Detecting Mild Cognitive Impairment through Excitation-Inhibition Imbalance
Poster No:
196
Submission Type:
Abstract Submission
Authors:
Guoshi Li1, Pew-Thian Yap1
Institutions:
1University of North Carolina at Chapel Hill, Chapel Hill, NC
First Author:
Co-Author:
Introduction:
Alzheimer's disease (AD) is a serious neurodegenerative disorder characterized by progressive cognitive decline and irreversible memory loss. Mild cognitive impairment (MCI) is a prodromal stage of AD that provides a critical window for early intervention. Recent studies indicate the critical role of excitation-inhibition (E-I) imbalance in AD pathology (Vico et al., 2019), raising the possibility of E-I imbalance as a potential early AD biomarker. However, it is unclear whether E-I balance can be used to accurately identify MCI. To address this question, we apply a Large-scale nEural Model Inversion (LEMI) framework to a resting-state functional MRI (fMRI) dataset to estimate regional E-I balance in MCI and normal control (NC) subjects, and use machine learning to classify MCI from NC based on E-I balance.
Methods:
We selected 48 MCIs (31/17 males/females, age: 73.9 ± 10 years) and 48 NCs (26/22 males/females, age: 73.4 ± 6.5 years) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Regional averaged BOLD time series were extracted from 46 regions covering the default mode, salience, executive control and limbic networks based on the Desikan-Killiany atlas (Desikan et al., 2016). In the LEMI framework (Fig. 1) the neural network dynamics is described by a neural mass model (Galán, 2008). Each network node contains two coupled excitatory (E) and inhibitory (I) neural populations and the E neural populations are connected via long-range fibers. First, empirical BOLD signals are Wiener-deconvolved to obtain the composite neural activity y(t). Second, Kalman filter is applied to estimate neural activity x(t) and the predicted error of the composite neural activity is calculated. Lastly, connection parameters (W) are optimized by minimizing the prediction error using a gradient descent algorithm. Regional E-I balance is evaluated by both intra-regional E-I balance (difference between local recurrent excitation and inhibition) and inter-regional E-I balance (difference between the sum of incoming excitatory inter-regional connections and the sum of incoming inhibitory inter-regional connections). Regional E-I balance is compared between NC and MCI using two-sample t-test and multiple comparisons are corrected by FDR. For MCI classification, we employ a support vector machine (SVM) classifier with a Gaussian kernel and select features using chi-squared statistic. The evaluation is performed for ten repetitions of 10-fold SVM classification with randomly partitioned training and testing sets and we report accuracy, sensitivity, specificity, and the area under the ROC curve (AUC). The E-I based classification is compared to classification based on functional connectivity (FC) calculated by Pearson's correlation of BOLD signals.
·Figure 1. The LEMI diagram.
Results:
We observe that intra-regional E-I balance is widely reduced in MCI than NC (Fig. 2A). Among 46 regions, the E-I balance of 29 regions is significantly decreased (p<0.05, FDR corrected), while that of 11 regions is marginally reduced (p<0.05, uncorrected). In contrast, only two regions show marginally significant difference in inter-regional E-I balance (p<0.05, uncorrected; Fig. 2B). Using intra-regional E-I balance as features, SVM achieves an accuracy of 71.8% (sensitivity: 60.0%, specificity: 83.9%, AUC: 74.2%), which is similar to the accuracy using FC features (72.9%). By comparison, using inter-regional E-I balance as features, SVM obtains an accuracy of 68.2% (sensitivity: 68.2%, specificity: 68.3%, AUC: 74.2%), significantly lower than the FC-based accuracy (p<0.05).
·Figure 2. Disruption of E-I balance in MCI.
Conclusions:
We show that intra-regional E-I balance is significantly reduced in MCI, consistent with the progressive disruption of synaptic transmission during AD progression (Sheng et al., 2012). The discriminative ability of E-I balance is confirmed by MCI classification which achieves a relatively high accuracy similar to that of FC-based classification. Our study suggests that E-I balance could be a useful biomarker for early AD diagnosis.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 1
Modeling and Analysis Methods:
Classification and Predictive Modeling 2
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling
Keywords:
Computational Neuroscience
Degenerative Disease
FUNCTIONAL MRI
Machine Learning
Modeling
Other - Excitation-inhibition balance
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Are you Internal Review Board (IRB) certified? Please note: Failure to have IRB, if applicable will lead to automatic rejection of abstract.
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
2. Galán, R.F. (2008). On how network architecture determines the dominant patterns of spontaneous neural activity. PLoS ONE, 3(5), e2148.
3. Sheng, M. et al. (2012). Synapses and Alzheimer's disease. Cold Spring Harb Perspect Biol, 4(5), a005777.
4. Vico, V.E. et al. (2019). Excitatory–inhibitory imbalance in Alzheimer’s disease and therapeutic significance. Neurobiology of Disease 127, 605–615.