Software suite for skull segmentation that can facilitate MR image-based
attenuation correction of PET brain imaging.
- A post-processing software suite for skull segmentation that requires no
manual intervention while maintaining high accuracy.
- Can be used to analyze images from independent PET and MRI scanners as
well as the fully integrated imaging device.
- Insensitive to noise and able to process MR images with low quality.
Each year 6.6 million MRI head scans are performed in the United States. In
2010, 1.74 million clinical PET patient studies were conducted, and the number
is estimated to be 2.5 million in 2011. The combination of these two imaging
technologies is considered to be the next generation of multi-modality imaging
because it synergizes the high-sensitivity and functional imaging of PET, and
the high-resolution anatomic information of MRI. To utilize this combined
imaging technology in clinical practice without added expense of purchasing a
new PET/MRI machine, post-processing software is indispensable.
Attenuation correction is an essential step in the study of quantitative PET
imaging of the brain. Because of the inability to offer transmission scans by
most PET/MRI hybrid systems, post-processing software is required to conduct the
MR-image based attenuation correction for PET, with skull segmentation as the
kernel of the process. However, without manual intervention, existing methods
for skull segmentation are suboptimal due to their lack of accuracy or
Emory researchers have developed an automatic post-processing software suite
for skull segmentation, and subsequently for MR image-based attenuation
correction of PET brain imaging. It is designed to co-register MR and PET images
of the brain, conduct segmentation, classification and finally correct the
attenuation for the PET image based on MR data. As a result, a single image of
the patient's skull is reconstructed. Extensive validation tests have
demonstrated its accuracy and insensitivity to random noise, even with
low-resolution MR images. In addition to analyzing images from hybrid PET/MRI
imaging device, the software suite can be used to process the results of two
separate scans taken on different scanners even with weeks apart. The software
suite can be used to facilitate combined PET/MRI imaging technology in detecting
and studying various neurological disorders.
Evaluation tests have been conducted on brain phantom database, simulated MR
images as well as real patient MRI and CT data sets.