Abstract:
HTML
Purpose: The purpose of our work is to improve 3D cardiac Rb-82 PET imaging
using Fourier-wavelet restoration techniques. 3D PET cardiac imaging can suffer
from high contributions of scatter and randoms due to the removal of septa.
However, 3D PET cardiac imaging has higher sensitivity, which might allow lower
doses and make Rb-82 PET cardiac imaging more affordable.
Methods and Materials: 3D myocardium Rb-82 studies were acquired following our
standard clinical protocol approved by Institutional Review Board of our
Medical Center. Activity of 740 MBq (20 mCi)of Rb-82 were injected
intravenously. Emission scans were performed for 6 min after a 3min delay for
blood pool clearance and decrease in randoms rate. Attenuation correction was
performed using an 8 min transmission scan. We used a GE Advance PET scanner,
which has retractable septa and can be used in both 2D and 3D mode. Randoms
subtraction was performed using the delayed coincidence method, which resulted
in negative values in the reconstructed data. The acquisition and
reconstruction matrix size was 128 x 128 and the pixel size was 4.3 mm. Images
were reconstructed with the Kinahan-Rogers 3D filtered backprojection
algorithm. First, the Fourier restoration was applied by using the point spread
function acquired separately. In Fourier domain, the inverse of modulation
transfer function (MTF) was multiplied with the Butterworth low-pass filter,
order n=6 and cut-off frequency f=0.35 cycles/pixel. In addition wavelet noise
suppression was applied by scalar shrinkage. The cardiac studies were evaluated
measuring the contrast between LV wall and LV cavity, using linear profiles and
by visual analysis.
Results: The results showed significant improvement in resolution, contrast and
background subtraction compared to the original unrestored image. The typical
contrast increase between LV wall and LV cavity was about 20%. However, the
restoration process slightly increased noise in the restored slices when
compared with the original image. The amount of noise was measured as a
coefficient of variation (COV, 100 x SD/mean (%)). It increased by only
1.2%-1.8% when Fourier-wavelet restoration was applied.
Conclusion: The quality of 3D myocardial Rb-82 PET images can be significantly
improved by Fourier-wavelet restoration filtering. (J. M. received research
support from Bracco Diagnostics and Medical General Electric.)
Questions about this event email: kknesaurek@mssm.edu
Knesaurek PhD, K,
Fourier-wavelet Restoration of 3D Myocardial PET Images. Radiological Society of North America 2003 Scientific Assembly and Annual Meeting, November 30 - December 5, 2003 ,Chicago IL.
http://archive.rsna.org/2003/3106752.html
