Motion Displacement Field Estimation using Time-of-Flight PET Histoimages

• Published in: IEEE conference record - Nuclear Science Symposium & Medical Imaging Conference. pp. 1 - 3
• Main Authors: Li, Yusheng, Karp, Joel S., Matej, Samuel
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2019
• Subjects: displacement field; Estimation; histoimage; Image reconstruction; Logic gates; Mathematical model; Motion estimation; Noise measurement; Positron emission tomography; positron emission tomography (PET); time-of-flight (TOF)
• ISSN: 2577-0829
• DOI: 10.1109/NSS/MIC42101.2019.9059910

Subject motion including cardiac and respiration motion during data acquisition can cause degraded image reconstruction. Time-of-flight (TOF) in PET provides additional useful information, which can be utilized for motion estimation. We aim to develop accurate motion field estimation and motion compensation from TOF PET data fully utilizing TOF information for improved quantitative PET imaging. In this work, we develop a new motion displacement field estimation method using TOF PET histoimages. The TOF list-mode data can be efficiently deposited into a TOF histoimage, and radioactive tracer distribution can be efficiently reconstructed using the DIRECT (Direct Image REConstruction for TOF) approaches since TOF histoimage has the same spatial coordinates and geometry as the reconstructed image. First we derive the optical flow equation for TOF histoimage, from which we formulate the global energy function including the total squared error of data consistency based on the optical flow equation and a displacement regularizer. We then obtain Euler-Lagrange equation of TOF histoimage by minimizing the global energy function based on the principle of least action. Inspired by Horn-Schunck method, we obtain an iterative algorithm for motion displacement field estimation, and displacement smoothness regularization is naturally included. We simulate a generic 2D TOF PET scanner using XCAT phantom with cardiac motion, and show accurate motion displacement field can be obtained from both noise-free and noisy TOF histoimages using the proposed method. In summary, we present a motion displacement field estimation method directly using TOF histoimages, which can be important for motion compensation and improved dynamic image reconstructions.