Premium
Technical Note: Rod phantom analysis for comparison of PET detector sampling and reconstruction methods
Author(s) -
Wollenweber Scott D.,
Kemp Brad J.
Publication year - 2016
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4964458
Subject(s) - imaging phantom , scanner , image quality , detector , image resolution , voxel , nuclear medicine , noise (video) , quality assurance , iterative reconstruction , computer science , artificial intelligence , biomedical engineering , optics , computer vision , physics , medicine , external quality assessment , pathology , image (mathematics)
Purpose This investigation aimed to develop a scanner quantification performance methodology and compare multiple metrics between two scanners under different imaging conditions. Most PET scanners are designed to work over a wide dynamic range of patient imaging conditions. Clinical constraints, however, often impact the realization of the entitlement performance for a particular scanner design. Using less injected dose and imaging for a shorter time are often key considerations, all while maintaining “acceptable” image quality and quantitative capability. Methods A dual phantom measurement including resolution inserts was used to measure the effects of in‐plane ( x , y ) and axial ( z ) system resolution between two PET/CT systems with different block detector crystal dimensions. One of the scanners had significantly thinner slices. Several quantitative measures, including feature contrast recovery, max/min value, and feature profile accuracy were derived from the resulting data and compared between the two scanners and multiple phantoms and alignments. Results At the clinically relevant count levels used, the scanner with thinner slices had improved performance of approximately 2%, averaged over phantom alignments, measures, and reconstruction methods, for the head‐sized phantom, mainly demonstrated with the rods aligned perpendicular to the scanner axis. That same scanner had a slightly decreased performance of −1% for the larger body‐size phantom, mostly due to an apparent noise increase in the images. Most of the differences in the metrics between the two scanners were less than 10%. Conclusions Using the proposed scanner performance methodology, it was shown that smaller detector elements and a larger number of image voxels require higher count density in order to demonstrate improved image quality and quantitation. In a body imaging scenario under typical clinical conditions, the potential advantages of the design must overcome increases in noise due to lower count density.