the malignant potential of these small nodules. The proposed CADx approach in classifying these small nodules is founded on Artificial Intelligence (AI) methods such as machine learning (ML) and deep learning (DL) algorithms. By providing these algorithms with a large dataset of nodules and hand-generated radiomic features, these data-driven models can determine which features are both statistically relevant and represent malignant behaviors. Moreover, most current AI methods are largely seen as 'blackboxes' to expert clinicians, and thus are not well trusted in the radiologist medical community as of yet. I propose developing further methods to surmount this hurdle of interpretability, while maintaining a high classification accuracy, for clinical deployment as another tool in the radiologist's toolbox. I will perform this task over the next 12 months. First, I will collect an assortment of public LDCT lung nodules. Second, I will monitor, implement, train, and study of state-of-the-art 2D and 3D ML and DL models, then validate these methods on an in-house lung nodule dataset. This CADx tool could effectively remove the 'wait and see' policy, address the cost-effective challenge of LDCT screening programs, stage shift cancerous nodules into earlier more treatable stages, eliminate extra burden on hospital resources, and save patients' lives."
Department
Program
Master's Award: Canada Graduate Scholarships
Recipient
Location
British Columbia, Canada
Recipient Type
Academia
Type
Grant
Source
CIHR Funding Decisions Database
As published
Id
428870
Name
Janzen Ian K
Region
British Columbia
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
"Utilization of Machine Learning, Deep Learning, and Radiomics for the classification of sub-cm lung nodules in early cancer diagnosis"
Programtype2
Studentship Programs
Orgnameinp2
University of British Columbia
Cihramount2
$17,500
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Scholarships
Keyworddelim
Cancer; Deep Learning; Early Detection; Lung Cts; Machine Learning; Radiomics
Abstract
"Low Dose Computed Tomography (LDCT) screening programs are well known for detecting small lung nodules in at risk lung cancer patients. Predicting the future behavior of small nodules has proven challenging even for expert clinicians. As such, nodules in this category often have a 'wait and see' plan assigned to them. Computer assisted diagnosis (CADx) tools are in high demand as a cost-effective diagnostic approach for predicting the malignant potential of these small nodules. The proposed CADx approach in classifying these small nodules is founded on Artificial Intelligence (AI) methods such as machine learning (ML) and deep learning (DL) algorithms. By providing these algorithms with a large dataset of nodules and hand-generated radiomic features, these data-driven models can determine which features are both statistically relevant and represent malignant behaviors. Moreover, most current AI methods are largely seen as 'blackboxes' to expert clinicians, and thus are not well trusted in the radiologist medical community as of yet. I propose developing further methods to surmount this hurdle of interpretability, while maintaining a high classification accuracy, for clinical deployment as another tool in the radiologist's toolbox. I will perform this task over the next 12 months. First, I will collect an assortment of public LDCT lung nodules. Second, I will monitor, implement, train, and study of state-of-the-art 2D and 3D ML and DL models, then validate these methods on an in-house lung nodule dataset. This CADx tool could effectively remove the 'wait and see' policy, address the cost-effective challenge of LDCT screening programs, stage shift cancerous nodules into earlier more treatable stages, eliminate extra burden on hospital resources, and save patients' lives."
Competitiondate
201912
Approvedterm2
1 yr 0 mth
Pinamesdelim
Janzen, Ian K
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $17,500~ $0~