Confidence Interval Methods for an Intra-Class Correlation Coefficient: An Application to T1, T2-Mappings in the Heart Using Magnetic Resonance Fingerprinting

• Activity Number: 342 - Novel Statistical Testing and Activation-Detection Methods for Imaging Data
• Type: Contributed
• Date/Time: Tuesday, August 1, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Pingfu Fu* and Shivani Pahwa and Seunghee Margevicius and Vikas Gulani
• Companies: Case Western Reserve University and Case Western Reserve University and Case Western Reserve University and Case Western Reserve University
• Keywords: Confidence interval ; intraclass correlation coefficient ; magnetic resonance fingerprinting ; random effects model ; MOLLI ; reproducibility

Abstract

The intraclass correlation coefficient (ICC) is widely used in biomedical research to assess the reproducibility of measurements between raters, labs, imaging techniques. ICC along with its confidence interval is a commonly reported summary. This study examines the performance of three different methods for constructing an interval in a two-way random effects model: original scale, logit transformation, and bootstrap method. We compare the coverage probabilities and widths of the different interval methods and apply the methods to the experiments on T1, T2-mappings in the heart using magnetic resonance fingerprinting (MRF) and modified look-locker inversion recovery (MOLLI) imaging. While different methods generally perform well and similarly when there are a large number of levels of each factor and data are normal, large differences between the methods emerge when the number of one or more factors is limited. In addition, all methods are shown to lack robustness to certain hard-to-detect violations of normality when the sample size is limited. MRF and MOLLI are in general agreement with moderate/strong ICC across 17-segment heart model.