無線腦波習慣化頻譜差異於開發偏頭痛分期系統

Abstract

偏頭痛是一種常見的神經性疾病會引起患者頭部某一個區域激烈悸動或是脈搏感覺的抽痛，並伴隨著噁心、嘔吐、極度畏光及畏聲的特性。偏頭痛是一個反覆發作且發作時使患者幾乎喪失行動能力。目前偏頭痛照護的標準是依照藥物治療，然而藥效會隨著偏頭痛的發生時間降低。預測偏頭痛的發生有機會使患者進行及早治療。過去研究指出習慣化指標會隨著偏頭痛週期變化而有不同的差異現象。此篇研究透過穩態視覺誘發電位實驗收集腦波資料並搭配由典型相關分析改良所開發出特徵擷取演算法，誘發並計算出習慣化指標，搭配無線腦機介面於偏頭痛周期中使用習慣化指標分類患者的頭痛週期，並以此指標建立偵測偏頭痛狀態的分類系統。在15Hz的視覺刺激下O1及O2通道的結果可以看到頭痛間期的病患表現出去習慣化的現象且頭痛前期的病患則表現出正規化的習慣化現象。頭痛分類系統在分四個時期的偏頭痛患者的情況下可以達到接近80%的準確率，在分兩個時期的偏頭痛患者(頭痛前期及頭痛間期)的情況下更可以達到90%的準確率。期望以此建立一可攜式量測系統應用於偵測患者偏頭痛發作時期讓患者能在居家使用並作為提供臨床個人服藥時機的最佳建議。

Migraine is a common neurological disease with intense throbbing headache and nausea, vomiting, and sensitivity to light and sound. Migraine is recurrent and cause severe disability during attacks. Currently, abortive treatment is the standard for acute migraine attacks. However, the efficacy is reduced if the medications start late after onset. To predict the onset of migraine may provide a higher chance for patients to take the medications earlier. Past studies have indicated that habituation trends of different sensory stimuli change at different states of the migraine cycle including inter-ictal, pre-ictal, ictal and post-ictal states. The aim of this study is exploring the habituation trends in four migraine states and compare the trends through inter and intra subjects by using Steady-State Visual Evoked Potentials (SSVEP) examination combined with an feature extraction method improved by canonical correlation analysis (CCA) and calculating the habituation index in order to develop a new migraine states classification system intergraded with feature extraction algorithm, which is able to identify the prodromal symptoms of migraine and thus let patients take therapy in advance. In our results, the inter-ictal group showed the dys-habituation phenomenon and the pre-ictal group showed the pre-ictal normalization in O1 & O2 channels under 15 Hz stimulus. In the classification results, the highest accuracy of migraine states classification can reach nearly 80% for the 4 different migraine states, an even up to almost 90% for 2 migraine states, i.e. pre-ictal and inter-ictal states. We hope to develop a portable system that can detect the timing of migraine attacks at home and provide a more timely opportunity for patients to use their migraine abortive treatment.