HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Chang, G. Articles by Mawlawi, O. PubMed Articles by Chang, G. Articles by Mawlawi, O.

Optimization of super-resolution (SR) processing using incomplete image sets

¹ Imaging Physics, MD Anderson Cancer Center, Houston, Texas; ² ECE, Rice University, Houston, Texas

1658

Objectives: We recently proposed a SR technique for PET imaging that combined multiple images generated from the same data set but reconstructed from different points of view (JNM 2007; 48:411P). This approach increases processing time due to the large number of required images. This abstract investigates 2 optimized SR (SR1 & SR2) techniques that combine incomplete image sets to reduce processing time and compares them with the complete SR (CSR) image set with respect to resolution and SNR.

Methods: An F-18 point source was imaged on a GE DRX scanner for 1 min. 16 images (64*64) were reconstructed from this data set in a 25.6 cm FOV while shifting the reconstruction grid. The centers of the 16 images formed a 4*4 square lattice (1mm apart). SR algorithm was applied to combine 16 (all), 7 (two sides of the square) and 4 (diagonal) images to generate the CSR, SR1 and SR2 respectively. Line profile was drawn along the point source for all the SR images and the FWHM was calculated. An IEC phantom containing six spheres (1-3.7 cm) was also scanned in 2D mode for 3 minutes. The CSR, SR1 and SR2 images were generated in a FOV of 51.2 cm using the same configuration as in the point source study. SNR and line profiles were plotted for each sphere for all three images. A 2-D FFT was also applied to investigate the noise spectrum for each SR image.

Results: Line profiles of the point source show that the three SR images have similar FWHM and signal amplitude. The IEC study shows that the three SR images have an average difference of 3.3% in SNR with respect to one another and contain similar noise structure.

Conclusions: SR images generated by combining an incomplete set of images can be used to reduce the total processing time while maintaining similar resolution, contrast, SNR and noise structure as the CSR.

This Article Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Chang, G. Articles by Mawlawi, O. PubMed Articles by Chang, G. Articles by Mawlawi, O.