Metal artifacts reduction in Computed Tomography (CT)
Qihui Lyu, Professor
UC San Francisco
Applications for Fall 2024 are closed for this project.
This project aims to improve metal artifacts in Computed Tomography (CT) images. In the presence of highly attenuating objects such as dental fillings, spinal screws/rods, hip prostheses, and gold fiducial markers, CT images are often corrupted by streak artifacts, making these images non-diagnostic and impacting the accuracy of radiotherapy treatment. This project uses deep learning methods, in combination with CT iterative reconstruction algorithms, to improve the image artifacts in the presence of metals.
Role: The student will be provide with a training dataset with 14,000 cases for training, validation, and algorithm development. The student will work directly with the faculty mentor, and will be trained on literature search, research problem formulation, coding, computational algorithms, and scientific writing. In particular, relevant to this project, the student will learn deep learning methods, CT image reconstruction algorithms, and image processing algorithms.
Qualifications: Applicants should have strong background in mathematical modeling, physics, and/or Computer Vision. Prior experience in mathematical modeling and coding is desirable but not essential. The student is expected to meet with the faculty mentor weekly.
Hours: 12 or more hours
Off-Campus Research Site: The student can choose to work remotely or work at the UCSF mission bay campus.
Related website: http://lyulab.ucsf.edu/