Migraine detection from EEG signals using tunable Q-factor wavelet transform and ensemble learning techniques
1Technical Sciences Vocational School, Gaziantep University, 27310, Gaziantep, Turkey. zulfikaraslan27@gmail.com.
Physical and Engineering Sciences in Medicine
|September 10, 2021
Summary
This study introduces a computer-aided diagnosis system using electroencephalogram (EEG) signals to detect migraine. The system achieved 89.6% accuracy, offering a valuable tool for migraine diagnosis support.
Area of Science:
- Neuroscience
- Medical Informatics
Background:
- Migraine is a debilitating neurovascular disease with significant impact.
- Accurate and timely diagnosis of migraine is crucial for effective management.
Purpose of the Study:
- To develop and evaluate a computer-aided diagnosis system for migraine detection using EEG signals.
- To analyze the oscillatory structure of EEG signals for diagnostic feature extraction.
Main Methods:
- Electroencephalogram (EEG) signals were analyzed using a tunable Q-factor wavelet transform (TQWT).
- EEG signals were decomposed into sub-bands, and statistical features were extracted.
- Kruskal Wallis test was used to assess feature significance; ensemble learning techniques, including Rotation Forest, were employed for classification.
Main Results:
- The proposed TQWT-based method successfully extracted distinguishing features from EEG sub-bands.
- Rotation Forest algorithm achieved the highest classification performance of 89.6% using features from Sub band 2.
- The study demonstrated the effectiveness of EEG signal analysis in differentiating migraine patients from healthy controls.
Conclusions:
- The developed computer-aided diagnosis system shows significant potential in supporting expert migraine diagnosis.
- EEG signal analysis, particularly with TQWT and ensemble learning, offers a promising avenue for objective migraine assessment.
