|Abstract: ||本論文，我們對睡眠呼吸中止症患者之心電圖訊號進行希爾伯特-黃轉換(HHT)，分析各七個，睡眠呼吸中止症患者和正常者持續10秒之心電圖樣本。 分析結果顯示，正常者的本質模態函數IMF1(intrinsic mode function 1，本質模態函數1)、IMF2和IMF3之平均能量比均大於10%，睡眠呼吸中止症患者的本質模態函數IMF1和IMF2之平均能量比均大於10%；在0-2Hz頻段，睡眠呼吸中止症患者的本質模態函數IMF2和IMF3之平均能量比為12.93%以及5.95%；在2-4Hz頻段，睡眠呼吸中止症患者的本質模態函數IMF3之平均能量比為9.34%；在10-20Hz頻段，睡眠呼吸中止症患者的本質模態函數IMF1之平均能量比為5.19%；0-2Hz頻段，正常者的本質模態函數IMF4和IMF5之平均能量比為6.07%和12.13%；在2-4Hz頻段，正常者的本質模態函數IMF2、IMF3和IMF4之平均能量比為6.48%、6.71%和6.22%；在6-8Hz頻段，正常者的本質模態函數IMF2之平均能量比為5.94%；在10-20Hz頻段，正常者的本質模態函數IMF1之平均能量比為5.30%；在20-80Hz頻段，正常者的本質模態函數IMF1之平均能量比為6.33%；基於能量-頻率-IMF分布特徵，使我們更明確的瞭解到，睡眠呼吸中止症患者心電圖的IMF特徵。|
This thesis uses the Hilbert-Huang transform (HHT) to analyze electrocardiography (ECG) signals in sleep apnea. The sample included five sleep apnea records, and five control ECG records, each having a 10 second duration. Results show that in the control group, the average energy ratios of the normal wave for intrinsic mode functions (IMF), 1, 2, and 3, to the refereed total energy are all larger than 10 %. This outcome is also obtained in IMF1, and IMF2 in the sleep apnea group. Average energy ratios of IMF2, and IMF3 for the sleep apnea group in the 0-2 Hz band are 12.93%, and 5.95%, respectively. Average energy ratios of IMF3 for the sleep apnea group in the 2-4 Hz band are 9.34%, respectively. The average energy ratio of IMF1 for the sleep apnea group in the 10-20 Hz bands is 5.19%. Average energy ratios of IMF4, and IMF5 for the control group in the 0-2 Hz band are 6.07%, and 12.13%, respectively. The average energy ratios of IMF2、IMF3, and IMF4 for the control group in the 2-4 Hz bands are 6.48%、6.71%, and 6.22%, respectively. The average energy ratio of IMF2 for the control group in the 6-8 Hz bands is 5.94%, respectively. The average energy ratio of IMF1 for the control group in the 10-20 Hz bands is 5.30%, respectively. The average energy ratio of IMF1 for the control group in the 20-80 Hz bands is 6.33%. The proposed feature extraction approach, which is based on energy distribution, enables us to better understand the differences between energy-frequency characteristics of IMFs and residual function of sleep apnea ECG signals.