Step-and-shoot Dynamic Whole-body PET Parametric Imaging with Weighted Expectation Maximization Algorithm

Dynamic whole-body PET parametric imaging has been widely investigated and demonstrated clinically feasibility. In the step-and-shoot dynamic protocol, an average time point is often assigned to the overlapped slices between adjacent beds. This sum-frame strategy is based on the assumption that the...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:IEEE conference record - Nuclear Science Symposium & Medical Imaging Conference. s. 1 - 3
Hlavní autori: Ye, Qing, Zhao, Yizhang, Xu, Hancong, Lv, Yang, Dong, Yun
Médium: Konferenčný príspevok..
Jazyk:English
Vydavateľské údaje: IEEE 16.10.2021
Predmet:
Heuristic algorithms
Image edge detection
Logic gates
Protocols
Reconstruction algorithms
Sensitivity
Whole-body PET
ISSN:2577-0829
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Dynamic whole-body PET parametric imaging has been widely investigated and demonstrated clinically feasibility. In the step-and-shoot dynamic protocol, an average time point is often assigned to the overlapped slices between adjacent beds. This sum-frame strategy is based on the assumption that the concentration is not varying significantly, which may not be guaranteed for all the voxels covered in the overlapped space. However, when separate time points are assigned, the unbalanced data noise due to the sensitivity drop in the edge slices of each bed causes additional image artifacts. In this study, we proposed to assign separate time points to the overlapped slices (separate-frame strategy) and to apply the weighted expectation maximization (EM) algorithm to the nested direct Patlak reconstruction. GATE simulations and clinical experiments were performed to evaluate the performance of weighted EM algorithm and the time assignment strategies. Compared with the conventional EM algorithm, the weighted EM algorithm effectively reduced image noise and eliminated the artifacts at the boundary of overlapped slices. Overall small differences were found in the overlapped slices between the sum-frame strategy and the separate-frame strategy, while the assumption of small concentration changes may not be satisfactory in some regions, such as the abdomen. Our proposed time assignment and reconstruction methods have been proven to be feasible and reliable.
ISSN:2577-0829
DOI:10.1109/NSS/MIC44867.2021.9875837