EFRE Logo

23.02.2017 - Medical Imaging Reconstruction Seminar - Termin verschoben!

Am Donnerstag, dem 23.02.2017 (ursprünglich für den 16.02.2017 geplant) findet im Konferenzraum der ExFa (1. Stock) ab 14.00 Uhr das nächste Medical Reconstruction Seminar statt.

Vortragender: Sebastian Bannasch (MSc.)


Vortragstitel: Model-based Perfusion using Orthogonal Basis Functions

 

Synopsis:

Purpose: The issue of model-based methods is to increase the image quality of reconstruction of tissue perfusion measurements using a slowly rotating X-ray based imaging system by temporal interpolation. There are several model-based methods which handle temporal undersampled X-ray-based tomographic data by including a priori knowledge. The aim of this work is to provide a fast algorithm including the full impact of model-based reconstruction.

Methods: Here, an analytic pre-processing in the projection space is focused. Consequently, in case of model-based imaging via linear spatio-temporal decomposition, also each projection can be described by a linear spatio-temporal decomposition model. This point of view provides a decomposition method, which transforms the initial dynamic into various static CT problems, such that a recourse to an arbitrary algorithm for reconstruction is possible.

Results: The presented theory of the Time Separation Technique for model-based computed tomography perfusion imaging is evaluated on an in-silico perfusion phantom. Here, the experimental setup is inspired by a C-Arm system with cone beam geometry and represents an appropriate platform for evaluation. By the use of deconvolution-based analysis of CT brain perfusion the reconstructions of the pre-processed data will be demonstrated and evaluated by perfusion parameter maps.

Conclusion: The Time Separation Technique provides a solution for handling the undersampled perfusion data without increasing the sampling rate of measured projections or needed X-ray dose, respectively. In this study, a technique is presented which is capable to close the gap between the accuracy of the linear spatio-temporal decomposition model application and with its connected high computational effort in X-ray-based imaging systems and indicates considerable potential for a practical application.