Click here to


Are you sure ?

Yes, do it No, cancel

Radiation Dose Constraint to Maintain CT Number Accuracy

R Zhang1*, J Cruz-Bastida1 , D Gomez-Cardona1 , J Hayes1 , K Li1 , G Chen1 , (1) University of Wisconsin-Madison, Madison, Wisconsin


(Sunday, 7/29/2018) 1:00 PM - 1:55 PM

Room: Room 205

Purpose: The accuracy of the CT number plays a key role in clinical diagnosis. For reconstruction algorithms that use post-log projection data, the reconstructed CT number may be biased, especially when radiation dose is reduced. The purpose of this work was to investigate the lower limit of radiation dose level if one desires to maintain a certain level of CT number accuracy.

Methods: A lesion-background model was developed to theoretically study the CT number bias for FBP reconstruction. Using the developed theoretical model, a map of CT number accuracy changes with the radiation exposure level and intrinsic image contrast was generated and experimentally validated. A Catphan phantom was scanned using a GE Discovery CT 750HD at different mAs levels. 50 repeated scans were performed for each mAs. The accuracy of CT number was evaluated for each combination of mAs and ROI with different contrast levels.

Results: Both theoretical and experimental studies showed that the CT number bias is inversely proportional to the radiation exposure level and linearly dependent on the CT number contrast between the lesion and the background, i.e. Bias=α/mAs(1+β∆HU) , where parameters α and β can be determined experimentally.

Conclusion: In this work, we showed that FBP reconstruction may introduce significant bias as mAs is reduced; both positive and negative bias can be observed depending on the contrast difference between a targeted ROI and its surrounding background. Using the theoretical model, an iso-contour map of the CT number accuracy can be generated, which can be used to determine the lower limit of radiation dose level for a given requirement of CT number accuracy.

Funding Support, Disclosures, and Conflict of Interest: This work is partially supported by an NIH grant R01CA169331.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email