Integrating across neuroimaging modalities boosts prediction accuracy of cognitive ability

Variation in cognitive ability arises from subtle differences in underlying neural architecture. Understanding and predicting individual variability in cognition from the differences in brain networks requires harnessing the unique variance captured by different neuroimaging modalities. Here we adopted a multi-level machine learning approach that combines diffusion, functional, and structural MRI data from the Human Connectome Project (N = 1050) to provide unitary prediction models of various cognitive abilities: global cognitive function, fluid intelligence, crystallized intelligence, impulsivity, spatial orientation, verbal episodic memory and sustained attention. Out-of-sample predictions of each cognitive score were first generated using a sparsity-constrained principal component regression on individual neuroimaging modalities. These individual predictions were then aggregated and submitted to a LASSO estimator that removed redundant variability across channels. This stacked prediction led to a significant improvement in accuracy, relative to the best single modality predictions (approximately 1% to more than 3% boost in variance explained), across a majority of the cognitive abilities tested. Further analysis found that diffusion and brain surface properties contribute the most to the predictive power. Our findings establish a lower bound to predict individual differences in cognition using multiple neuroimaging measures of brain architecture, both structural and functional, quantify the relative predictive power of the different imaging modalities, and reveal how each modality provides unique and complementary information about individual differences in cognitive function.

[1]  Paola Galdi,et al.  A distributed brain network predicts general intelligence from resting-state human neuroimaging data , 2018, bioRxiv.

[2]  Christos Davatzikos,et al.  Finding imaging patterns of structural covariance via Non-Negative Matrix Factorization , 2015, NeuroImage.

[3]  Timothy O. Laumann,et al.  Towards Reproducible Brain-Wide Association Studies , 2020, bioRxiv.

[4]  Essa Yacoub,et al.  The WU-Minn Human Connectome Project: An overview , 2013, NeuroImage.

[5]  Gael Varoquaux,et al.  Establishment of Best Practices for Evidence for Prediction: A Review. , 2019, JAMA psychiatry.

[6]  R. Kahn,et al.  The association between brain volume and intelligence is of genetic origin , 2002, Nature Neuroscience.

[7]  H T Siegelmann,et al.  The global landscape of cognition: hierarchical aggregation as an organizational principle of human cortical networks and functions , 2015, Scientific Reports.

[8]  Mert R. Sabuncu,et al.  Global signal regression strengthens association between resting-state functional connectivity and behavior , 2019, NeuroImage.

[9]  T. Verstynen,et al.  Affective brain patterns as multivariate neural correlates of cardiovascular disease risk , 2020, Social cognitive and affective neuroscience.

[10]  Stefan Haufe,et al.  On the interpretation of weight vectors of linear models in multivariate neuroimaging , 2014, NeuroImage.

[11]  Peter C. Hansen,et al.  Neural correlates of intelligence as revealed by fMRI of fluid analogies , 2005, NeuroImage.

[12]  Daniel S. Margulies,et al.  Predicting brain-age from multimodal imaging data captures cognitive impairment , 2016, NeuroImage.

[13]  P. Bonifazi,et al.  Group-Level Progressive Alterations in Brain Connectivity Patterns Revealed by Diffusion-Tensor Brain Networks across Severity Stages in Alzheimer’s Disease , 2017, bioRxiv.

[14]  Dick J. Veltman,et al.  Controlling for effects of confounding variables on machine learning predictions , 2020, bioRxiv.

[15]  R. Nathan Spreng,et al.  Fluid and flexible minds: Intelligence reflects synchrony in the brain’s intrinsic network architecture , 2017, Network Neuroscience.

[16]  R. Cattell Theory of fluid and crystallized intelligence: A critical experiment. , 1963 .

[17]  Kevin Murphy,et al.  Towards a consensus regarding global signal regression for resting state functional connectivity MRI , 2017, NeuroImage.

[18]  Sebastiano Stramaglia,et al.  Structure–function multi‐scale connectomics reveals a major role of the fronto‐striato‐thalamic circuit in brain aging , 2018, Human brain mapping.

[19]  S. Eickhoff,et al.  Empirical examination of the replicability of associations between brain structure and psychological variables , 2018, bioRxiv.

[20]  Jason B. Mattingley,et al.  Functional brain networks related to individual differences in human intelligence at rest , 2016, Scientific Reports.

[21]  Qiang Yang,et al.  A Survey on Transfer Learning , 2010, IEEE Transactions on Knowledge and Data Engineering.

[22]  Robert H. Paul,et al.  The relationship between frontal gray matter volume and cognition varies across the healthy adult lifespan. , 2006, The American journal of geriatric psychiatry : official journal of the American Association for Geriatric Psychiatry.

[23]  Timothy Verstynen,et al.  Integrating across neuroimaging modalities boosts prediction accuracy of cognitive ability , 2020, bioRxiv.

[24]  Timothy D. Verstynen,et al.  Population-averaged atlas of the macroscale human structural connectome and its network topology , 2018, NeuroImage.

[25]  Gaël Varoquaux,et al.  Scikit-learn: Machine Learning in Python , 2011, J. Mach. Learn. Res..

[26]  Nancy C. Andreasen,et al.  Intelligence and regional brain volumes in normal controls , 1997 .

[27]  M. Chun,et al.  Functional connectome fingerprinting: Identifying individuals based on patterns of brain connectivity , 2015, Nature Neuroscience.

[28]  Jesper Andersson,et al.  A multi-modal parcellation of human cerebral cortex , 2016, Nature.

[29]  W. Freeman,et al.  Combining fMRI with EEG and MEG in order to relate patterns of brain activity to cognition. , 2009, International journal of psychophysiology : official journal of the International Organization of Psychophysiology.

[30]  Fang-Cheng Yeh,et al.  Connectometry: A statistical approach harnessing the analytical potential of the local connectome , 2016, NeuroImage.

[31]  Fang-Cheng Yeh,et al.  NTU-90: A high angular resolution brain atlas constructed by q-space diffeomorphic reconstruction , 2011, NeuroImage.

[32]  David K. Yu,et al.  Superficial white matter fiber systems impede detection of long-range cortical connections in diffusion MR tractography , 2015, Proceedings of the National Academy of Sciences.

[33]  J. Shao Linear Model Selection by Cross-validation , 1993 .

[34]  S. Petersen,et al.  Brain Networks and Cognitive Architectures , 2015, Neuron.

[35]  C. Chabris,et al.  Neural mechanisms of general fluid intelligence , 2003, Nature Neuroscience.

[36]  Gaël Varoquaux,et al.  Transmodal Learning of Functional Networks for Alzheimer's Disease Prediction , 2016, IEEE Journal of Selected Topics in Signal Processing.

[37]  R. Kahn,et al.  Efficiency of Functional Brain Networks and Intellectual Performance , 2009, The Journal of Neuroscience.

[38]  Ian J. Deary,et al.  Exploring Possible Neural Mechanisms of Intelligence Differences Using Processing Speed and Working Memory Tasks: An fMRI Study. , 2009 .

[39]  Leo Breiman,et al.  Stacked regressions , 2004, Machine Learning.

[40]  Arthur W Toga,et al.  Relationships between IQ and regional cortical gray matter thickness in healthy adults. , 2007, Cerebral cortex.

[41]  Alessandro Daducci,et al.  Microstructure Informed Tractography: Pitfalls and Open Challenges , 2016, Front. Neurosci..

[42]  Jing Sui,et al.  Neuroimaging-based Individualized Prediction of Cognition and Behavior for Mental Disorders and Health: Methods and Promises , 2020, Biological Psychiatry.

[43]  Rainer Goebel,et al.  Combining multivariate voxel selection and support vector machines for mapping and classification of fMRI spatial patterns , 2008, NeuroImage.

[44]  Dustin Scheinost,et al.  Ten simple rules for predictive modeling of individual differences in neuroimaging , 2019, NeuroImage.

[45]  David H. Wolpert,et al.  Stacked generalization , 1992, Neural Networks.

[46]  Abraham Z. Snyder,et al.  Function in the human connectome: Task-fMRI and individual differences in behavior , 2013, NeuroImage.

[47]  Fang-Cheng Yeh,et al.  Generalized ${ q}$-Sampling Imaging , 2010, IEEE Transactions on Medical Imaging.

[48]  D. Leopold,et al.  Anatomical accuracy of brain connections derived from diffusion MRI tractography is inherently limited , 2014, Proceedings of the National Academy of Sciences.

[49]  M. A. Muñoz,et al.  A novel brain partition highlights the modular skeleton shared by structure and function , 2014, Scientific Reports.

[50]  Stuart J. Ritchie,et al.  Resting-State Connectivity and Its Association With Cognitive Performance, Educational Attainment, and Household Income in the UK Biobank , 2018, Biological psychiatry. Cognitive neuroscience and neuroimaging.

[51]  Jun Li,et al.  Brain Anatomical Network and Intelligence , 2009, NeuroImage.

[52]  Rex E. Jung,et al.  Structural brain variation and general intelligence , 2004, NeuroImage.

[53]  Tor D. Wager,et al.  Common and unique components of response inhibition revealed by fMRI , 2005, NeuroImage.

[54]  Michael W. Cole,et al.  Mapping the human brain's cortical-subcortical functional network organization , 2018, NeuroImage.

[55]  Mark Jenkinson,et al.  The minimal preprocessing pipelines for the Human Connectome Project , 2013, NeuroImage.

[56]  A. Dale,et al.  Fluid and crystallised intelligence are associated with distinct regionalisation patterns of cortical morphology , 2020, bioRxiv.

[57]  Joelle Zimmermann,et al.  Unique Mapping of Structural and Functional Connectivity on Cognition , 2018, The Journal of Neuroscience.

[58]  Alan C. Evans,et al.  Changes in thickness and surface area of the human cortex and their relationship with intelligence. , 2015, Cerebral cortex.

[59]  Wiro J. Niessen,et al.  Structural and diffusion MRI measures of the hippocampus and memory performance , 2012, NeuroImage.

[60]  Stefan Klöppel,et al.  Estimating the age of healthy subjects from T1-weighted MRI scans using kernel methods: Exploring the influence of various parameters , 2010, NeuroImage.

[61]  Fang-Cheng Yeh,et al.  Local connectome phenotypes predict social, health, and cognitive factors , 2017, bioRxiv.