Evaluation of non-negative matrix factorization of grey matter in age prediction

Abstract

The relationship between grey matter volume (GMV) patterns and age can be captured by multivariate pattern analysis, allowing prediction of individuals’ age based on structural imaging. Raw data, voxel-wise GMV and non-sparse factorization (with Principal Component Analysis, PCA) show good performance but do not promote relatively localized brain components for post-hoc examinations. Here we evaluated a non-negative matrix factorization (NNMF) approach to provide a reduced, but also interpretable representation of GMV data in age prediction frameworks in healthy and clinical populations. This examination was performed using three datasets: a multi-site cohort of life-span healthy adults, a single site cohort of older adults and clinical samples from the ADNI dataset with healthy subjects, participants with Mild Cognitive Impairment and patients with Alzheimer’s disease (AD) subsamples. T1-weighted images were preprocessed with VBM8 standard settings to compute GMV values after normalization, segmentation and modulation for non-linear transformations only. Non-negative matrix factorization was computed on the GM voxel-wise values for a range of granularities (50 to 690 components) and LASSO (Least Absolute Shrinkage and Selection Operator) regression were used for age prediction. First, we compared the performance of our data compression procedure (i.e., NNMF) to various other approaches (i.e., uncompressed VBM data, PCA-based factorization and parcellation-based compression). We then investigated the impact of the granularity on the accuracy of age prediction, as well as the transferability of the factorization and model generalization across datasets. We finally validated our framework by examining age prediction in ADNI samples. Our results showed that our framework favorably compares with other approaches. They also demonstrated that the NNMF based factorization derived from one dataset could be efficiently applied to compress VBM data of another dataset and that granularities between 300 and 500 components give an optimal representation for age prediction. In addition to the good performance in healthy subjects our framework provided relatively localized brain regions as the features contributing to the prediction, thereby offering further insights into structural changes due to brain aging. Finally, our validation in clinical populations showed that our framework is sensitive to deviance from normal structural variations in pathological aging.

Charting shared developmental trajectories of cortical thickness and structural connectivity in childhood and adolescence

“…Reconstruction error was relatively stable across all component sets, as was age prediction error for all levels except for n = 20. Overall, age prediction error was in range with previous "brain age" estimates (Ball, Adamson, et al, 2017;Cole et al, 2017;Franke, Luders, May, Wilke, & Gaser, 2012) confirming that utility of NMF for providing useful low-rank representations of large imaging data sets (Varikuti et al, 2018).…”

“…Overall, age prediction error was in range with previous "brain age" estimates (Ball, Adamson, et al, 2017;Cole et al, 2017;Franke, Luders, May, Wilke, & Gaser, 2012) confirming that utility of NMF for providing useful low-rank representations of large imaging data sets (Varikuti et al, 2018). We performed NMF using five different levels: 2, 5, 10, 15, and 20 and compared how well the resulting components could reconstruct the original data, and how well the resulting timecourses could be used to predict chronological age.…”

Deep Neural Networks and Kernel Regression Achieve Comparable Accuracies for Functional Connectivity Prediction of Behavior and Demographics

“…For example, many studies utilized either lifespan (Cole et al, 2017;Liem et al, 2017) or developmental (Sturmfels et al, 2018;Nielsen et al, 2019) cohorts, while the UK Biobank comprised older adults (more than 45 years old). Furthermore, many studies preferred to use structural MRI, instead of RSFC, for predicting age (Cole et al, 2017;Sturmfels et al, 2018;Varikuti et al, 2018). Liem and colleagues (2017) achieved MAEs ranging from 5.25 to 5.99 when using RSFC for predicting age in a lifespan dataset comprising 2354 subjects, which was worse than our MAE.…”

“…Like before, prediction accuracies for age, fluid intelligence and pairs matching were evaluated based on the Pearson's correlation between predicted and actual measures across subjects within the test set. Since the age prediction literature often used mean absolute error (MAE) as an evaluation metric (Liem et al, 2017;Cole et al, 2018;Varikuti et al, 2018), we included MAE as an evaluation metric. For completeness, we also computed MAE for pairs matching and fluid intelligence.…”

Personality and local brain structure: their shared genetic basis and reproducibility

“…The Schaefer parcellation is derived using functional MRI data from ~1500 subjects, by integrating local approaches that detect abrupt transitions in functional connectivity patterns and global approaches that cluster similar functional connectivity patterns (Schaefer A et al 2018). Previous research found that the Schaefer parcellation showed convergence with partitions based on structure alone (Varikuti DP et al 2018). A combination of within-area micro circuitry, proxied by brain morphometry, and between-area connectivity enables each area to perform a unique set of computations (Van Essen DC and MF Glasser 2018).…”