Classification of Alzheimer's Dementia vs. Healthy subjects by studying structural disparities in fMRI Time-Series of DMN (2407.19990v1)

Abstract: Time series from different regions of interest (ROI) of default mode network (DMN) from Functional Magnetic Resonance Imaging (fMRI) can reveal significant differences between healthy and unhealthy people. Here, we propose the utility of an existing metric quantifying the lack/presence of structure in a signal called, "deviation from stochasticity" (DS) measure to characterize resting-state fMRI time series. The hypothesis is that differences in the level of structure in the time series can lead to discrimination between the subject groups. In this work, an autoencoder-based model is utilized to learn efficient representations of data by training the network to reconstruct its input data. The proposed methodology is applied on fMRI time series of 50 healthy individuals and 50 subjects with Alzheimer's Disease (AD), obtained from publicly available ADNI database. DS measure for healthy fMRI as expected turns out to be different compared to that of AD. Peak classification accuracy of 95% was obtained using Gradient Boosting classifier, using the DS measure applied on 100 subjects.

• The paper presents a novel DS measure to identify structural disparities in fMRI time-series for distinguishing Alzheimer's patients from healthy subjects.
• An autoencoder model reconstructs DMN data to extract key stochastic features, achieving a peak classification accuracy of 95%.
• The findings highlight the potential of DS as a non-invasive biomarker for early diagnosis of Alzheimer's disease.

Classification of Alzheimer's Dementia Using fMRI and Deviation from Stochasticity

The paper "Classification of Alzheimer's Dementia vs. Healthy subjects by studying structural disparities in fMRI Time-Series of DMN" explores an innovative method to differentiate between healthy individuals and those affected by Alzheimer's Disease (AD) using a novel measure called "Deviation from Stochasticity" (DS). This study leverages the potential of functional magnetic resonance imaging (fMRI) time series from the default mode network (DMN) to illuminate patterns indicative of AD, aligning with the ongoing efforts to improve diagnostic tools for this prevalent neurodegenerative disorder.

Methodological Overview

A central proposition of this research is the application of the DS measure to characterize fMRI time series, aiming to identify structural differences in brain activity that distinguish healthy subjects from those with AD. The study utilizes an autoencoder-based model to learn efficient representations of fMRI data, designed to reconstruct the input data and quantify deviations. The dataset, consisting of 50 healthy controls and 50 AD subjects from the ADNI database, was analyzed using this approach.

The DS measure is predicated on assessing the stochasticity in fMRI time series, capturing the level of randomness versus structure in the signal. This is achieved through calculating metrics like the Difference in Reconstruction (DR) and the Density of Peaks measure (DP) for various regions of interest (ROIs) in the DMN. These measures help discern the signal's complexity and variability, which can be indicative of pathological changes.

Key Findings

The study finds significant differences in DS measures between healthy and AD-affected individuals. A peak classification accuracy of 95% was achieved using a Gradient Boosting Classifier. This robust numerical result underscores the potential applicability of the DS measure as a biomarker for diagnosing AD. The clear distinction within the DS values between the two subject groups highlights the technique's efficacy in identifying structural disparities in brain activity.

Implications

The implications of this research are multifold. Practically, it suggests a non-invasive, highly accurate method for early diagnosis and classification of AD, using widely available fMRI technology. Theoretically, it advances the understanding of neural activity's stochastic behavior in the context of neurodegeneration, emphasizing the role of stochasticity as a feature of brain function.

Future Directions

The potential for further development lies in expanding the dataset and refining the autoencoder model to improve classification accuracy and generalizability. Future research could explore integrating this method with other biological markers or imaging modalities to enhance diagnostic precision. Additionally, applying the DS measure to other neurological disorders could uncover broader applications of this technique.

In conclusion, this paper provides a significant contribution to AD classification methodologies by presenting a novel approach based on deviation from stochasticity. The successful implementation and strong results suggest promising applications in both clinical diagnostics and theoretical neuroscience.