Variation of the Hounsfield Unit On CT Scanning Parameters and Reconstruction and Its Effect On Dose Calculations
J Lee*, R Chan, Medstar Washington Hospital Center, Washington, DCSU-E-J-171 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: To assess the variation of Hounsfield units (HU) when using different CT scanning parameters. The effect of HU-to-electron density (ED) table on the accuracy of dose calculations was evaluated.
Methods: A CIRS (model 62) phantom was used for all the tests. Various kVp, field-of-view(FOV), reconstruction algorithms with contribution of adaptive statistical iterative reconstruction(ASiR™) (0-80%) were applied during the scanning process(GE lightspeed CT simulator). HU for each insert was measured. A HU-ED table was acquired for Philips PET/CT scanner as well. A phantom and clinical cases were selected for dosimetric comparisons using different HU-ED tables.
Results: No significant difference in HU for low electron density material was observed regardless the scanning parameters. The dependency of HU on tube voltage and FOV was observed for bone- equivalent materials. HU for dense bone scanned with 80kVp was 39.4% higher than HU with 120kVp. HU with small FOV relative to large FOV was 11.9% higher for trabecular bone. The maximum deviation was observed for bone-equivalent material when the bone algorithm, producing a 6.2% change in HU, or 80% of ASiR™ contribution, were applied; however, the difference in HU was still within the standard deviation. The difference in HU between two scanners was less than the SD observed on GE CT simulator, except for liver- and bone-equivalent materials. Using a HU-ED table with the worse case scenario (80kVp, bone reconstruction with 80% ASiR™), it was shown that the deviation was within 1% for phantom studies and -7.4%-2.7% for clinical cases. The maximum of 5.9% deviation was observed when calculated with a table from the different scanner.
Conclusion: The maximum deviation in HU was observed for bone-equivalent materials when low kVp was applied. Use of a wrong HU-ED table could introduce -7.4%-2.7% dosimetric error at certain regions of interest.