心音時序演算法精確計算脈搏速率

Abstract

脈搏速率對於臨床醫師診療以及評估病人具有相當重要的意義，但在許多急救場合中，現行常用的血氧計量測指尖血氧濃度換算脈搏訊號這種間接方式常會受到傷者斷肢或是指甲裝飾(指甲油、彩繪)等干擾，當病人遇到危急情況如無脈搏心臟電氣生理活動(Pulseless electronic activity, PEA)時，可能會因誤判而施予不當之急救措施或是拉長判斷時間，錯失急救良機，故本研究提出直接以心音反映脈搏訊號的脈搏計算心音時序演算法以及搭載此演算法的DS301來解決問題。 心音時序精密演算法嘗試將48KHz的聲音訊號經過降低取樣、帶通濾波器、TT濾波函數、TTMA動態平均法、波峰定位、標準差演算與TT演算輸出等七道程序來簡化並抓出頻率較強波段，藉由不斷重複的週期特徵、標準差的計算與閥值條件限制過濾出乾淨的第一、第二心音(S1, S2)來對應脈搏訊號，此種方式有效降低環境存在大型雜音時錯誤計算發生的機率。利用搭載此演算法的原型產品DS301測試18位實驗對象包含肺動脈瓣、主動脈瓣、三尖瓣、二尖瓣等五處量測點的脈搏訊號，並以MP70生理監視器(Philips intellivue mp70)所測得的脈搏訊號作為基準比較，結果統合150筆測試資料，TSA的偵測確率為91.33%，且平均偵測速率為7.025秒，具有相當潛力。

Pulse rate measurement is important for clinicians to access patient treatment as well as evaluate patient status. However, in many emergency situations, oxygen concentration was obtained by pulse oximeters and transformed into pulse rate, which is very unstable and can be influenced by the injured limb or nails decoration (nail polish, painting). When patients are experiencing critical situations such as pulseless electronic activity (PEA), it can lead to misjudgment and administered inappropriate first aid measures or lengthen the judgment time. In this study, we develop an algorithm, which can relate to the heart sound to calculate heart rates. The DS301 is equipped with this algorithm to measure the exact heart rate by the heart sound. This timing precision algorithm collected the 48KHz heart sound signal through reduce sampling, a band-pass filter, TT filtering function, TTMA moving average method, the peak position, and standard deviation calculations to simplify and catch the frequency strong band. With repeated cycles, characteristic standard deviation calculation and threshold conditions to filter out clean first, second heart sound (S1, S2) to the corresponding pulse signal. In this way, we greatly reduce the environmental noise, which diminishes the probability of inaccuracy in the heart rate measurement. To test this algorism, the prototype DS301 was used on 18 different subjects to measure heart rate included pulmonary valve, aortic pulse signal, tricuspid, and mitral valves at five diferent points. The MP70 physiological monitors (Philips intellivue mp70) was used as a reference to measure the heart rate. The TSA detection rate was 91.33% and an average detection rate of 7.025 seconds, which indicate that DS301 is with considerable potential.