Akademik Veri Yönetim Sistemi
2nd Joint German-Turkish Symposium on Human Neuroscience, Ankara, Türkiye, 7 - 09 Eylül 2023
Objective: Migraine headache is frequently misdiagnosed in clinical settings. Today, there is no method (blood test, cerebrospinal fluid, neuroimaging, etc.) that provides precise and accurate results in the clinical diagnosis of migraine. The main aim of the study is to be able to classify individuals into three groups: healthy controls, migraine with aura, and migraine without aura by using certain machine learning approaches.
Materials and Methods: The changes in the prefrontal oxy-hemoglobin (HbO) concentrations were measured by a 16-channel functional near-infrared spectroscopy (fNIRS) device during the Victoria Stroop task. Features were extracted by analyzing differences in HbO concentrations between different stages of the Stroop task according to the time domain. Subsequently, feature selection methods were employed to identify the most influential features in the classification process.
Results: Through the application of machine learning techniques, Support Vector Machine (SVM) algorithm achieved an accuracy of 78% based on preliminary analysis, successfully classifying healthy controls, migraine with aura, and migraine without aura.
Conclusion: This study revealed the potential role of fNIRS-based prefrontal oxygenation parameters during a cognitive task to differentiate healthy controls, migraine with aura, and migraine without aura. The findings indicated that combining prefrontal oxygenation patterns obtained through fNIRS with extracted features and machine learning techniques is effective for classifying individuals with different migraine conditions. This research may contribute to our understanding of the hemodynamic correlates of migraines and could pave the way for the development of objective diagnostic methods in the future.