Program Information
Mammography Accreditation Phantom Variation: Effects On Quality Control Measurements
R Fisher*, K Hulme , The Cleveland Clinic, Cleveland, OH
Presentations
TU-C1-GePD-I-3 (Tuesday, August 1, 2017) 9:30 AM - 10:00 AM Room: Imaging ePoster Lounge
Purpose: To quantify variation in vendor-specified signal-to-noise (SNR) and contrast-to-noise (CNR) measurements made in mammography quality control phantoms as a result of phantom construction.
Methods: Twenty-five Gammex Model 156 phantoms were phototimed (28kVp, W/Rh, average glandular dose 0.8 mGy) 12 times each over the course of two days on a Siemens Inspiration. CNR and SNR values were measured for each image in three ways, corresponding to two different Siemens QC Manual specifications and one Hologic specification. All three methods measured background pixel values in the wax insert for SNR calculations. The Siemens methods measured CNR using 0.2 cm² and 0.4 cm² regions-of interest, respectively, within the largest mass object in the phantom. The Hologic method measured pixels within an acrylic disc on top of the phantom for CNR. Measurements were averaged and variation between phantoms was investigated.
Results: Signal-to-noise ratios calculated via all three methods were very similar, showing a coefficient of variation (CV) of less than 1% for any single method and <2% across all 900 measurements. Contrast-to-noise ratios varied by method, and differences within each method were observed. Using the Hologic specification, the CV of CNR measurements was <1%. Both Siemens methods showed CVs of ~6%, though the mean CNR for the smaller ROI was 12% higher than the larger, equating to a range of ~0.5 around mean values of 2.67 and 2.39, respectively.
Conclusion: A CV of 6% was observed between phantoms when calculating CNR based on embedded mass objects, while using an acrylic contrast disc resulted in a CV < 1%. Phantom variation could result in difficulty meeting minimum CNR specifications for some vendors for a particular phantom. Due to variation in CNR measurements, it is recommended that Siemens units be calibrated to a target SNR or mean glandular dose, as opposed to a target CNR.
Contact Email: