Abstract
655
Objectives: In clinical applications of simultaneous PET/MR abdominal imaging, such as regular and delayed scans of patients with liver cancer, respiratory motion often causes imaging artifacts in both PET and MR images. Motion artifacts can also induce inaccuracy in PET attenuation correction due to spatial mismatch between PET and MR images at the diaphragm. Traditional gating involves additional hardware such as respiratory belt or camera, which increases the complexity of patient positioning and more radiation exposure to the operator. Moreover, these external gating device based methods are often lack of accuracy because they are designed to track belly movement rather than directly measuring organ location. To overcome these difficulties, in this work, an adaptive PET/MR motion correction technology based on real-time 2D excitation MR navigator has been designed for high definition abdominal imaging and assessed on phantom and patients with liver cancer.
Methods: A pencil beam navigator excited by a 2D RF pulse is positioned along the motion direction and the projection profile is obtained by the Fourier transform of the measured signal. The respiratory motion vector is extracted by utilizing an edge detection algorithm. Then the estimated motion is deployed to prospectively trigger the MR data acquisition and retrospectively gate PET data. 4 Frames of PET images and attenuation map are transformed to a reference frame through a non-rigid motion compensation method and then attenuation corrected to create a high SNR PET image without motion blur (Figure 1). To verify this approach, in the phantom study, three Na-22 point-sources were positioned on a MRI phantom which is driven by the patient couch to move back and forth to simulate a respiratory motion. In the human study, navigator FOV was placed across the right hemi-diaphragm along the superior-inferior direction and a respiratory belt was placed around the upper abdomen to acquire respiratory signals for comparison. PET data was acquired simultaneously with navigator and navigator triggered T2 weighted fast spin echo sequence.
Results: Navigator successfully tracked the abdominal motion and the motion vector extracted by the edge detection algorithm well delineated the diaphragm movement (Figure 2). MR signal to noise ratio (SNR), MR ghost to image ratio (GIR, defined as the mean value within ghost area divided by the signal mean value around the region of interest) and FWHM values of lesions in PET images have been evaluated quantitatively with phantom and patient. In phantom studies, SNR increased from 163.89 to 662.23 whereas GIR reduced from 24.87% to 2.94%, suggesting a significant improvement compared to those without motion correction. Average FWHM value of Na-22 point source was reduced from 42.8mm to 4.4mm. In clinical images, FWHM values drawn from multiple locations of the lesions decreased from 5.47mm to 3.59mm (Figure 4) compared to PET images without motion correction.
Conclusions: MR navigator guided adaptive motion correction offers a clinical solution for high definition simultaneous PET/MR imaging of abdominal with streamlined patient handling. Validation result in both phantom and clinical imaging shows that this technology precisely capture abdominal motion and effectively eliminate motion blurring.