Program Information
Quantitative Ultrasound Estimates in Three Liver-Mimicking Phantoms Are Repeatable and Reproducible Across Five Operators, Three Clinical Scanners and Multiple Transducers
M Andre1,2* , A Han3 , D Chen2 , J Cui2 , W Henderson2 , E Sy2 , C Sirlin2 , R Loomba2 , W OBrien3 , J Erdman3 , (1) VA Medical Center, San Diego, CA, (2) University of California, San Diego, La Jolla, California, (3) University of Illinois at Urbana-Champaign, Urbana, Illinois
Presentations
WE-DE-708-7 (Wednesday, August 2, 2017) 10:15 AM - 12:15 PM Room: 708
Purpose: Assess reproducibility and repeatability (R&R) of multiple quantitative ultrasound (QUS) parameters for five operators and two clinical scanners with multiple transducers in three phantoms mimicking a wide range of liver tissue acoustic properties.
Methods: RF data were acquired on GE Logiq e9 with C1-5 transducer and two Siemens S3000 with 4C1 and 6C1HD transducers from three well-characterized phantoms having backscatter and attenuation coefficients (BSC, AC) and sound speed comparable to human in vivo liver tissue for normal, mild and high fat content as determined in our previous study. Five operators each acquired 5 frames of RF separated by several seconds for each combination of phantom, scanner and transducer. AC and BSC spectra were estimated using a reference phantom method over the respective system bandwidths. Mean AC and log BSC values were averaged over 2.8-3.0 MHz. R&R was estimated for each phantom using unweighted sums of squares ANOVA by applying a two-factor random-e effects model for operator and system (scanner-transducer combination).
Results: Mean AC values for the phantoms were 0.68, 0.35 and 0.54 dB/cm-MHz, and mean BSC were –34.0, –46.1, and –19.4 dB, respectively. Estimated total measurement variability (standard deviation, SD) was 0.025-0.029 dB/cm-MHz for AC, and 0.62-1.04 dB for BSC. Total variability was decomposed into repeatability (variability under same operator, same system) and reproducibility (variability due to operator and/or system effects). The estimated repeatability SD was 0.019-0.021 dB/cm-MHz for AC and 0.35-0.43 dB for BSC. Reproducibility SD was 0.015-0.020 dB/cm-MHz for AC and 0.49-0.95 dB for BSC. The operator effect was minimal and reproducibility variability, also low, was mainly due to system effect. Overall measurement variability was small compared to the AC and BSC range of the three phantoms.
Conclusion: QUS estimates in three liver-mimicking phantoms are repeatable and reproducible across five operators, two clinical scanners and multiple transducers.
Funding Support, Disclosures, and Conflict of Interest: National Institutes of Health R01DK106419, Research Agreement, Siemens Medical Systems, Research Agreement, GE Medical Systems
Contact Email: