J Nucl Med. 2007; 48 (Supplement 2):13P
General Clinical Specialties: Musculoskeletal
Infection and Improved Scintigraphic Techniques |
Reduction in scan duration or injected dose in planar bone scintigraphy enabled by Pixon® post-processing
Osama Mawlawi1,
Amos Yahil2,
Hans Vija3,
William Erwin1 and
Homer Macapinlac1
1 Imaging Physics, MD Anderson Cancer Center, Houston, Texas;
2 ImageRecon LLC, Stony Brook, New York;
3 Siemens Medical Solutions, Hoffman Estates, Illinois
43
Objectives: Pixon processing has been shown to be capable of preserving image quality in count-reduced planar imaging (JNM Vol. 47 (5) 365P 2006). The focus of this investigation was Pixon processing of planar bone scans that have been subsampled by 50%, thereby simulating a 50% reduction in scan duration or injected dose. The objective is to compare the Pixon processed images to the original non-subsampled bone scans with regard to diagnosis and reader confidence. Methods: 24 planar bone scans were evaluated by 7 experienced nuclear medicine physicians. 50% count-reduced images were first obtained from the original scans using binomial (Poisson preserving) subsampling followed by Pixon image processing. These images were then linearly blended with the unprocessed, 50% subsampled images. Two Pixon blending fractions were used: 30% (mimicking the SNR of the original images) and 50% (mimicking images with double the original counts), resulting in a total of 48 (24*2) bone scans. Custom software, consisting of two adjacent display windows with slide bars to manipulate the window/level and gamma, was written to present the images in random order. In each case, one window displayed the original (non-subsampled) bone scan while the other showed the Pixon processed and blended image. For each case, the physicians were asked to rate the images for diagnostic value and reader confidence on a 5 point scale. Results: [table] Conclusions: For 30% Pixon blending, the diagnostic performance and reader confidence was equal or better than the original image in 96% and 88% of the cases respectively. For 50% blending the results were 83% and 69% respectively. This finding suggests that Pixon processing can reduce scan duration or injected dose by 50% with minimal impact on the clinical value of the resulting images. In this regard, Pixon processing with image blending could increase patient throughput, minimize potential motion blur or reduce overall patient dose.
| |
Frequencies of Scores
|
|
