Interval-based reconstruction for uncertainty quantification in PET.

Interval-based reconstruction for uncertainty quantification in PET.
Phys Med Biol. 2017 Dec 01;:
Authors: Kucharczak F, Loquin K, Buvat I, Strauss O, Mariano-Goulart D
Abstract
A new directed interval-based tomographic reconstruction algorithm, called Non-additive Interval Based Expectation Maximization (NIBEM) is presented. It uses non-additive modeling of the forward operator that provides intervals instead of single-valued projections. The detailed approach is an extension of the Maximum Likelihood - Expectation Maximization (ML-EM) algorithm based on intervals. The main motivation for this extension is that the resulting intervals have appealing properties for estimating the statistical uncertainty associated with the reconstructed activity values. After reviewing previously published theoretical concepts related to interval-based projectors, this paper describes the NIBEM algorithm and gives examples that highlight the properties and advantages of this interval valued reconstruction.
PMID: 29192605 [PubMed - as supplied by publisher]

Source: Physics in Medicine and Biology - December 1, 2017 Category: Physics Authors: Kucharczak F, Loquin K, Buvat I, Strauss O, Mariano-Goulart D Tags: Phys Med Biol Source Type: research