Reducing the Number of Measurements in Liver Point Shear-Wave Elastography: Factors that Influence the Number and Reliability of Measurements in Assessment of Liver Fibrosis in Clinical Practice

Abstract

Purpose To identify the minimum number of measurements required for the noninvasive assessment of liver fibrosis by using point shear-wave elastography (pSWE) and determine whether the use of a reliability indicator such as interquartile range [IQR]-to-median ratio will affect diagnostic performance. Materials and Methods Ten liver shear-wave velocity (SWV) measurements by pSWE were obtained in 232 participants. Interclass correlation coefficients (ICC) between the median of the first two through the first nine measurements and all 10 measurements were calculated; the minimum number of measurements with ICC greater than 0.95 versus all 10 measurements was determined. The diagnostic performance of the minimum number of measurements and 10 measurements in identifying significant (Ishak stage, ≥3) and severe fibrosis or cirrhosis (Ishak stage, ≥5) was compared by using areas under the receiver operating characteristic curve. These were compared between measurements that demonstrated higher or lower reliability (IQR-to-median ratio of ≤ 30% and IQR-to-median ratio of > 30%, respectively). Results Compared with 10 measurements, a minimum of six SWV measurements was required. The overall area under the curve for diagnosing significant (areas under the receiver operating characteristic curve, 0.828 vs 0.839; P = .487) and severe fibrosis or cirrhosis (0.953 vs 0.969, respectively; P = .145) did not differ according to number of measurements (six vs 10); a median of six measurements resulted in only limited disagreement (nine of 232 [3.9%]) versus histologic evaluation. When using 10 measurements, higher reliability measurements showed a lower percentage of discordance between pSWE and significant fibrosis and severe fibrosis or cirrhosis (22 [14.7%] and three [2.0%] of 150 cases, respectively) compared with lower reliability measurements (26 [31.7%] and eight [9.8%] of 82 cases, respectively). Significant fibrosis was an independent predictor for lower reliability (hazard ratio, 2.22; P < .020). Conclusion A limited number of SWV measurements (median six vs median 10) were required for the assessment of liver fibrosis by using pSWE. The number of measurements had less influence on the diagnostic accuracy compared with lower reliability measurements. RSNA, 2018 Online supplemental material is available for this article.

Impact of contouring variability on oncological PET radiomics features in the lung

“…This analysis was performed for each of the 25 radiomics features being investigated. Selection of radiomics features least impacted by contouring variability was based on the lower confidence interval of ICC (ICC LB ) for all 10 raters and a threshold of ≥0.950 (i.e., an error of less than 5% for worst case scenario) was used as the cutoff value 33 . The impact of decreasing the number of raters for the top performing features was explored by employing a leave-p-out analysis 34 .…”

General advice in ultrasound based elastography of pediatric patients

“…In the available literature, the number of acquisitions taken in children varies from 1 to 24 [11]. Less than 10 measurements may be required in future studies as excellent reproducibility and accuracy have been reported with a lower number of acquisitions [30]. It has been reported that the number of acquisitions may have little influence on the diagnostic accuracy compared to the variability of the measurements [30].…”

“…Less than 10 measurements may be required in future studies as excellent reproducibility and accuracy have been reported with a lower number of acquisitions [30]. It has been reported that the number of acquisitions may have little influence on the diagnostic accuracy compared to the variability of the measurements [30]. Therefore, in pediatric studies, the number of measurements taken may be based on the child's age and the cooperation during the examination.…”

Different techniques for ultrasound liver elastography

“…On practical grounds pSWE techniques tend to require a higher number of valid measurements to achieve an accurate median stiffness value (recommendation is 6 measurements or even more if reliability criteria are not fully met). 9,10 This is expected to be related to the fact that they sample a smaller portion of tissue and are therefore more prone to suffer the heterogenous involvement of the liver in case of significant fibrosis (since the sampled portion may fall in a regenerative nodule as well as in a large scar). On the other hand, bidimensional SWE, is often accurate after only 35 measurements, but may be more prone to subjective judgement in all those technologies in which the location of the Region of Interest (ROI) within the stiffness box is freely chosen by the operator.…”