Detecting type 2 diabetes mellitus cognitive impairment using whole-brain functional connectivity

Area of Science:

• Neuroscience
• Medical Imaging
• Endocrinology

Background:

• Type 2 diabetes mellitus (T2DM) is associated with cognitive decline and brain alterations.
• Resting-state functional magnetic resonance imaging (rs-fMRI) is crucial for diagnosing neurodegenerative conditions like Alzheimer's disease and vascular dementia.
• The utility of rs-fMRI for early diagnosis in T2DM patients with mild cognitive impairment (T2DM-MCI) requires further investigation.

Purpose of the Study:

• To investigate the functional connectivity (FC) patterns in patients with T2DM-MCI.
• To determine if rs-fMRI can aid in the early clinical diagnosis of T2DM-MCI.
• To identify brain regions critical for differentiating T2DM-MCI from T2DM without cognitive impairment (T2DM-NCI) and normal controls (NC).

Main Methods:

• rs-fMRI data were collected from three groups: T2DM-MCI (n=37), T2DM-NCI (n=93), and NC (n=69).
• The XGBoost machine learning model was employed to analyze functional connectivity data.
• Classification accuracy was assessed for T2DM-MCI vs. T2DM-NCI and T2DM-NCI vs. NC.

Main Results:

• The XGBoost model achieved 87.91% accuracy in classifying T2DM-MCI versus T2DM-NCI.
• An accuracy of 80% was obtained in classifying T2DM-NCI versus NC.
• The thalamus, angular gyrus, caudate nucleus, and paracentral lobule were identified as the most significant regions for classification.

Conclusions:

• rs-fMRI analysis reveals distinct functional connectivity patterns in T2DM-MCI patients.
• These findings support the potential of rs-fMRI for early clinical diagnosis and prediction of T2DM-related cognitive impairment.
• The identified brain regions offer a basis for future research and targeted interventions.