Grant: Alliance Grants
École Polytechnique de Montréal
$112,000.00
AI tools for the integration of MRI in planning and motion management phases in cone-beam CT workflows for upper abdominal and lung cancer treatments
EF 2023-24
Résumé
In image-guided radiotherapy, magnetic resonance imaging (MRI) is becoming increasingly important due to the improved soft tissue contrast and physiologic imaging it provides. Current clinical workflows will often display MRI separately, while MRI to CT or cone-beam CT (CBCT) registration solutions provide mostly global alignment, with a need to improve local registration near the tumor region. Furthermore, real-time integration of MRI during radiation therapy in abdominal organs (lungs, liver) is an important challenge for accurate targeted dose delivery, particularly where breathing motion will significantly impact dose delivery and enhancing tumor visualization is important to verify margins. However, anticipating motion and deformations from live navigator images or surrogate signals remains an open problem. The overall goal of this proposal is to improve the integration of MRI in the treatment planning and real-time treatment phases in CBCT workflows. Through an industrial partnership with Varian (a Siemens Healthineers company), we plan to address the unmet needs of improved MRI integration in both the CBCT planning and treatment workflows by exploiting recent advances in deep learning, with synthetic image generation methods to simplify the multimodal deformable registration process, and conditional-based Transformer networks for predicting motion during free-breathing interventions. The multi-modal (MRI/CT/CBCT) registration problem is simplified to a mono-modal registration solution using synthetic images and improve in local deformation field estimation, near the target region of interest. Next, we plan to learn a spatio-temporal representation of organ displacement from respiration in free-breathing sequences (4DCT or 4DMRI), defined as 4D flows described with low-dimensional features to achieve the ability to parameterize an intrinsic model. This will allow to predict future volumes at different respiratory phases, driven by an optical tracking system monitoring the abdominal breathing, with benefits to patients with reduced toxicity. The project will enable the training of highly qualified personnel, creating new scientific developments in AI and computational medical imaging.
Programme
Alliance Grants
Bénéficiaire
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data
Tel que publié
Application id
585700-2023
Name nom
Kadoury, Samuel S
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
2023
Competition year année de concours
2023
Award amount
112000
Program id
ALLRP
Program name en
Alliance Grants
Program name fr
Subventions Alliance
Group en
RESEARCH PARTNERSHIPS
Group fr
PARTENARIATS DE RECHERCHE
Committee code
96
Committee name en
RPP Internal Decision Cttee
Committee name fr
Cté de décision interne - PPR
Area of application code
518
Area of application group en
Manufacturing processes and products
Area of application group fr
Procédés et produits industriels
Area of application en
Medical equipment and apparatus
Area of application fr
Appareils et instruments médicaux
Research subject code
2603
Research subject group en
Robotics
Research subject group fr
Computer vision
Research subject en
Robotique
Research subject fr
Vision artificielle
Application title
AI tools for the integration of MRI in planning and motion management phases in cone-beam CT workflows for upper abdominal and lung cancer treatments
Keyword
Deep learning; Deformable registration; Image synthesis; Medical imaging; Motion modeling; Predictive modelling; Radiotherapy; Tumor tracking
Application summary
In image-guided radiotherapy, magnetic resonance imaging (MRI) is becoming increasingly important due to the improved soft tissue contrast and physiologic imaging it provides. Current clinical workflows will often display MRI separately, while MRI to CT or cone-beam CT (CBCT) registration solutions provide mostly global alignment, with a need to improve local registration near the tumor region. Furthermore, real-time integration of MRI during radiation therapy in abdominal organs (lungs, liver) is an important challenge for accurate targeted dose delivery, particularly where breathing motion will significantly impact dose delivery and enhancing tumor visualization is important to verify margins. However, anticipating motion and deformations from live navigator images or surrogate signals remains an open problem. The overall goal of this proposal is to improve the integration of MRI in the treatment planning and real-time treatment phases in CBCT workflows. Through an industrial partnership with Varian (a Siemens Healthineers company), we plan to address the unmet needs of improved MRI integration in both the CBCT planning and treatment workflows by exploiting recent advances in deep learning, with synthetic image generation methods to simplify the multimodal deformable registration process, and conditional-based Transformer networks for predicting motion during free-breathing interventions. The multi-modal (MRI/CT/CBCT) registration problem is simplified to a mono-modal registration solution using synthetic images and improve in local deformation field estimation, near the target region of interest. Next, we plan to learn a spatio-temporal representation of organ displacement from respiration in free-breathing sequences (4DCT or 4DMRI), defined as 4D flows described with low-dimensional features to achieve the ability to parameterize an intrinsic model. This will allow to predict future volumes at different respiratory phases, driven by an optical tracking system monitoring the abdominal breathing, with benefits to patients with reduced toxicity. The project will enable the training of highly qualified personnel, creating new scientific developments in AI and computational medical imaging.