Quality guarantee for tomographic images developed

CWI researcher Wagner Fortes has developed new methods to determine the quality of tomographic images. Tomography is an imaging technique used in for instance medical CT scanners and electron microscopes that uses penetrating beams to record a series of two-dimensional images from a range of angles.

Publication date: 12-09-2013

Space InvaderCWI researcher Wagner Fortes has developed new methods to determine the quality of tomographic images. Tomography is an imaging technique used in for instance medical CT scanners and electron microscopes that uses penetrating beams to record a series of two-dimensional images from a range of angles.

Mathematical analysis of this set of images results into a 3D reconstruction of the interior of the object. The methods developed by Fortes aim to give a theoretical guarantee that this reconstruction resembles the actual scanned object. Fortes will defend his thesis ‘Error bounds for discrete tomography’ on Wednesday 18 September 2013 at Leiden University.

When performing a CT scan of a patient, the judgment of a doctor relies on the quality of the reconstructed images. Any measure of the quality of the image resulting from the mathematical analysis of the scans is therefore most welcome. Providing quality bounds for the mathematical method used is however very difficult. The method is an approximation, scans are usually corrupted by noise and an infinite number of theoretically possible images can be reconstructed from a single set of scans.

Making clever use of very basic prior knowledge about the nature of the scanned object, Fortes has succeeded in formulating quality guarantees that can be used if the scanned object is homogeneous, i.e. consisting of a single material. These guarantees tell us to which extent the reconstructed image can be different from the scanned object. It is the first time that such a guarantee can be formulated. These error bounds for homogeneous objects can be an important step in finding quality bounds for more complex images.

Wagner Fortes is a PhD student at Centrum Wiskunde & Informatica under the supervision of prof. dr. Joost Batenburg in the Scientific Computing group. In 2011 he won a Best Paper Award at the 16th IAPR International Conference on Discrete Geometry for Computer Imagery. His work is funded by the EU Erasmus Mundus programme and Leiden University.

Thesis: ‘Error bounds for discrete tomography’
By: Wagner Fortes (CWI Scientific Computing group)
Promoters: prof. dr. K.J. Batenburg (CWI and UL) and prof. dr. ir. B. Koren (Eindhoven University of Technology)
Date: 18 September 2013, 13.45h
Place: Academiegebouw Universiteit Leiden

Image: A typical example of an homogeneous object