癲癇腦波與抗癲癇藥物作用之關聯性分析

The instantaneous EEG frequency assessment among various AEDs usage for epileptic treatment

謝懿(Yi Hsieh)

指導教授 ： 蔡瑞章

共同指導教授 ： 饒敦(Tun Jao)

國立臺灣大學/醫學院/醫療器材與醫學影像研究所/碩士(2019年)

https://doi.org/10.6342/NTU201904014

摘要

根據世界衛生組織在2015 年的報告顯示，癲癇疾患每年影響了世界上3.3% 的人口。有高達32.5% 的患者因為各種原因而放棄的抗癲癇藥物的治療。其中有 12.4% 的患者因為藥物副作用所造成的困擾而放棄治療 ; 有 11.6% 因為所開立的抗癲癇藥物無顯著療效而選擇性地放棄治療 ; 最後，有剩下的 8.5% 的患者因為上述兩者原因而最終放棄治療。因此，如何提升患者依從性與有效率的選用藥劑成為相當重要的環節。為了能夠從腦波中擷取足夠的潛在資訊，在此研究中使用了時域上的頻率變化轉換。在這個研究中使用了八個種演算來針對腦波非線性與非穩態之特性來進行特徵提取。研究結果顯示，針對癲癇發生時之腦波進行轉換與演算後再進行支持性向量機的雙向比對所得到對於服用Phenobarbital 與Phenytoin 的患者有73.95%的辨認精準度 ; 對於服用Phenobarbital 與 Levetiracetam 的患者有64.75% 的辨認精準度 ; 對於服用Phenytoin 與 Levetiracetam的患者有68.25% 的辨認精準度。然而，完整的腦波 (其中包含了發作期間、正常期間、發作前期間) 經過運算與特徵擷取後放入支持向量機分類取得了96%, 91.25%, and 97.5% 的比對精準度。最後針對三種用藥者的腦波進行運算後的預兆期 (發作前 10ms) 與發作期進行比對後針對Phenobarbital 得到了80% ; 針對Levetiracetam 得到了75% ; 針對 Phenytoin 得到了85%的辨認精準度。最後結論，此研究證明了癲癇腦預兆期的存在並且可以使用 EEG 演算後偵測到。另外，對於針對時域-頻率變化上，訊號震盪與了解癲癇波段的腦波最有關聯性。

關鍵字

癲癇 ； 腦波 ； 支持向量機 ； 希爾伯特黃轉換 ； 抗癲癇藥物 ； 瞬間頻率

並列摘要

In accordance to the World Health Organization (WHO)’s statistical review in 2015, epilepsy has been affecting almost 3.3% of global population every year. According to previous research. There are more than 32.5% of patient has invalid control of epilepsy. Among them, there are 12.4% quit the treatment because of the adverse effect of AED, 11.6% stop accepting the medication due to the lack of efficacy of the AED prescribed, and 8.5% of patients quit because of mixture of two issues Thus providing a method for reducing the non-adherence problem during the AED therapy is crucial. In order to extract adequate encrypted information from extracranial EEG, a temporal-frequency study has been performed in data preparation. Additionally, eight features were taken into consideration to recognize the non-stationary and non-linear epileptiform of EEG. In the result, an average of 73.95% accuracy, 64.75% accuracy, and 68.25% accuracy were performed in an SVM pattern recognition for the pair-comparison of ical EEG segment between Phenobarbital versus Phenytoin, Phenobarbital versus Levetiracetam, and Phenytoin versus Levetiracetam respectively. With the pair-comparison the accuracy for classifying of the whole spectrum EEG of Phenobarbital versus Phenytoin, Phenobarbital versus Levetiracetam, and Phenytoin versus Levetiracetam intakes were 96%, 91.25%, and 97.5% respectively. For the recognition of ictal and pre-ictal segregation, the EEG of patients gotten prescribed with Phenobarbital, Levetiracetam, and Phenytoin were 80%, 75%, and 85% accuracy respectively. In the conclusion, this research proved the existence of pre-ical activity, and the signal fluctuation of temporal-frequency information is the most relatable features for recognizing onset movement of a seizure.

並列關鍵字

Hilbert-Huang Transformation ； epilepsy ； Support vector machine ； instantaneous frequency ； anti-epileptic drug ； Encephalography

參考文獻

Abdulhay, E., Alafeef, M., Abdelhay, A., & Al-Bashir, A. (2017). Classification of Normal, Ictal and Inter-ictal EEG via Direct Quadrature and Random Forest Tree. Journal of medical and biological engineering, 37(6), 843-857.

Amorim, B. O., Covolan, L., Ferreira, E., Brito, J. G., Nunes, D. P., de Morais, D. G., . . . Hamani, C. (2015). Deep brain stimulation induces antiapoptotic and anti-inflammatory effects in epileptic rats. Journal of Neuroinflammation, 12. doi:ARTN 162

10.1186/s12974-015-0384-7

Bhardwaj, A., Tiwari, A., Krishna, R., & Varma, V. (2016). A novel genetic programming approach for epileptic seizure, detection. Computer Methods and Programs in Biomedicine, 124, 2-18. doi:10.1016/j.cmpb.2015.10.001

Brodtkorb, E., Samsonsen, C., Sund, J. K., Brathen, G., Helde, G., & Reimers, A. (2016). Treatment non-adherence in pseudo-refractory epilepsy. Epilepsy Research, 122, 1-6. doi:10.1016/j.eplepsyres.2016.02.001