MRI Accuracy

[Guest guest]

I remember a few months back the topic about the accuracy of MRI's

and the topic bounced back and forth....not sure if this article is

old, but just a FYI read......

" Neurologists sometimes monitor their patients using periodic MRIs

so they can evaluate disease activity and progression in the form of

new lesions, brain atrophy and lesion volume. The objective and

quantitative nature of MRIs can be helpful in determining status and

course of action. However, as with any measurement technique there

are sources of variability and error, and therefore some of the

change between any two sequential MRIs may be due not to disease

progression but to variability in the MRI process itself. A team of

radiologists set out to estimate this built-in variability to guide

clinicians in evaluating MRI results. They recruited 20 subjects with

MS and scanned each twice within 30 minutes. To simulate normal

conditions, each subject got up after the first scan, left the room,

and re-entered it, and a different technician did the positioning for

the second scan. An automated image analysis technique was used to

measure brain atrophy (brain parenchymal fraction, or BPF) and T2

lesion volume for each scan.

After comparing the measurements between each pair of scans, the

researchers determined that a BPF variation of 0.0056 and a T2 lesion

volume variation of 0.65 mL could safely be attributed to variations

in the scanning and measurement process. The authors note that the T2

variation of 0.65 mL is approximately half of the average yearly

change for people with RRMS and is about equal to the average yearly

change in people with SPMS. These figures are likely a conservative

estimate of inherent measurement process variability because in the

real world, other factors may further increase the level of

variability (including use of different scanners or upgrades of

software versions between scans and use of non-automated or semi-

automated analysis methods). "