Neuropathic Pain Detection: An EEG-Based Brain Functional Network Approach
Chronic neuropathic pain presents a multifaceted clinical challenge, originating from nerve damage or dysfunction, leading to persistent pain and altered sensory processing in the nervous system. This condition poses formidable hurdles in both diagnosis and treatment, often intertwined with a myriad...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Transactions on Neural Systems and Rehabilitation Engineering |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11059912/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Chronic neuropathic pain presents a multifaceted clinical challenge, originating from nerve damage or dysfunction, leading to persistent pain and altered sensory processing in the nervous system. This condition poses formidable hurdles in both diagnosis and treatment, often intertwined with a myriad of other medical conditions. The complexity of these challenges arises primarily from the absence of definitive biomarkers, necessitating heavy reliance on subjective clinical evaluations. However, this reliance introduces potential biases and significantly complicates the attainment of precise diagnostic outcomes. Given this complex scenario, there is a pressing need for innovative approaches that can provide objective insights into the underlying mechanisms of neuropathic pain (NP). Despite the inherent potential of brain functional networks (BFNs) analysis based on electroencephalogram (EEG) signals to reveal the underlying mechanisms of various nervous system diseases, it has been relatively underexplored in the context of NP. To address these challenges, our study utilizes electroencephalogram (EEG) signals to analyze BFNs through the weighted phase lag index (wPLI), capturing unique connectivity patterns associated with NP. Additionally, we employ a minimum dominating set (MDSet) approach to assess network control, revealing functional disruptions across various stages of NP severity. By extracting persistent homology features from these networks, we developed a support vector machine (SVM)-based classifier that distinguishes control subjects from those experiencing varying degrees of NP. In binary classification between NP and control subjects, the approach achieved a high accuracy of at least 97%. However, accuracy declines when distinguishing between NP groups of different severity and further decreases in multiclass analysis. These findings highlight specific neural signatures underlying NP, contributing to a more comprehensive understanding of the condition and potentially informing targeted therapeutic interventions. |
|---|---|
| ISSN: | 1534-4320 1558-0210 |