Die Zeitseparationstechnik : eine effiziente modellbasierte Rekonstruktionstechnik für die computertomographische Perfusionsbildgebung [article]

Sebastian Bannasch, Armin Iske, Georg Rose, Gerald Warnecke, Martin-Luther Universität, Universitäts- Und Landesbibliothek Sachsen-Anhalt
2020
Computed tomography (CT) is an X-ray based imaging system. It is used for medical diagnostics and non-destructive testing. The composition of different two-dimensional layers forms a three-dimensional sampled tomographic image, the CT-image. The CT-image provides an insight into the interior of an object. The slices are reconstructed from a large number of X-ray images. The mathematical-physical model of a CT reconstruction problem leads to a so-called ill posed inverse problem. Generally, an
more » ... em. Generally, an ill posed inverse problem is very hard or impossible to solve. The best known modelling approach is based on the work of Johann Radon. The Radon transformation is an essential component in the field of integral geometry. The CT reconstruction requires stable algorithms that can robustly calculate the inversion of a CT system. This work focuses on clinical use, so that the standards for the techniques are very high. This is due to the fact that X-rays are ionising (electromagnetic) radiation that can even increase the probability of developing cancer. A reduction of the X-ray dose leads to a degeneration of the data quality. Hence, new methods and techniques are needed that can provide rapid and accurate CT images for diagnostics. Therefore, computed tomography is a very diverse field of research with contributions from mathematics, physics, computer science and engineering are made. Dynamic computer tomography, using an angiography system for stroke diagnosis, results in high difficulties. In particular, the slow acquisition times for a so-called CT perfusion measurement make it extremely hard to reconstruct an in time resolved CT image. The reason for this is that the dynamic data are temporally under-sampled. In this thesis, priori knowledge is integrated into the reconstruction algorithm by means of a model approach in order to handle this under-sampled problem. The quintessence of the model-based time separation technique (TST) presented here is the strong reduction of computational complexity. While a model-based approa [...]
doi:10.25673/32356 fatcat:i2sxgiqamvc3dipqwrp6tllevu