MR-based Attenuation Correction for PET/MR - Correction Techniques in Emission Tomography - page 221

mission images using a surface matching technique. The co-registered MR
images are then segmented into 3 classes (Table 11.1). Air was considered to
be present only outside the patient. Appropriate linear attenuation coecient
values at 511 keV are then assigned to these tissue classes. Le Goff-Rougetet
et al. evaluated their MR-AC method for phantoms and only 1 patient scan.
Using ROIs within selected axial images of the phantom the authors found a
maximum difference of 11% between PET MRAC and PET TXAC . The patient
study revealed a maximum difference of 12%, primarily in the occipital cortex.
In a publication by El Fakhri and colleagues [12] MR-AC was also men-
tioned. However, the authors provided neither details of their implementation
nor a performance evaluation. Personal communication with the authors sug-
gests that they acquired 2 MR sequences for each subject and performed a
cluster identification on the joint histogram prior to assigning the correspond-
ing attenuation values.
An alternative method for MR-AC in brain PET was suggested by Zaidi et
al. [41]. The authors had previously shown that the quality of PET neurology
imaging was insucient when standard PET attenuation correction methods
were applied [40]. Therefore, the authors studied the use of MR-AC in brain
PET (Table 11.1). They presented a workflow based on the availability of
co-registered PET images, following standard (ellipse-fitted) attenuation cor-
rection, and MR images [41]. Using a fuzzy-logic-based segmentation method,
the co-registered MR images were segmented into 5 tissues classes that were
assigned attenuation coecients at 511 keV. The entire process for MR-AC
was reported to take 10 min on a Sun SPARC with minimal user intervention.
Similar to Le Goff-Rougetet, the authors accounted for the head holder before
using the segmented MR-based attenuation map for MR-AC.
Figure 11.2 shows brain images of an FDG-PET study from Zaidi et al
[41] who compared segmented MR-AC with a standard, ellipse-fitted AC. The
image quality of the PET following MR-AC appears somewhat improved,
which can be attributed to the lower noise levels in the MR-based attenuation
maps (Figure 11.2B). Analysis of the differences of the two methods of AC
was performed across 10 patient sets. Despite a tendency of the method to
lead to activity overestimation, overall correlation of ROI activity values on
PET MRAC and PET TXAC was good (R 2 =0.91), indicating the feasibility of
segmented MR-AC as suggested by the authors (Table 11.1).
11.2.2 Atlas approaches
A viable alternative to multi-step segmentation procedures [41, 10] is to use
atlas-registration (Figure 11.3). An atlas typically consists of a template MR
image together with a corresponding attenuation label image. The template
MR image can be obtained as an average of registered MR images from several
patients. The label image could represent a segmentation into different tissue
classes (e.g., air, bone, and soft tissue) or a co-registered attenuation map from
a PET transmission scan or a CT scan with continuous attenuation values.