Notice and Invitation
Oral Presentation of Dissertation Proposal
Department of Bioengineering
Volgenau School of Engineering, George Mason University

Ko-Tsung Hsu

Bachelor of Science in Biotechnology, Ming Chuan University, 2015
Master of Science in Bioinformatics and Computational Biology, George Mason University, 2019
Image Quality Enhancement of Photoacoustic and Ultrasonic Imaging with Data-Driven Methods


Tuesday, August 3, 2021, 3:00-4:30 pm
Via Zoom<https://gmu.zoom.us/j/93109807019?pwd=eDNVQ3FyUW5ZMDJkcWFONDE3aTdkZz09>
Meeting ID: 931 0980 7019
Passcode: 756268

All are invited to attend.

Committee
Dr. Parag Chitnis, Dissertation Supervisor
Dr. Qi Wei, Committee Chair
Dr. Siddhartha Sikdar
Dr. Vadim Sokolov

Abstract:
The medical imaging community has recently extensively investigated the benefits of using deep learning for many applications in different imaging modalities. For instance, a low-dose CT protocol that delivers fewer X-ray photons can be employed to avoid the patient at risk of exposure to excessive radiation. However, the resulting images acquired using the low-dose CT protocol are affected by the low signal-to-noise ratio and server streaking artifacts, which cannot be adjusted by routine clinically used algorithms such as filter back projection. This similar scenario occurs in photoacoustic tomography, where the target of interest must be imaged under sparse sensing strategies or geometric constraints, resulting in inferior image quality due to insufficient sampling. In ultrafast ultrasound imaging, high frame rate ultrasound modes come at the cost of lower contrast and reduced lateral resolution due to the nature of the unfocused transmission. To realize the aforementioned medical imaging applications and acquisitions in practice, the image quality needs to be sacrificed inevitably if conventional methods are exploited for reconstruction. For this reason, a much more advanced algorithm is necessary to develop against these scenarios. Here, we will apply data-driven methods to address the intractable scenarios specifically in photoacoustic tomography and ultrafast ultrasound imaging.



Carol McHugh
Academic Program Assistant
Department of Bioengineering
3100 Peterson Family Health Services Hall
703-993-5846