Medical Physics Seminar – Monday, September 27, 2021
Interior tomographic reconstruction via direct backprojection and deep learning.
Chengzhu Zhang
Medical Physics Graduate Student (Advisor: Dr. Chen)
In this work, a new pathway is presented to address the interior tomographic reconstruction problem. The new scheme consists of two steps: 1) Direct backprojecting the acquired fully-truncated divergent beam projection data to form a backprojection image B0 without the conventional differentiation operations and 2) The true image is then reconstructed from the blurred image B0 using a trained deep neural network architecture. Experimental results demonstrated that the trained network can reconstruct small ROIs at arbitrary locations
Visualization and quantification of immune-related adverse events by 18F-FDG PET/CT in immune-checkpoint blockade-treated metastatic melanoma
Daniel Huff
Medical Physics Graduate Student (Advisor: Dr. Jeraj)
Immune-checkpoint inhibitors (ICI) have greatly improved outcomes for metastatic melanoma patients over the past decade. While ICI patients regularly receive 18F-FDG PET/CT scans to monitor their disease, information about inflammatory side effects caused by ICI, called immune-related adverse events (irAE), can also be gained from analyzing their PET scans. Here, we present a retrospective study of clinical 18F-FDG PET/CT images which makes use of automated organ segmentation tools to quantify 18F-FDG uptake as an indicator of irAE in the thyroid, lung, and bowel. We find that patients who experience irAE associated with these organs demonstrate elevated 18F-FDG uptake within the affected organ as compared to patients without irAE.