Spatiotemporal Decomposition in Object-Space along Reconstruction in Emission Tomography

Abstract : Emission tomography has provided a new insight in brain mechanisms past years. Although reconstructions are nowadays mostly static, trend is going toward dynamic acquisitions and reconstructions. This opens a new range of investigations, for instance for drugs discovery. Indeed new drugs are studied through the dynamic ability of tissues to catch them. However, it is required to know radiotracer concentration of blood that irrigates tissues in order to draw conclusions on potentials of these drugs. This concentration is called 'input function' and this paper presents a new method for measuring it in a non-invasive way. Our new method relies on simultaneous estimations of vessels kinetics and vessels spatial distribution. These estimations are performed during the reconstruction process and take into account the statistical nature of measured signals. Indeed, this method is based on the maximisation of the likelihood of counts in detectors. It takes advantages of a non-negative matrix factorisation which separate spatial and temporal components. Results are very promising, since it estimates arterial input function accurately although object emits just a limited amount of photons, especially within the first minutes.
Document type :
Conference papers
Liste complète des métadonnées
Contributor : Enzo Ferrante <>
Submitted on : Saturday, December 14, 2013 - 1:03:12 PM
Last modification on : Thursday, March 7, 2019 - 3:34:14 PM

Links full text




Xavier Hubert, Dominique Chambellan, Samuel Legoupil, Régine Trébossen, Jean-Robert Deverre, et al.. Spatiotemporal Decomposition in Object-Space along Reconstruction in Emission Tomography. Medical Image Computing and Computer-Assisted Intervention - MICCAI 2008, 2008, New-York, United States. pp.255-262, ⟨10.1007/978-3-540-85990-1_31⟩. ⟨hal-00918726⟩



Record views