Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Engage Grants Program

École Polytechnique de Montréal

$25,000.00

Versatile training and evaluation frameworks of convolutional neural networks for medical image segmentation

FY 2014-15

Summary
Image segmentation plays a crucial role in many medical imaging applications, such as in oncology orneuroimaging, by automating or facilitating the delineation of anatomical structures, lesions and other regionsof interest. The growing size and number of these medical images have necessitated the use of computationalmethods to facilitate diagnosis, longitudinal assessment of a disease, treatment planning and computer-aidedsurgery. Still, important challenges such as inter and intra-patient appearance variability, multimodal imagingand shape complexity are yet to be fully resolved. Recent advances in machine learning and high performancedata computation have opened a new path to learn from thousands of objects from millions of images using amodel with a large learning capacity. Convolutional neural networks (CNNs) have demonstrated the capacity torecognize various objects within a scene, which outperforms state-of-the-art object recognition methods. Still,the use of CNNs has been very limited for medical image segmentation and are seldom available in commercialsoftware platforms, due to their exhaustive training procedures, which are often very application specific. Thisproject proposes to elliviate this problem by developing versatile training and evaluation platforms for medicalimage segmentation using deep convolutional neural networks, that will be integrated into the ORS (ObjectResearch System) medical imaging software (ORS Visual). This will be geared towards applied problemsencountered in medical imaging, such as liver cancer and cognitive brain diseases. First, a data normalizationprotocol will be developed in partnership with ORS to standardize imaging formats in order to process thesefor training CNNs. Second, a model decision algorithm will be created to select the appropriate architecturewhich will be trained for a specific medical application. Finally, an evaluation platform will be able to processnew test images provided from any given segmentation problem. These innovative platforms will be of greatvalue for a number of applications in medical imaging, by providing the required tools to process largeamounts of complex, high-dimensional medical images for image segmentation problems.
Program
Engage Grants Program
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
476972-2014
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
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
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
Versatile training and evaluation frameworks of convolutional neural networks for medical image segmentation
Keyword
NSERC
Application summary
Image segmentation plays a crucial role in many medical imaging applications, such as in oncology orneuroimaging, by automating or facilitating the delineation of anatomical structures, lesions and other regionsof interest. The growing size and number of these medical images have necessitated the use of computationalmethods to facilitate diagnosis, longitudinal assessment of a disease, treatment planning and computer-aidedsurgery. Still, important challenges such as inter and intra-patient appearance variability, multimodal imagingand shape complexity are yet to be fully resolved. Recent advances in machine learning and high performancedata computation have opened a new path to learn from thousands of objects from millions of images using amodel with a large learning capacity. Convolutional neural networks (CNNs) have demonstrated the capacity torecognize various objects within a scene, which outperforms state-of-the-art object recognition methods. Still,the use of CNNs has been very limited for medical image segmentation and are seldom available in commercialsoftware platforms, due to their exhaustive training procedures, which are often very application specific. Thisproject proposes to elliviate this problem by developing versatile training and evaluation platforms for medicalimage segmentation using deep convolutional neural networks, that will be integrated into the ORS (ObjectResearch System) medical imaging software (ORS Visual). This will be geared towards applied problemsencountered in medical imaging, such as liver cancer and cognitive brain diseases. First, a data normalizationprotocol will be developed in partnership with ORS to standardize imaging formats in order to process thesefor training CNNs. Second, a model decision algorithm will be created to select the appropriate architecturewhich will be trained for a specific medical application. Finally, an evaluation platform will be able to processnew test images provided from any given segmentation problem. These innovative platforms will be of greatvalue for a number of applications in medical imaging, by providing the required tools to process largeamounts of complex, high-dimensional medical images for image segmentation problems.