Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Engage Grants Program

École Polytechnique de Montréal

$25,000.00

Detection of liver metastases with deep convolutional neural networks

EF 2014-15

Résumé
An important area of application of computer vision is medical image analysis, which often consists processing3D imaging modalities, such as computed tomography (CT) and magnetic resonance imaging (MRI). Standardproblems in this area include tissue classification, lesion detection and organ segmentation, with the ultimategoal of assisting diagnosis and improving clinical outcomes. An application of medical image analysis iscomputer aided diagnosis (CADx) of cancer tumors, a condition that will affect almost half of us during ourlifetime. Recent work in this field has applied standard machine learning algorithms, such as Support VectorMachines (SVMs) and Conditional Random Fields (CRFs) to those problems, using standard features such asHistogram of Oriented Gradients (HOG). However, these features are often tuned to a specific organ type andmodality, requiring significant expertise and severely limiting the scope of a given model. On the other hand,the fundamental motivation for deep learning approaches is that, given enough data, they are capable ofautomatically learning these features, reaching a level of accuracy surpassing previous approaches. In thisproject, we propose to generalize recent work in brain tumor segmentation competition with a model relying ondeep learning to detect metastatic liver tumors of different etiologies and modalities. A challenge with deeplearning architectures is that they require significant amounts of data to train good-performing models, achallenge compounded in medical imaging applications due to the scarcity of subjects that may exhibit tumortypes of interest. Using readily-available data of sufficient quantity from the liver through a private datasethand-segmented by medical collaborators to this project, we plan to develop a framework that will detect tumorlocations in a liver segmented from CT images. This approach opens new prospects for medical imagingapplications, by proposing of models that exhibit high accuracy with minimal human intervention, broadeningtheir use and ultimately leading to improved patient care.
Programme
Engage Grants Program
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
477403-2014
Name nom
Kadoury, Samuel
Department département
Génie informatique et génie logiciel
Organization id
64
Institution établissement
École Polytechnique de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2014
Competition year année de concours
2014
Award amount
25000
Program id
EGP
Program name en
Engage Grants Program
Program name fr
Programme de subventions d'engagement partenariat
Group en
Research partnerships programs
Group fr
Programmes de partenariat
Committee code
1552
Committee name en
Quebec Internal Decision Committee
Committee name fr
Comité de décision interne pour le Québec
Area of application code
1211
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Medical and health sciences
Area of application fr
Sciences médicales et de la santé
Research subject code
2800
Research subject group en
Artificial intelligence
Research subject group fr
Intelligence artificielle
Research subject en
Artificial intelligence (computer vision, use2603)
Research subject fr
Intelligence artificielle (vision artificielle, utiliser 2603)
Partie
1
Nb partie
1
Application title
Detection of liver metastases with deep convolutional neural networks
Keyword
NSERC
Application summary
An important area of application of computer vision is medical image analysis, which often consists processing3D imaging modalities, such as computed tomography (CT) and magnetic resonance imaging (MRI). Standardproblems in this area include tissue classification, lesion detection and organ segmentation, with the ultimategoal of assisting diagnosis and improving clinical outcomes. An application of medical image analysis iscomputer aided diagnosis (CADx) of cancer tumors, a condition that will affect almost half of us during ourlifetime. Recent work in this field has applied standard machine learning algorithms, such as Support VectorMachines (SVMs) and Conditional Random Fields (CRFs) to those problems, using standard features such asHistogram of Oriented Gradients (HOG). However, these features are often tuned to a specific organ type andmodality, requiring significant expertise and severely limiting the scope of a given model. On the other hand,the fundamental motivation for deep learning approaches is that, given enough data, they are capable ofautomatically learning these features, reaching a level of accuracy surpassing previous approaches. In thisproject, we propose to generalize recent work in brain tumor segmentation competition with a model relying ondeep learning to detect metastatic liver tumors of different etiologies and modalities. A challenge with deeplearning architectures is that they require significant amounts of data to train good-performing models, achallenge compounded in medical imaging applications due to the scarcity of subjects that may exhibit tumortypes of interest. Using readily-available data of sufficient quantity from the liver through a private datasethand-segmented by medical collaborators to this project, we plan to develop a framework that will detect tumorlocations in a liver segmented from CT images. This approach opens new prospects for medical imagingapplications, by proposing of models that exhibit high accuracy with minimal human intervention, broadeningtheir use and ultimately leading to improved patient care.