以相異位置PPG訊號間脈衝傳遞時間建立血壓量測模型

Abstract

現今的血壓量測方法仍是以脈壓袖帶為基礎的接觸式量測為主，包含聽音診斷法與示波振幅法。傳統的接觸式血壓量測方法具有高準確度的優點，卻也伴隨著不少缺點，如操作上的不便性、容易造成受試者不舒適與不適合做長時間的量測等。再者也不適用於動脈硬化與心房顫動的患者。因此為實現非接觸式的血壓量測方法，本研究開發以脈衝傳遞時間 (Pulse Transit Time, PTT) 為基礎的血壓量測方法，以同時量測手指與耳朵的光體積描記 (Photoplethysmography, PPG) 訊號來改善接觸式量測的缺點。本方法僅需將附有LED與光感測器的PPG裝置貼附皮膚表面以取代傳統的脈壓袖帶進行量測，計算由二相異位置的PPG訊號間的PTT並與足夠的血壓資訊建立血壓迴歸模型，將可實現簡便的連續血壓監測。 在非接觸式血壓量測方法中，PTT為血壓模型迴歸的泛用參數。一般來說，PTT定義為心電圖 (Electrocardiography, ECG) 中R波峰值 (R peak) 與PPG足點 (Foot point) 間的時間差，二者在生理意義上相近，因此定義其時間差為脈搏波傳遞至目標位置所需的時間，亦即PTT。然而，ECG訊號容易受電壓干擾、汗水與電極貼片狀態等因素影響，進而影響PTT計算的穩定度，因此在本研究中以同時量測二相異位置的PPG波形計算所得之PTT取代傳統PTT定義。許多研究指出PTT與血壓間具有高度相關，包含收縮壓、舒張壓與脈壓。以PWV或PTT為參數所建立的血壓模型主要針對心搏力對血壓造成的影響，亦即脈壓，因此本研究中以脈壓為目標建立迴歸模型。 為計算PTT作為脈壓迴歸的參數，本研究建立一套系統用以同時量測耳朵與手指的PPG訊號，並透過商用儀器進行系統驗證以確保訊號準確度，再針對PPG波形上的特徵開發演算法以針對波形上各特徵點進行擷取，即使是在波形特徵較不明顯的非典型PPG波形中仍適用，在此以二訊號中最大斜率點 (Max slope point) 的時間差定義為PTT。在脈壓迴歸模型中，本研究以PTT為基礎參數，再加以結合流體力學與材料力學理論與適當假設所推導出的脈壓迴歸模型，以力學模型為基礎選取適合的參數進行迴歸，並探討流體力學中體積流率與光學PPG波形特徵間斜率的對應關係後引入脈壓迴歸模型作為參數，最後將本研究所推導出的脈壓迴歸模型與目前文獻中提出的各脈壓模型做比較，並分析模型中各項參數與脈壓的關係及比重以探討各參數對脈壓的影響程度並驗證血管力學模型。最後由單一受試者和複數受試者的迴歸結果可看出，將參數取對數後進行迴歸的脈壓模型在擬合程度上均呈現出顯著的改善，證明脈壓與各項參數之間存在指數關係，並驗證該迴歸模型可適用於不同受試者。

Nowadays, cuff-based blood pressure (BP) monitoring is still a wild spread standard method for health care, which including auscultatory and oscillometric. Both of these methods provide high accuracy. Nevertheless, cuff-based BP monitoring methods present disadvantages such as inconvenience, discomfort, and is inappropriate for long time measuring. Moreover, cuff-based BP measurement is not applicable for subjects with stiff artery or atrial fibrillation problems. To achieve the goal of blood pressure measurement without cuff-based method, an innovative BP monitoring method based on pulse transit time (PTT) was proposed in this study. By measuring photoplethysmography (PPG) signals at finger and ear simultaneously, the issues of cuff-based BP measurement mentioned above could be improved. Thus, two PPG pods with LED and photodetector which required simply contact to skin surface were utilized to replace the inflated cuff. With the two different positioned PPG pods, the PTT could be calculated from the PPG waveforms. Combined with adequate BP data, a simple continuous BP measurement via successive PPG will be fulfilled. For cuff-less BP monitoring, PTT is a wild used parameter. In traditional, the PTT was defined as the time difference between the R-peak of an electrocardiogram (ECG) and the foot point of PPG waveform which represents the moment when blood flow is rushing out of aortic valve. However, the ECG signal is affected by voltage interference, sweat and the state of patches easily which could make the instability of the PTT calculation based on ECG and PPG signals. In this study, the traditional PTT calculation was replaced by detecting two different positioned PPG waveforms at the same time. Several researches have shown the high correlation between PTT and BP, both of systolic blood pressure (SBP), diastolic blood pressure (DBP) and pulse pressure (PP). The BP regression model based on PWV or PTT was established for investigating the effect of heart beat force on BP, that is PP. Therefore, PP is the target of regression in this study. In order to calculate PTT for PP regression model, this study built a system for measuring ear and finger PPG signals at the same time and validated the system by a commercial instrument to ensure the accuracy of measurement. Before PTT calculation, this study developed an algorithm for PPG feature extraction on both typical and untypical PPG waveforms, then defined PTT as the time difference between max slope points on two PPG signals. Besides adapting PTT as the parameter for PP regression model, more parameters were chosen from artery mechanics model which was based on fluid and material mechanics theory and with appropriate assumptions. Here introduced the slope on PPG waveform between characteristic points as the regression parameter which was corresponding to flow rate in artery. Furthermore, this study analyzed the effect and weighting of each parameter in PP regression model to validate the artery mechanics model. Finally, by comparing with the existed PP regression models, the new PP regression model which was proposed in this study by introducing, and showed great improvement on R2 performance either in the results of single subject or multiple subjects regression.