Jeffrey Steinberg, Presenter: Nothing to Disclose

Guang Jia PhD, Abstract Co-Author: Nothing to Disclose

Zarine Ketul Shah MD, Abstract Co-Author: Nothing to Disclose

Steffen Sammet MD, PhD, Abstract Co-Author: Nothing to Disclose

Nathan C. Hall MD, PhD, Abstract Co-Author: Nothing to Disclose

Michael Vinzenz Knopp MD, PhD, Abstract Co-Author: Nothing to Disclose

To segment and identify different tissue types in MRI for the purpose of generating an attenuation map for image reconstruction in PET for MRI/PET hybrid systems.

PET, CT, and MRI examinations of the thoracic and abdominal regions of a patient were performed. These regions were chosen due to the large number of tissues of different attenuation in the torso. The MR images (matrix size: 512x512) were registered to PET images (305x305) and then segmented using simple thresholding of anatomical structures. That is, tissues were identified based on the signal intensity of each pixel. Thresholding for T1 weighted MR images were used to identify anatomical structures due to the high contrast between lung and soft tissues. The segmented tissues are then assigned a specific attenuation value based on the average attenuation of the tissue. The resulting attenuation map is smoothed with a low pass filter and used for PET image reconstruction. These images are compared to CT attenuated PET images for accuracy.

An approximate PET image is obtained using an attenuation map with only air, lung, and soft tissue segmented. The absolute difference between the resulting image and the CT attenuated image is 17.3%, which is much less than the 86.5% difference between unattenuated and CT attenuated images. The difference image reveals a lack of definition in anatomical structures such as pulmonary vessels, intestines, and bone. Despite the differences, the PET images are capable of distinguishing important anatomical structures, such as lung, liver, pancreas, and mediastinum.

The MRI-segmented attenuation maps in this study matched the basic anatomical structures to produce PET images comparable to those generated from CT-based attenuation maps. Although segmentation of air, lung, and soft tissue is sufficient for a rudimentary reconstruction, further segmentation of fine details in the anatomical structure is necessary for PET images with the quality of CT-based attenuation.

Calculating MRI-based attenuation maps is important in the development of MRI/PET, which uses MRI’s high inherent soft tissue contrast to potentially improve lesion localization and characterization.

Steinberg, J, Jia, G, Shah, Z, Sammet, S, Hall, N, Knopp, M, MRI-Segmented Attenuation Maps for PET Reconstruction.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/5011468.html