Aoife Mary McErlean MBBCh, FFRRCSI, Presenter: Nothing to Disclose

David M. Panicek MD, Abstract Co-Author: Nothing to Disclose

Emily Craig Zabor, Abstract Co-Author: Nothing to Disclose

Chaya Moskowitz, Abstract Co-Author: Nothing to Disclose

Richard Bitar MD, Abstract Co-Author: Nothing to Disclose

Hedvig Hricak MD, PhD, Abstract Co-Author: Nothing to Disclose

Michelle Segall Ginsberg MD, Abstract Co-Author: Nothing to Disclose

The purpose of this study is to determine the intra- and interobserver variability in measurements of lesions of various sizes and morphology at computed tomography (CT) and to calculate error rates.

The institutional review board approved this study. Lymph nodes, pulmonary and hepatic lesions, stratified by size and sharpness of contour, were selected. Bidimensional orthogonal axial measurements of each lesion were independently performed by 17 blinded radiologists with varying levels of experience. Repeat measurements were performed by each reader at an interval of at least 2 weeks. Intra-Class Correlation Coefficients and Bland-Altman plots were used to assess intra- and inter observer variability.

The 320 measured lesions had median dimension of 14 mm (range, 2-126 mm) for long axis and 10 mm (range, 2-95 mm) for short axis. The overall intraobserver agreement was 0.957 (95% CI: 0.947, 0.966) and 0.945 (95% CI: 0.933, 0.955), and interobserver agreement was 0.954 (95% CI: 0.943, 0.963) and 0.941(95% CI: 0.929, 0.951) for long- and short-axis measurements, respectively. Both intra- and interobserver agreement differed by lesion size and reader experience. Interobserver variability was 0.928 (95% CI: 0.903, 0.946) for fellows, compared to 0.961(95% CI: 0.949, 0.969) for attendings in short-axis measurements. Across all readers, agreement was highest for lung lesions and lowest for lymph nodes. Lesions with a smooth rather than spiculated contour had less variability in measurements. Measurement error decreased with increasing lesion size; the 95% limits of agreement for short-axis measurements were within ±7% for lesions <10 mm versus ±5% for lesions ≥20 mm.  

Overall intra- and interobserver variability rates are similar; therefore, in clinical practice, serial CT measurements can be safely performed by different readers. Smooth contour, larger lesion size, and increased reader experience improve measurement consistency. For subcentimeter lesions, a change of >7% in short-axis dimension can be considered a true change rather than a measurement error.

Intra- and interobserver variability rates are similar for lesions of different sizes and contour; therefore, in clinical practice, serial CT measurements can be safely performed by different readers

McErlean, A, Panicek, D, Zabor, E, Moskowitz, C, Bitar, R, Hricak, H, Ginsberg, M, Intra- and Interobserver Variability in Lesion Measurements on Computed Tomography.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12026368.html