Deep learning-based quality control for tissue motion tracking in 2D-cine MRI-guided radiotherapy

With real-time acquisition of 2D imaging planes, 2D-cine MRI is often used to visualize rapidly moving tumors and organs-at-risk during radiotherapy, and automatic image registration of 2D-cine MRIs at different time points can assist in tracking tissue displacements. However, when sudden large motions occur, automatic registration algorithms can fail to track the radiation target, potentially resulting in sub-optimal therapeutic outcomes and damaging health tissues. Unfortunately, there is still a lack of automatic methods to detect such events. Therefore, the proposed project will establish novel deep learning algorithms to efficiently and robustly detect large tissue motions and the associated causes, as well as failed tissue motion tracking during radiotherapy. The resulting algorithm is expected to allow effective quality control for existing tissue motion tracking systems in radiotherapy for optimized and consistent radiation dose delivery to the treatment target, leading to an improved quality of life for cancer patients.

Faculty Supervisor:

Yiming Xiao;Hassan Rivaz

Student:

Partner:

Elekta

Discipline:

Computer science

Sector:

Health and Related Sciences & Technology; Manufacturing

University:

Concordia University

Program: