This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Ischaemia / Infarction, Imaging sequences, Contrast agent-intravenous, CT-Quantitative, CT, Interventional vascular, Computer applications
Authors:
G. Van Eyndhoven1, J. Sijbers2, J. Batenburg3; 1Wilrijk/BE, 2Antwerp (Wilrijk)/BE, 3Amsterdam/NL
DOI:
10.1594/ecr2014/C-0282
Methods and materials
Data acquisition
Data acquisition is performed by rotating source and detector multiple times around the patient.
In each 180° rotation,
data is sampled equiangularly with a starting angle chosen such that no angle is sampled twice during the entire acquisition protocol.
Reconstruction method
Overview
In contrast to the conventional approach where reconstructions are calculated independently per 180° projection data set,
our algorithm exploits the temporal correlation of the reconstructions.
This is achieved in two ways: static regions (e.g.,
bone and air) are forced to be constant over time and dynamic regions (e.g.,
artery and vessels) are forced to have locally similar enhancement.The temporal correlation is exploited at every step of the algorithm,
resulting iteratively in more refined TCCs and improved perfusion maps.
Furthermore,
at every iteration,
an intermediate optimization of the arterial input function (AIF) is performed,
since the calculation of the perfusion maps depend heavily on the AIF's accuracy.
Detailed description
The different steps of the proposed algorithm are visualized in the flowchart of Fig.
2.
First,
an initial reconstruction per 180° rotation of source and detector (i.e.,
per point in time) is calculated with the simultaneous iterative reconstruction technique (SIRT) [4].
Next,
this estimate is refined by iteratively repeating the following four steps.
The first step consists of adding a traditional SIRT update step to each time point reconstruction.
Next,
the SIRT update step in the static regions of each time point is added to the reconstruction on all other time points.
The third step starts by calculating the local enhancement based on the current reconstruction,
which is defined as the ratio between each pair of reconstructions (at different time points) of the average attenuation values in a small neighborhood (typically a 3x3 neighborhood) surrounding each pixel.
This local enhancement is utilized to add a weighted version of the SIRT update step in each time point to the reconstruction at all other time points.
Finally,
a least square solution for the set of coefficients corresponding to gamma variate basis functions is fitted to minimize the projection distance (i.e.,
the distance between simulated and acquired projection data) in each pixel belonging to an artery.
This last step ensures an accurate estimate of the arterial input function (AIF),
which is of critical importance for the estimation of CBF and CBV map.