Each week, we will be publishing layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.

The article featured today is from the Canadian Journal of Statistics, with the full article now available to read in Early View here.

Wang, Z., Sair, H.I., Crainiceanu, C., Lindquist, M., Landman, B.A., Resnick, S., Vogelstein, J.T. and Caffo, B. (2021), On statistical tests of functional connectome fingerprinting. Can J Statistics. https://doi.org/10.1002/cjs.11591

Since the development of functional neuroimaging techniques, reliability and reproducibility of complex and heavily preprocessed multivariate data has been an issue of key interest. Recently, the concept of functional connectome fingerprinting, which attempts to quantify data reproducibility via a rather direct and intuitive procedure of subject identification by matching repeated functional connectivity measurements, has drawn the attention of the field. The number or percentage of correct matches while subject labels being blinded is usually reported as a statistic, which is then used in the permutation tests. Despite the simplicity and increasing popularity of such procedures, the soundness of the statistical tests, the power, and the factors impacting the test are unstudied.

In this manuscript, the authors investigate the statistical aspects of fingerprinting through the fundamentals of an exchangeability null hypothesis. On one hand, it is found to be a meaningful weak null where (under conditions) the rejection of exchangeability implies the rejection of independence between true signal and measurement, which is an essential feature of reproducible measurement procedures. Based on such formulation, the authors provide a nearly universal Poisson(1) approximation for different matching schemes, which simplifies future implementation. On the other hand, interpretation of the test relies on distributional assumptions on true signals, and therefore precautions should be taken regarding violations, such as covariate effects. A power analysis via both asymptotic approximations and simulation shows that significance can be in an unintended direction, such as clustering due to familial status or demographics. Follow-up studies of covariate impact are recommended. These results can improve the practice of fingerprinting analyses in the future.

The authors perform a numerical study on two functional magnetic resonance imaging (fMRI) resting state datasets, the Human Connectome Project (HCP) and the Baltimore Longitudinal Study of Aging (BLSA). These datasets are instructive, as the HCP includes technical replications of long scans and includes monozygotic and dizygotic twins as well as non-twin siblings. In contrast, the BLSA study incorporates more typical length resting state scans in a longitudinal study. Finally, a study of single regional connections is performed on the HCP data.