VarCoNet: A variability-aware self-supervised framework for functional connectome extraction from resting-state fMRI

TL;DR

VarCoNet, an enhanced self-supervised framework for robust functional connectome (FC) extraction from resting-state fMRI (rs-fMRI) data, is introduced, providing a versatile and robust framework for FC analysis in rs-fMRI.

Abstract

Accounting for inter-individual variability in brain function is key to precision medicine. Here, by considering functional inter-individual variability as meaningful data rather than noise, we introduce VarCoNet, an enhanced self-supervised framework for robust functional connectome (FC) extraction from resting-state fMRI (rs-fMRI) data. VarCoNet employs self-supervised contrastive learning to exploit inherent functional inter-individual variability, serving as a brain function encoder that generates FC embeddings readily applicable to downstream tasks even in the absence of labeled data. Contrastive learning is facilitated by a novel augmentation strategy based on segmenting rs-fMRI signals. At its core, VarCoNet integrates a 1D-CNN-Transformer encoder for advanced time-series processing, enhanced with a robust Bayesian hyperparameter optimization. Our VarCoNet framework is evaluated on two downstream tasks: (i) subject fingerprinting, using rs-fMRI data from the Human Connectome Project, and (ii) autism spectrum disorder (ASD) classification, using rs-fMRI data from the ABIDE I and ABIDE II datasets. Using different brain parcellations, our extensive testing against state-of-the-art methods, including 13 deep learning methods, demonstrates VarCoNet's superiority, robustness, interpretability, and generalizability. Overall, VarCoNet provides a versatile and robust framework for FC analysis in rs-fMRI.

Figures (11)

• Figure 1: A hypothetical example illustrating the importance of reducing the ratio between intra- to inter-individual variability. The upper panel shows the computation of FCs from rs-fMRI data (subject 1, session 1) using PCC and VarCoNet, including parcellation, time-series extraction, and FC calculation. With a 4-region atlas, each FC yields six distinct values ($(4\times(4-1))/2$), which can be represented as a point in a 6-D scatter plot. For simplicity, FCs are shown as points in a reduced 2-D space. The lower panel compares scatter plots of PCC-based (left) and VarCoNet-based (right) FCs. Colors denote subjects, while multiple points per color indicate sessions. Shaded areas reflect intra-individual variability; distances between shaded areas reflect inter-individual variability. Assuming subjects 1–4 are neurotypicals and subjects 5–8 have ASD, the figure highlights how accounting for variability facilitates clearer decision boundaries.
• Figure 2: Block diagram of VarCoNet's contrastive training. For illustration purposes, a batch of two subjects is shown. After parcellation using one of two atlases, each rs-fMRI signal is augmented to generate a pair of views. These views are processed by a shared 1D-CNN-Transformer encoder, producing embeddings of size $R \times L$, where $R$ is the number of ROIs and $L$ denotes the number of tokens. Cosine similarity is then computed among the ROIs to obtain the VarCoNet-based FCs as $R \times R$ matrices. These matrices are vectorized by discarding the lower triangular part and undergo a contrastive process that favors similarity between FCs of the same subject while minimizing similarity between FCs of different subjects. The lower part of the figure provides a detailed view of embedding computation. Specifically, the 1D-CNN slides over each ROI’s time-series of length $T$, extracting $L$ tokens of shape $R \times K$, where $K$, the number of kernels. Global average pooling reduces each token to shape $R \times 1$. Positional encodings are then added before the Transformer processes the token sequence.
• Figure 3: Block diagram of the subject fingerprinting-based objective function used in Bayesian optimization. Given a set of hyperparameters proposed by the optimizer, VarCoNet is initialized and trained on the training data. At each epoch, rs-fMRI signals from the validation set ($N_{val} = 200$ subjects) are segmented using sliding windows of lengths $L_{w1}$, $L_{w2}$, and $L_{w3}$ (10 segments per length). VarCoNet extracts vectorized FC embeddings of shape $1 \times R(R-1)/2$, where $R$ is the number of ROIs. Subject fingerprinting is then performed across all six segment-length combinations ($L_{w1}$-$L_{w1}$, $L_{w1}$-$L_{w2}$, ..., $L_{w3}$-$L_{w3}$) using data from both sessions. Each combination yields 10 fingerprinting scores, which are averaged to obtain six final scores. The objective function is computed as the harmonic mean of the average and minimum of these six scores. The highest objective function value across all training epochs is used to update the optimizer.
• Figure 4: Hyperparameter values for each Bayesian optimization trial and the corresponding objective function value. Each of the 125 runs, represented by a curved line, corresponds to a unique combination of $FF_{dim}$, $N_{heads}$, $N_{layers}$, batch size, learning rate ($lr$), and temperature parameter ($\tau$).
• Figure 5: Contrastive training loss of VarCoNet trained on 930 rs-fMRI scans from the HCP dataset, using the AAL3 (a) and AICHA (b) atlases for parcellation.