TY - GEN
T1 - Full 4π emission data collection and reconstruction for small animal PET imaging
AU - Zhang, Yuxuan
AU - Baghaei, Hossain
AU - Li, Hongdi
AU - Liu, Shitao
AU - Ramirez, Rocio
AU - An, Shaohui
AU - Wang, Chao
AU - Wong, Wai Hoi
N1 - Copyright:
Copyright 2020 Elsevier B.V., All rights reserved.
PY - 2011
Y1 - 2011
N2 - Most of the current animal PET detector systems are cylindrical or similar multi-sided polygonal geometries with limited axial field of view. The object is placed near the center of the detector ring during the emission data collection. The signal that can be detected has a limited angular range; the paraxial signal cannot be detected. Moreover, the sensitivity for the objects positioned at different locations inside the FOV is different. The central part of the FOV has higher sensitivity than that near the end along the axis. The lack of paraxial detectability which means non-uniform sampling, along with the non-uniform sensitivity of a PET system, will affect the uniformity of the overall image quality. Also, currently widely used PET data processing and reconstruction algorithms are sinogram based, which usually uses different sizes for the voxel in axial and transaxial directions. Because of these non-uniformities, the resolution and quality of current PET image are anisotropic. In order to achieve isotropic results, the emissions from the object must be collected in full 4π space and reconstructed accordingly. Here we propose a full 4π emission data collection method, which involving the rotation of the object during the emission collection in a plane parallel to the detector ring axis. The full 4π emission data are then reconstructed and processed to generate the 3D image set with cubic voxel, uniform resolution and signal-to-noise ratio in all directions and locations. Both Monte Carlo simulations and experiments are carried out with simulated and real mouse phantoms. Emission data in both 4π and conventional modes are collected and then processed. Point source is used in the experiment as the fiducial mark. Our results show that with full 4π collection method, the image qualities are substantially improved in several aspects such as the axial distortions and the uniformity of the SNR. Moreover, the axial strip-like artifacts in the conventional mode are canceled in the full 4π mode, therefore less smooth window is needed during the reconstruction which led to higher resolution.
AB - Most of the current animal PET detector systems are cylindrical or similar multi-sided polygonal geometries with limited axial field of view. The object is placed near the center of the detector ring during the emission data collection. The signal that can be detected has a limited angular range; the paraxial signal cannot be detected. Moreover, the sensitivity for the objects positioned at different locations inside the FOV is different. The central part of the FOV has higher sensitivity than that near the end along the axis. The lack of paraxial detectability which means non-uniform sampling, along with the non-uniform sensitivity of a PET system, will affect the uniformity of the overall image quality. Also, currently widely used PET data processing and reconstruction algorithms are sinogram based, which usually uses different sizes for the voxel in axial and transaxial directions. Because of these non-uniformities, the resolution and quality of current PET image are anisotropic. In order to achieve isotropic results, the emissions from the object must be collected in full 4π space and reconstructed accordingly. Here we propose a full 4π emission data collection method, which involving the rotation of the object during the emission collection in a plane parallel to the detector ring axis. The full 4π emission data are then reconstructed and processed to generate the 3D image set with cubic voxel, uniform resolution and signal-to-noise ratio in all directions and locations. Both Monte Carlo simulations and experiments are carried out with simulated and real mouse phantoms. Emission data in both 4π and conventional modes are collected and then processed. Point source is used in the experiment as the fiducial mark. Our results show that with full 4π collection method, the image qualities are substantially improved in several aspects such as the axial distortions and the uniformity of the SNR. Moreover, the axial strip-like artifacts in the conventional mode are canceled in the full 4π mode, therefore less smooth window is needed during the reconstruction which led to higher resolution.
UR - http://www.scopus.com/inward/record.url?scp=84863346626&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84863346626&partnerID=8YFLogxK
U2 - 10.1109/NSSMIC.2011.6152506
DO - 10.1109/NSSMIC.2011.6152506
M3 - Conference contribution
AN - SCOPUS:84863346626
SN - 9781467301183
T3 - IEEE Nuclear Science Symposium Conference Record
SP - 2865
EP - 2869
BT - 2011 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS/MIC 2011
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2011 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS/MIC 2011
Y2 - 23 October 2011 through 29 October 2011
ER -