Comparison of different figure of merit functions for dynamic single photon emission computed tomography (dSPECT)

Abstract

Dynamic Single Photon Emission Computed Tomography (dSPECT) is a technique which visualizes changing activity distributions in the human body, using dynamic emission data acquired during a single rotation of a standard SPECT camera system. The reconstruction process in dSPECT is based on nonlinear regularization and optimization techniques, and we presently compare a number of possible problem oriented figures of merit based on different spatial and temporal regularization techniques. The accuracy of dSPECT in terms of temporal and spatial resolution is tested in a simulated dynamic cardiac study and a dynamic renal patient study.

The inverse problem of emission tomography

“…which is simpler than (9). The major advantage of ( 12) over its alter ego ( 9) is that as soon as µ(x) is known, the data may be corrected for attenuation, and the problem reduces to inverting the Radon transform.…”

Fully 4D dynamic image reconstruction by nonlinear constrained programming

“…The vector included to improve stability of the otherwise seriously ill-conditioned problem and several of them have been discussed in the literature e.g. [23]. The data mismatch term…”

“…in principle assumes Gaussian distributed measurements % , but an appropriate weighting may be used to address the more realistic Poisson distribution [23].…”

“…This allows the method to adjust the tentative peak position , so that errors imposed by the prior information may to some extent be corrected or reduced. Prior peak positions may be calculated either by the method proposed in [18], or by fitting "hat functions" [23] or similar functions. This work is currently in progress.…”

“…In this latter case, the temporal position of the peak activity is needed. The methods discussed in [23,18], in particular the "hat method" may be used to obtain prior estimates of the peak positions. However, in those reconstructions, the temporal mask was not calculated, but directly defined using the true activity distributions.…”