Technical guidelines for liver MR elastography, developed under the Quantitative Imaging Biomarkers Alliance and now continued by the Quantitative Medical Imaging Coalition, set out how to acquire, reconstruct, check and analyse liver stiffness measurements so that results are comparable across scanners and sites.

Their central quantitative claim is that, with the standard followed, the within-subject coefficient of variation is about 7%, so a change in liver stiffness of at least 19% is needed to be 95% confident the change is biological rather than measurement variation.

This matters for the growing use of elastography to follow patients with fatty liver disease, viral hepatitis and other chronic liver disease, and for trials using stiffness as an endpoint. A reported fall from 4.0 to 3.5 kPa looks like improvement but sits within the noise. Reports should give the number of slices and region of interest used, and comparisons are most reliable on the same scanner with the same protocol.

Treat a change in liver stiffness of less than about 19% between MR elastography studies as possibly measurement variation.
Repeat follow-up elastography on the same scanner and protocol where possible.
Report the method — slices, regions of interest and quality checks — so the next reader can judge comparability.
Explain to hepatologists that small stiffness changes should not drive treatment decisions on their own.

Why it matters

Serial elastography is increasingly used to judge response, and small apparent changes are easily over-interpreted.

The statistics, in plain English

A within-subject coefficient of variation of 7% means repeat scans in the same person typically differ by about that much. Allowing for variation in both scans and 95% confidence gives a threshold of roughly 2.77 × 7%, about 19%.