Iterative image reconstruction using geometrically ordered subsets with list-mode data

• Main Authors: Popescu, L.M., Matej, S., Lewitt, R.M.
• Format: Conference Proceeding
• Language: English
• Subjects: Change detection algorithms; Event detection; Histograms; Image reconstruction; Iterative algorithms; Maximum likelihood detection; Maximum likelihood estimation; Memory management; Positron emission tomography; Reconstruction algorithms
• ISSN: 1082-3654; 2577-0829
• DOI: 10.1109/NSSMIC.2004.1466649

In positron emission tomography (PET), the format in which the data is stored has a major influence on the image reconstruction procedure. The use of the list-mode format preserves all of the measured attributes of the detected photon pairs but the events are stored in the order that they were measured, which allows only sequential access to the data. This fact limits the number of applicable algorithms and often computing speed or memory capacity constraints require the use of algorithms that do not make full use of the original precise information in the data. In this paper we show how through a change of the format in which the data is stored one can keep all the initial information about the individual events while providing random access to subsets of events belonging to given geometrical regions, thus making possible the use of maximum likelihood ordered subsets (OSEM) type algorithms with data provided as a collection of individual events (list-mode), and facilitating the adaptation of other types of algorithms. The structured data format also allows for more compact (compressed) storage of the information compared to the simple list-mode format.