The Shape of the Brain's Connections Is Predictive of Cognitive Performance: An Explainable Machine Learning Study

Abstract

The shape of the brain's white matter connections is relatively unexplored in diffusion magnetic resonance imaging (dMRI) tractography analysis. While it is known that tract shape varies in populations and across the human lifespan, it is unknown if the variability in dMRI tractography-derived shape may relate to the brain's functional variability across individuals. This work explores the potential of leveraging tractography fiber cluster shape measures to predict subject-specific cognitive performance. We implement two machine learning models (1D-CNN and Least Absolute Shrinkage and Selection Operator [LASSO]) to predict individual cognitive performance scores. We study a large-scale database from the Human Connectome Project Young Adult study (n = 1065). We apply an atlas-based fiber cluster parcellation (953 fiber clusters) to the dMRI tractography of each individual. We compute 15 shape, microstructure, and connectivity features for each fiber cluster. Using these features as input, we train a total of 210 models (using fivefold cross-validation) to predict 7 different NIH Toolbox cognitive performance assessments. We apply an explainable AI technique, SHapley Additive exPlanations (SHAP), to assess the importance of each fiber cluster for prediction. Our results demonstrate that fiber cluster shape measures are predictive of individual cognitive performance. The studied shape measures, such as irregularity, diameter, total surface area, volume, and branch volume, are generally as effective for prediction as traditional microstructure and connectivity measures. The 1D-CNN model generally outperforms the LASSO method for prediction. Further interpretation and analysis using SHAP values from the 1D-CNN suggest that fiber clusters with features highly predictive of cognitive ability are widespread throughout the brain, including fiber clusters from the superficial association, deep association, cerebellar, striatal, and projection pathways. This study demonstrates the strong potential of shape descriptors to enhance the study of the brain's white matter and its relationship to cognitive function.