Work Package 6 - Toolkit
Research lines initiated by coordinating postdoc
Project 1 - Normative modelling for neuroimaging data
This project provides a novel collaboration between methodological, clinical and applied language research. It combines expertise in experimental design, data acquisition and analytic methodology with clinical expertise and high quality datasets. The case control paradigm has been dominant in many areas of neuroscience, including the study of language. This is useful for detecting group level effects, but is not optimised for detecting patterns of variation in individual subjects and does not accurately capture the clinical decision making process where diseases in individual patients are recognized as deviations from a normal pattern of functioning. Within the context of the toolkit WP of Language in Interaction, a generic analysis approach is developed for understanding and mapping variation in experimental and clinical populations. This approach constitutes a fundamental departure from classical analysis approaches that compare the means between different groups or that seek correlations between experimental variables. Instead, the approach aims to estimate a normative model for phenotypic or biological variation across large experimental cohorts and is analogous to the use of growth charts to map child development in terms of height and weight as a function of age, where deviations from a normal growth trajectory manifest as outliers within the normative range at each age. This enables us to: (i) use data from large cohorts to learn a normative distribution that characterises the study population; (ii) make probabilistic statements about which participants deviate from the normative pattern; (iii) statistically map the brain regions underlying these deviations on a case-by-case basis and (iv) identify outliers in the distribution, who may have an atypical profile with respect to the population.
A conceptual framework for normative modelling has been developed and implemented. This has been applied as proof of concept to a publicly available neuroimaging dataset and has been published in the following paper: - Marquand A.F., Rezek I., Buitelaar J., Beckmann C.F. (2016) Understanding heterogeneity in clinical cohorts using normative models: beyond case control studies. Biological Psychiatry, 80:552-61
A set of software tools implementing this approach are under development. A beta version of these tools are freely distributed under an open source license and released on GitHub for use by the scientific community. The normative modelling approach also forms an important basis for ‘Big’ question 4 (BQ4): Variability in language processing and in language learning. The details and time frame for this project are currently under discussion within the remit of BQ4 but at the very least, the normative modelling tools will be instrumental in mapping the neurobiological underpinnings of variability in language abilities across a young, healthy population. Dr. Marquand has participated in all BQ4 meetings to date and continues to contribute according to the evolving needs of the working group.
The value of the normative modelling approach in clinical conditions (including those relevant to language) has also been surveyed in the following publication:
Marquand, A, Wolfers, T., Mennes, M., Buitelaar, J and Beckmann, C. (2016) Beyond Lumping and Splitting: A Review of Computational Approaches for Stratifying Psychiatric Disorders. Biological Psychiatry Cogn. Clin. Neuro. 1: 433-447
In addition to the work on normative modelling, Dr. Marquand has collaborated with other members of the LiI consortium on two publications investigating structural asymmetries in the human brain. These publications are:
Tulya Kavaklioglu, Tulio Guadalupe, Marcel Zwiers, Andre F. Marquand, Marten Onnink, Elena Shumskaya, Han Brunner, Guillen Fernandez, Simon E. Fisher, Clyde Francks (2017) Structural asymmetries of the human cerebellum in relation to cerebral cortical asymmetries and handedness. Brain Structure and Function doi: 10.1007/s00429-016-1295-9.
Guadalupe T, et al (2016) Human subcortical brain asymmetries in 15,847 people worldwide reveal effects of age and sex. Brain Imaging and Behaviour (in press).