Habeck, C. G. | Search bioRxiv

bioRxiv2022

Persistent Homology-based Functional Connectivity Explains Cognitive Ability: Life-span Study

Brain-segregation attributes in resting-state functional networks have been widely investigated to understand cognition and cognitive aging using various approaches (e.g., average connectivity within/between networks and brain system segregation). While these approaches have assumed that resting-state functional networks operate in a modular structure, a complementary perspective assumes that a core-periphery or rich club structure accounts for brain functions where the hubs are tightly interconnected to each other to allow for integrated processing. We introduce a novel method, persistent homology (PH)-based functional connectivity, to quantify the pattern of information during the integrated processing. We also investigate whether PH-based functional connectivity explains cognitive performance and compare the amount of variability in explaining cognitive performance for three sets of independent variables: (1) PH-based functional connectivity, (2) graph theory-based measures, and (3) brain system segregation. Resting-state functional connectivity data were extracted from 279 healthy participants, and cognitive ability scores were generated in four domains (fluid reasoning, episodic memory, vocabulary, and processing speed). The results first highlight the pattern of brain-information flow over whole brain regions (i.e., integrated processing) accounts for more variance of cognitive abilities than either brain system segregation or the graph theory-based network topology measure. The results also show that fluid reasoning and vocabulary performance significantly decrease as the strength of the additional information flow on functional connectivity with the shortest path increases.

neuroscience↗

bioRxiv2022

Transfer Learning for Cognitive Reserve Quantification

Cognitive reserve (CR) has been introduced to explain individual differences in susceptibility to cognitive or functional impairment in the presence of age or pathology. We developed a deep learning model to quantify the CR as residual variance in memory performance using the structural MRI data from a lifespan healthy cohort. The generalizability of the sMRI-based deep learning model was tested in two independent healthy and Alzheimers cohorts using transfer learning framework. Structural MRIs were collected from three cohorts: 495 healthy adults (initially aged 20-80) from RANN, 620 healthy participants (age 36-100) from lifespan Human Connectome Project Aging (HCPA), and 941 subjects from Alzheimers Disease Neuroimaging Initiative (ADNI). Region of interest (ROI)-specific cortical thickness and volume measures were extracted using the Desikan-Killiany Atlas. Cognitive reserve was quantified by residuals which subtract the predicted memory from the true memory. Cascade neural network (CNN) models were used to train RANN dataset for memory prediction. Transfer learning was applied to transfer the T1 imaging-based model from source domain (using RANN) to the target domain (HCPA or ADNI). The CNN model trained on the RANN dataset exhibited strong linear correlation between true and predicted memory based on the chosen T1 cortical thickness and volume predictors. In addition, the model generated from healthy lifespan data (RANN) was able to generalize to an independent healthy lifespan data (HCPA) and older demented participants (ADNI) across different scanner types. The estimated CR was correlated with CR proxies such education and IQ across all three datasets. The current findings suggest that the transfer learning approach is an effective way to generalize the residual-based CR estimation. It is applicable to various diseases and may flexibly incorporate different imaging modalities such as fMRI and PET, making it a promising tool for scientific and clinical purposes. HighlightO_LIQuantification of cognitive reserve using brain measures for pre-symptomatic Alzheimers patients can be estimated by leveraging lifespan data. C_LIO_LIMulti-center, multi-scanner, multi-sequence can affect the performance of the quantification. C_LIO_LILeveraging lifespan data from a single site can improve the performance. C_LIO_LITransfer learning allows the pre-trained network to successfully reconstruct the dataset acquired from different domains or age groups. C_LI

bioinformatics↗