NIST logo

Publication Citation: Statistical Analysis of Reader Measurement Variability in Nodule Sizing with CT Phantom Imaging Data

NIST Authors in Bold

Author(s): John Lu; Charles D. Fenimore; Nicholas Petrick; Rongping Zeng; Marios A. Gavrielides; David Clunie; Kristin Borradaile; Robert Ford; Hyun J. Kim; Michael McNitt-Gray; Binsheng Zhao; Andrew Buckler;
Title: Statistical Analysis of Reader Measurement Variability in Nodule Sizing with CT Phantom Imaging Data
Published: November 23, 2012
Abstract: RSNA has conducted a phantom quantitative imaging biomarker (QIBA) study to assess reader measurement variability of both spherical and non-spherical nodules using CT imaging. Statistical analysis of intra-reader and inter-reader variability of volume measurements is performed. We argue that unlike conventional approach in the literature there is a need to examine the measurement performance at each nodule level in terms of shape and size as important covariates before one can aggregate to get the overall performance. We recommend the data analysis strategy based on careful data graphics and meaningful statistics metrics. We think the approach based on careful definition of metrology concepts to be important, not only for this study but also important for QIBA studies on quantitative imaging analysis on objects of vastly different shapes or sizes with severely limited data on experiments with complex factor settings.
Citation: NIST Interagency/Internal Report (NISTIR) - 7879
Keywords: Quantitative imaging as biomarkers, measurement variability, analysis of variance, quantitative image analysis, measurement effect of nodule shape and size.
Research Areas: Statistics, Imaging
DOI: http://dx.doi.org/10.6028/NIST.IR.7879