Click here to


Are you sure ?

Yes, do it No, cancel

Toward a Meaningful Figure-Of-Merit for CT Image Quality Assessment During Routine QC

V Mamyan*, E Marshall , Z Lu , I Reiser , The University of Chicago, Chicago, IL


(Monday, 7/15/2019) 3:45 PM - 4:15 PM

Room: Exhibit Hall | Forum 9

Purpose: The limitations of CNR as an image quality metric in CT are well known, particularly its failure to correlate with changes in low-contrast detectability when iterative reconstruction is employed. Despite this limitation, CNR is proportional to detectability so long as noise correlations in the image do not change. CNR is ideal for use in routine QC owing to its simplicity. Based on these observations, we propose a novel figure-of-merit for CT that avoids the pitfalls of CNR and provides information that allows for meaningful comparison of image quality between CT scanners. The purpose of this work was to investigate the reproducibility of the new metric, and to identify and minimize sources of uncertainty.

Methods: The proposed image quality index (IQI) is defined as the ACR-defined CNR divided by the square-root of CTDI100 (center) and slice thickness, but with CNR measured on images reconstructed with a standard smooth kernel and without iterative reconstruction, regardless of clinical protocol settings. In order to eliminate uncertainty from ROI placement, a systematic approach for placing all ROIs on the same radius was developed. The ACR CT Accreditation Phantom was scanned on CT scanners from two different vendors. Reproducibility was assessed from five repeat scans. Independence of IQI from dose and slice thickness was verified and sources of deviation and uncertainty were examined.

Results: The error in CNR measurements was less than 10% for both vendors and head and abdomen protocols. The expected dose and slice-thickness dependence of the standard deviation was observed for a 5cm² ROI at the phantom center, but deviated for a 1cm² ROI.

Conclusion: The systematic approach enables reproducible CNR measurements. The proposed IQI allows for comparison of CT scanners with similar reconstruction filter shape, and has potential applications in scanner harmonization across a diverse CT scanner fleet.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email