MR-based Attenuation Correction for PET/MR - Correction Techniques in Emission Tomography - page 224

The template MR is warped to the patient-specific MR image volume. When
applying the same spatial transformation to the atlas attenuation image a
corresponding patient-specific attenuation map is generated.
Atlas-based approaches to MR-AC were presented by Rota Kops et al.
[33] and Hofmann et al. [16]. Rota Kops and colleagues generated a template
of PET transmission images from 10 patient datasets that is matched to the
PET transmission template within SPM2 [13]. The MR template within SPM2
(which is already aligned with the PET transmission template) was normalized
to the MR image of the patient. The resulting transformation was then applied
to the template attenuation image, thus yielding an attenuation image for
this patient. The same group has also employed MR-AC based on an MR
segmentation method by Dogdas et al. [10]. Rota Kops et al. [33] validated
their MR-AC algorithm with 4 patients (Figure 11.4). An analysis of the rather
large ROIs defined on cortical and sub-cortical structures demonstrated that
PET MRAC differed from PET TXAC by up to 10%. When using the MR-derived
segmentation by Dogdas, maximum differences were observed in the occipital
cortex and caudate nucleus, with up to 10% difference from PET TXAC .
Hofmann et al. suggested a revised atlas approach to MR-AC [16]; see
Figure 11.5. Here, the authors utilize a set of aligned MR-CT image volumes
of 17 patients. Each of the available 17 MR image volumes from the MR-CT
pairs was co-registered to the MR image volume from the PET/MR study. The
co-registration vectors were applied to the corresponding CT image volumes
thus generating 17 CT image sets that were aligned to the MR set from the
patient. Subsequently, a pattern recognition approach was used to match the
MR image of the patient with the appropriate CT information from that
MR-CT data set that best matched the patient information. This voxel-based
approach can merge partial sub-volumes from independent data sets into a
single CT-volume that is used for MR-AC of the patients. This atlas-based
algorithm was validated on 3 clinical data sets comparing MR-AC to the gold
standard CT-AC [16]. Automated ROI analysis of PET MRAC and PET CTAC
yielded a mean absolute dierence of 3% 2:5%. Mean dierences for the
standard brain regions were smaller 10% and a maximum difference of 10%
was observed for a meningioma located directly next to the skull.
11.3 Methods for torso imaging
Due to the current lack of prototype systems for a whole-body PET/MR
system, studies of MR-AC algorithms for extra-cranial applications are scarce.
Beyer et al. studied 10 patients who underwent routine torso scans (with arms
up) on a combined PET/CT tomograph [3]. Within 1 day of the PET/CT
exam, complementary MR scans were acquired. MR imaging was performed
on a 1.5 T system with patients being positioned with their arms down.