Related Experiment Video
Updated: Apr 14, 2026
A Sectioning, Coring, and Image Processing Guide for High-Throughput Cortical Bone Sample Procurement and Analysis for Synchrotron Micro-CT
Published on: June 12, 2020
[Investigation of the Optimal Parameters for Block Sequential Regularized Expectation Maximization (BSREM)
Hayato Sato1, Hayato Odagiri1,2, Nao Yamaguchi1
1Department of Diagnostic Image Analysis, Tohoku University Graduate School of Medicine.
Purpose:
This research aims to optimize the block sequential regularized expectation maximization (BSREM) reconstruction in bone single photon emission computed tomography (SPECT) images using a ring-type SPECT/computed tomography (CT) system in terms of image quality and accuracy of quantitative evaluation.
Methods:
SPECT images were acquired using a ring-type SPECT/CT and from a SIM2 bone phantom containing a 99m-Tc solution. BSREM parameters were assessed with coefficient of variation (CV), recovery coefficient (RC), contrast noise ratio (CNR), standardized uptake value max (SUVmax), and its root mean square error (RMSESUVmax). Clinical images were reconstructed, and the reconstructions were evaluated as superior based on phantom results.
Results:
Increasing γ led to higher RC and CV, while increasing β lowered these values. CNR showed no consistent trend. In terms of quantitative evaluation, γ value 5.0, β values 0.3-0.9, and γ value 10, βvalues 0.3-0.9 resulted in lower RMSESUVmax compared to Anger type. Based on the phantom experiment, γ value 5.0, β values 0.5-0.9, and γ value 10, β values 0.7-0.9 were evaluated as superior. Clinical images under these reconstructions, determined from the phantom experiments, with γ value 5.0, β value 0.9, showed the smoothest of normal bone and the clearest separation of neighboring tumors.
Conclusion:
As reconstruction parameters for BSREM reconstruction in bone SPECT on a ring-type semiconductor SPECT/CT system, a γ value of 5.0 and a β value of 0.9 were considered suitable for image quality and quantitative evaluation.
