EEG-Based MDD Diagnosis Using Feature Extraction and Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing major depressive disorder, such as fMRI, are limited by high costs, spatial and temporal restrictions, and do not adequately consider altered cognitive processes, making accurate diagnosis challenging due to similar symptoms with other disorders and varying severity.
Innovation Solution
A method and device using brain wave data from selected electrode channels and frequency bands to extract features like power spectrum densities, functional connectivity, and network indices, applying a classification model to determine the presence of major depressive disorder with high accuracy, reducing the number of channels to prevent overfitting and improve computational efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fMRI is used for diagnosis, then neural activity can be visualized, but the system becomes expensive and has spatial and temporal restrictions
Solution Approach 1:
The patent replaces the complex fMRI system with a simpler EEG-based system. Instead of using magnetic resonance imaging to visualize neural activity, the invention uses electroencephalography to detect brain wave patterns, thereby reducing system complexity and cost while maintaining diagnostic capability through different measurement approaches
Solution Approach 2:
The patent creates a simplified copy of the fMRI diagnostic function using EEG technology. Rather than directly measuring neural activity with complex fMRI equipment, the system uses brain wave data as a proxy indicator that can be captured with simpler, more affordable EEG devices, achieving similar diagnostic purposes through a less complex measurement system
2Measurement precision
If fMRI is used for diagnosis, then neural activity is captured, but temporal resolution is limited and cognitive processes are not adequately considered
Solution Approach 1:
The patent substitutes fMRI's spatial focus with EEG's temporal focus. While fMRI provides good spatial resolution for neural activity localization, it has poor temporal resolution. The EEG system compensates by measuring brain wave oscillations that occur in real-time, capturing temporal dynamics of cognitive processes as they unfold, thereby resolving the temporal resolution limitation of fMRI
3Measurement precision
If multiple brain wave channels are used, then diagnostic accuracy improves, but computational complexity and overfitting risk increase
Solution Approach 1:
The patent extracts and selects only the most informative brain wave channels from the full set of available channels. Instead of using all channels which would increase computational complexity and overfitting risk, the system identifies and extracts a subset of channels that provide sufficient diagnostic information, thereby reducing system complexity while maintaining accuracy
Solution Approach 2:
The patent changes the parameter of channel number from maximum possible to an optimized subset. By adjusting this parameter based on diagnostic effectiveness and computational efficiency considerations, the system finds the optimal balance between accuracy and complexity, preventing overfitting while maintaining sufficient diagnostic capability
Data Source
AI summary
The present invention provides a method for providing information on a major depressive disorder implemented by a processor. Provided are a method for providing information on a major depressive disorder and a device using the same, the method comprising the steps of: receiving brain wave data of a subject; extracting feature data of at least one of power spectrum densities (PSDs), a functional connectivity, and a network index with respect to the brain wave data; and determining whether the subject has a major depressive disorder on the basis of at least one feature data, by using a classification model trained to output whether a subject has a major depressive disorder on the basis of the at least one feature data as an input, wherein the subject is a subject suspected of suffering from a major depressive disorder without having a history of drug use.