Sheela Agarwal MD, MS, Presenter: Nothing to Disclose

Tarik K. Alkasab MD, PhD, Abstract Co-Author: Nothing to Disclose

Peter F. Hahn MD, PhD, Abstract Co-Author: Stockholder, Abbott Laboratories Stockholder, Covidien AG Stockholder, CVS Caremark Corporation Stockholder, Kimberly-Clark Corporation Stockholder, Landauer, Inc

Sanjay Saini MD, Abstract Co-Author: Nothing to Disclose

Mukesh Gobind Harisinghani MD, Abstract Co-Author: Nothing to Disclose

Debra Ann Gervais MD, Abstract Co-Author: Research Grant, Covidien AG

Because hepatobiliary contrast liver MRI offers special challenges, these studies undergo enhanced peer-review. We evaluated the effectiveness of two randomized peer-review methods compared to double reading for detection and characterization of liver lesions and incidental findings.

A single expert in gadoxetate-enhanced MRI double read the 544 studies performed during a 6 month period. Rates of change in diagnosis served as the historical control. Each study had been read initially by a fellowship trained abdominal radiologist before expert over-reading. During the same 6 month period in the following year, 798 liver MRIs were performed with gadoxetate disodium. These examinations were randomly sampled for a total of 2% of all cases, or 16 cases, which were then over-read (Arm 1). During this same time frame, all gadoxetate MRIs that were randomly selected during the department-wide peer-review system were evaluated, as part of the ACR recommended target rate of 2% of all examinations per radiologist (Arm 2).

Using a double read paradigm, changes in interpretation occurred on 50/544 examinations (9.2%) with 23 (4.2%) leading to a potential change in clinical management. All 28 readers were evenly sampled (100%). Using the 2% paradigm the following year (Arm 1), changes in interpretation occurred on 3 (19%) examinations, with 2 (12.5%) leading to potential change in management. 12 (39%) radiologists were sampled using this method. During the department-wide 2% peer-review process, 53 abdominal MRIs were reviewed, of which 6 (0.75%) were gadoxetate-enhanced liver MRI (Arm 2). One (17%) led to a change in interpretation, which was not clinically significant, and 6 (21%) radiologists were sampled.

Department wide peer-review, which randomly selects cases from all abdominal MRIs performed, under-sampled gadoxetate MRI and radiologists, therefore capturing a lower rate of error compared to the double-read paradigm. Instead, specifically sub-selecting two percent of all gadoxetate studies leads to a significantly more accurate reflection of misinterpretation rates.

Double reading of gadoxetate MRI results in clinically significant improvement in read accuracy, though this approach is resource intensive and thus impractical in a busy hospital setting. Accurate methodologies for sampling are essential to capture ongoing interpretive challenges with gadoxetate MRI.

Agarwal, S, Alkasab, T, Hahn, P, Saini, S, Harisinghani, M, Gervais, D, Peer-Review Sampling: Comparison of Two Paradigms Against Double Reading of Gadoxetate-Enhanced MRI.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14011937.html