Yiran Wang, a Ph.D. student of Dr. Guobao Wang and Dr. Simon Cherry, won the IEEE NPSS Christopher J Thompson Student Awards “2nd Best Oral Paper” at the 2021 IEEE Medical Imaging Conference. His paper
Y Wang, E Berg, Y Zuo, E Li, BA Spencer, RD Badawi, SR Cherry, GB Wang. Voxel-wise kinetic model selection using single-subject deep learning for total-body PET parametric imaging. 2021 Virtual IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS&MIC), October 16-23, 2021.
is one of the two Best Oral Papers that were selected from 141 student applicants.
Yiran has been working on total-body PET kinetic modeling and parametric imaging and exploring deep learning approaches in this area. In this work, Yiran developed a single-subject deep learning approach to address the voxel-wise model selection problem for total-body PET parametric imaging on EXPLORER. His approach uses a small fraction of voxels from a subject to build a temporal neural network model and then applies the learned model to fast predict the appropriate kinetic model type of a vast amount of voxels of the same subject. As compared to a standard solution, the deep-learning approach leads to improved total-body PET parametric images with reduced artifacts without a significant increase in computational time.
This work is supported in part by NIH grants R01 CA206187, R01DK124803.