Internetwork dynamic connectivity effectively differentiates schizophrenic patients from healthy controls

Hui Shen, Zhenfeng Li, Ling‐Li Zeng, Lin Yuan, Fanglin Chen, Zhening Liu, Dewen HuNeuroreport 2014 · DOI

TL;DR

This study introduces a novel approach using dynamic functional connectivity analysis of resting-state fMRI to differentiate individuals with schizophrenia from healthy controls. By examining the temporal fluctuations in brain network interactions, the researchers achieved a high classification accuracy, suggesting that these dynamic connectivity patterns could serve as a valuable biomarker for schizophrenia.

What Problem Does This Paper Address?

Schizophrenia is a complex neurological disorder characterized by disruptions in brain connectivity. While previous neuroimaging studies have identified static abnormalities in functional connectivity networks, they often assumed that these connections remain constant over time. This assumption overlooks the dynamic nature of brain activity, especially during resting states. The paper addresses the gap in understanding how the temporal dynamics of functional connectivity, specifically between different brain networks, might be altered in schizophrenia and whether these dynamic changes can be used to distinguish patients from healthy individuals.

Methods

The researchers employed a sliding window approach to analyze resting-state functional MRI (fMRI) data. This technique allows for the examination of how functional connectivity changes over time. Specifically, they:

  1. Preprocessed fMRI data: Standard preprocessing steps were applied, including motion correction, normalization to a standard brain space, and spatial smoothing.
  2. Calculated Amplitude of Low-Frequency Fluctuations (ALFFs): ALFFs, which represent the amplitude of brain activity within a specific low-frequency band (0.01–0.08 Hz), were measured. This was done on the dynamic functional connectivity matrices derived from the sliding windows.
  3. Constructed Dynamic Networks: A sliding window of a fixed length (40 seconds) was used to calculate functional connectivity between 160 predefined regions of interest (ROIs). This process was repeated across the entire fMRI scan, generating a series of connectivity matrices.
  4. Feature Extraction and Selection: ALFF values from these dynamic connectivity matrices served as features. Two-tailed two-sample t-tests were used to identify connections with significant ALFF differences between patients and controls.
  5. Support Vector Machine (SVM) Classification: A multivariate pattern analysis using SVMs with a polynomial kernel was performed to classify participants. To ensure robust evaluation, a leave-one-out cross-validation (LOOCV) strategy was implemented.

What's novel here is the focus on the temporal dynamics of ALFFs within functional connectivity, rather than just static connectivity measures.

Datasets

The study utilized fMRI data from:

  • 24 patients with schizophrenia recruited from the Second Xiangya Hospital of the Central South University, Changsha, China.
  • 25 healthy controls recruited from Changsha city, China.

All participants were right-handed Chinese speakers. The patients were assessed using the Positive and Negative Syndrome Scale (PANSS) to evaluate symptom severity.

Key Insights and Ideas

  • Dynamic connectivity is key: The study strongly suggests that the temporal variability of functional connections, not just their static strength, is significantly altered in schizophrenia.
  • Internetwork connections are critical: The majority of the abnormal dynamic connections identified were located between different intrinsic connectivity networks (ICNs), highlighting the importance of inter-network communication in the disorder.
  • ALFF as a biomarker: The amplitude of low-frequency fluctuations in dynamic functional connectivity appears to be a sensitive indicator that can effectively differentiate schizophrenic patients from healthy controls.
  • Default Mode Network (DMN) involvement: The DMN showed the highest occurrence of feature weights, indicating its significant role in the abnormal temporal variability observed in schizophrenia.

Key Results

  • High Classification Accuracy: The multivariate pattern analysis using SVMs achieved an overall accuracy of 81.3% in classifying participants (83.3% for patients and 80.0% for healthy controls). This result was statistically significant (P<0.0001).
  • Consensus Features: Through LOOCV, 11 "consensus" connections were consistently identified as discriminative features across all validation folds.
  • Network Contributions: The study found that abnormal dynamic connections were predominantly internetwork, with specific patterns observed between the DMN and other networks (e.g., cingulo-opercular, frontoparietal, cerebellum, sensorimotor).

Limitations and Future Work

The authors acknowledge several limitations:

  • Limited Sample Size: The findings need to be confirmed with a larger sample size due to the relatively small number of participants (n=49).
  • Medication Effects: Some patients were receiving antipsychotic medication, which previous studies suggest can alter brain connectivity. This could potentially influence the observed results.

Future work could involve:

  • Investigating these dynamic connectivity patterns in larger, more diverse cohorts, including unmedicated patients and individuals at different stages of the illness.
  • Exploring the relationship between specific dynamic connectivity alterations and different symptom clusters in schizophrenia.
  • Further validating the potential of these dynamic connectivity measures as reliable biomarkers for diagnosis and treatment monitoring.

Summary updated

Abstract

Increasingly more neuroimaging studies have shown that the complex symptoms of schizophrenia are linked to disrupted neural circuits and dysconnectivity of intrinsic connectivity networks. Previous studies have assumed temporal stationarity of resting-state functional connectivity, whereas temporal dynamics have rarely been explored. Here, we utilized resting-state functional MRI with a sliding window approach to measure the amplitude of low-frequency fluctuations (ALFFs) in functional connectivity in 24 patients with schizophrenia and 25 healthy controls. We found that there were significant differences in the ALFFs of specific connections, the majority of which were located between the intrinsic connectivity networks. Importantly, the experimental results of a multivariate pattern analysis of these ALFF measures showed that 81.3% (P<0.0009) of the participants were correctly classified as either schizophrenic patients or healthy controls by leave-one-out cross-validation. Our results show significant abnormality in the dynamics of internetwork functional connectivity in schizophrenia, which contributes toward the characterization and differentiation of schizophrenic patients, and may be used as a potential biomarker.
BibTeX
@inproceedings{shen2014internetwork,
  author = {Hui Shen and Zhenfeng Li and Ling‐Li Zeng and Lin Yuan and Fanglin Chen and Zhening Liu and Dewen Hu},
  title = {{Internetwork dynamic connectivity effectively differentiates schizophrenic patients from healthy controls}},
  booktitle = {Neuroreport},
  year = {2014},
  doi = {10.1097/wnr.0000000000000267},
}
Figure