Hierarchical Bayesian Regression for Multi-Site Normative Modeling of Neuroimaging Data

@article{Kia2020HierarchicalBR,
  title={Hierarchical Bayesian Regression for Multi-Site Normative Modeling of Neuroimaging Data},
  author={Seyed Mostafa Kia and Hester Huijsdens and Richard Dinga and Thomas Wolfers and Maarten Mennes and Ole Andreas Andreassen and Lars Tjelta Westlye and Christian F. Beckmann and Andre F. Marquand},
  journal={ArXiv},
  year={2020},
  volume={abs/2005.12055}
}
Clinical neuroimaging has recently witnessed explosive growth in data availability which brings studying heterogeneity in clinical cohorts to the spotlight. Normative modeling is an emerging statistical tool for achieving this objective. However, its application remains technically challenging due to difficulties in properly dealing with nuisance variation, for example due to variability in image acquisition devices. Here, in a fully probabilistic framework, we propose an application of… 

Federated Multi-Site Normative Modeling using Hierarchical Bayesian Regression

TLDR
A federated probabilistic framework using hierarchical Bayesian regression (HBR) for multi-site normative modeling that closes the technical loop for applying normative modeling across multiple sites in a decentralized manner and will facilitate applications of normative modeling as a medical tool for screening the biological deviations in individuals affected by complex illnesses such as mental disorders.

Modeling Multi-Site Neuroimaging Data Using Hierarchical Bayesian Neural Networks to Study Structural Brain Variability

  • Computer Science
  • 2020
TLDR
The model is able to model the mean and variance in datasets, also when there is non-linearity in the mean or variance of the data, or when the variance follows a skew Gaussian distribution and the model offers additional flexibility to handle site-related variation and non- linearity in neuroimaging datasets.

Accommodating site variation in neuroimaging data using normative and hierarchical Bayesian models

TLDR
The results show the value of hierarchical Bayesian regression methods for accommodating site variation in neuroimaging data, which provides an alternative to harmonization techniques.

Normative modeling of neuroimaging data using generalized additive models of location scale and shape

TLDR
GAMLSS is presented, a flexible modeling framework that can model heteroskedasticity, non-linear effects of variables, and hierarchical structure of the data, and it allows for an automatic model order selection, thus improving the accuracy of normative models while mitigating problems of overfitting.

Explanatory latent representation of heterogeneous spatial maps of task-fMRI in large-scale datasets

TLDR
A data-driven approach to learn interpretable and generalizable latent representations that link cognition with underlying brain systems; a three-dimensional autoencoder is applied to two large-scale datasets to find an interpretable latent representation of high dimensional task fMRI image data.

Learning Robust Hierarchical Patterns of Human Brain across Many fMRI Studies

TLDR
This paper introduces a simple yet effective matrix factorization based formulation to reduce site-related effects while preserving biologically relevant variations in fMRI data acquired on multiple sites and leverages adversarial learning in the unsupervised regime to improve the reproducibility of the components.

Revealing Individual Neuroanatomical Heterogeneity in Alzheimer's Disease

TLDR
This work aimed to characterise individual differences and outliers in cortical thickness in patients with Alzheimers disease, people with mild cognitive impairment and cognitively normal controls, and assessed the relationships between cortical thickness heterogeneity and cognitive function, amyloid-beta, tau, ApoE genotype.

References

SHOWING 1-10 OF 31 REFERENCES

Harmonization of cortical thickness measurements across scanners and sites

Harmonization of resting-state functional MRI data across multiple imaging sites via the separation of site differences into sampling bias and measurement bias

TLDR
It is essential to collect large neuroimaging data from as many sites as possible to appropriately estimate the distribution of the grand population, and a novel harmonization method was developed that removed only the measurement bias by using traveling-subject dataset.

Harmonization of multi-site diffusion tensor imaging data

Detecting and harmonizing scanner differences in the ABCD study - annual release 1.0

TLDR
It is argued that accounting for scanner variance, including even minor differences in scanner hardware or software, is crucial for any analysis and can be harmonized using an empirical Bayes approach known as ComBat.

Dissecting the Heterogeneous Cortical Anatomy of Autism Spectrum Disorder Using Normative Models

Toward discovery science of human brain function

TLDR
A universal architecture of positive and negative functional connections, as well as consistent loci of inter-individual variability are demonstrated, which demonstrate that independent R-fMRI datasets can be aggregated and shared.

Data sharing in neuroimaging research

TLDR
The INCF Task Force on Neuroimaging Datasharing, in conjunction with several collaborative groups around the world, has started work on several tools to ease and eventually automate the practice of data sharing, hoped that such tools will allow researchers to easily share raw, processed, and derived neuroimaging data, with appropriate metadata and provenance records, and will improve the reproducibility of neuroim imaging studies.

The ENIGMA Consortium: large-scale collaborative analyses of neuroimaging and genetic data

TLDR
The ENIGMA Consortium has detected factors that affect the brain that no individual site could detect on its own, and that require larger numbers of subjects than any individual neuroimaging study has currently collected.