Abstract:Dynamic functional connectivity (dFC) analysis of resting-state fMRI data is commonly performed by calculating sliding-window correlations (SWC), followed by k-means clustering in order to assign each window to a given state. Studies using synthetic data have shown that k-means performance is highly dependent on sliding window parameters and signal-to-noise ratio. Additionally, sources of heterogeneity between subjects may affect the accuracy of group-level clustering, thus affecting measurements of dFC state temporal properties such as dwell time and fractional occupancy. This may result in spurious conclusions regarding differences between groups (e.g. when comparing a clinical population to healthy controls). Therefore, is it important to quantify the ability of k-means to estimate dFC state temporal properties when applied to cohorts of multiple subjects, and to explore ways in which clustering performance can be maximised. Here, we explore the use of dimensionality reduction methods prior to clustering in order to map high-dimensional data to a lower dimensional space, providing salient features to the subsequent clustering step. We assess the use of deep autoencoders for dimensionality reduction prior to applying k-means clustering to the encoded data. We compare this deep clustering method to dimensionality reduction using principle component analysis (PCA), uniform manifold approximation and projection (UMAP), as well as applying k-means to the original feature space using either L1 or L2 distance. We provide extensive quantitative evaluation of clustering performance using synthetic datasets, representing data from multiple heterogeneous subjects. In synthetic data we find that deep clustering gives the best performance, while other approaches are often insufficient to capture temporal properties of dFC states. We then demonstrate the application of each method to real-world data from human subjects and show that the choice of dimensionality reduction method has a significant effect on group-level measurements of state temporal properties.

Improved Detection of Time Windows of Brain Responses in Fmri Using Modified Temporal Clustering Analysis.

Iterative Temporal Clustering Analysis for the Detection of Multiple Response Peaks in Fmri

Derivative temporal clustering analysis: detecting prolonged neuronal activity.

Clustered Cnmf for Fmri Data Analysis

Investigation of the Components of Signal Detected by Temporal Clustering Analysis in the Cerebral Functional MRI Data Processing

Fmri Time Series Analysis Based on Stationary Wavelet and Spectrum Analysis

Clustering Analysis for Fmri Dataset Based on ISODATA Algorithm

Clustered Pixels Analysis for Functional MRI Activation Studies of the Human Brain

A Preliminary Functional MRI Study with Temporal Clustering Analysis Fof Localization of Epileptic Activity

Topological Cluster Statistic (TCS): Towards structural-connectivity-guided fMRI cluster enhancement

Comparison of TCA and ICA techniques in fMRI data processing.

Assessment and Optimization of Functional MRI Analyses

Using deep clustering to improve fMRI dynamic functional connectivity analysis

Analysis of dynamic susceptibility contrast MRI time series based on unsupervised clustering methods

Altered Temporal Organization of Brief Spontaneous Brain Activities in Patients with Alzheimer’s Disease

Discovering hidden brain network responses to naturalistic stimuli via tensor component analysis of multi-subject fMRI data

Exploring Functional Connectivity Networks in fMRI Data Using Clustering Analysis.

ICA analysis of fMRI with real-time constraints: an evaluation of fast detection performance as function of algorithms, parameters and a priori conditions

Synchronization-based approach for detecting functional activation of brain.

Determination of the Time Window of Event-Related Potential Using Multiple-Set Consensus Clustering

Analysis of fMRI Time-Series Revisited