200816953 九、發明說明： 【發明所屬之技術領域】 本發明係有關於醫療用測定器、生物信號波形抽出方 法以及已記錄生物信號波形抽出程式的媒體，尤其係有關 於連續地檢測從生物信號所得之波形的醫療用測定器、生 物信號波形抽出方法以及已記錄生物信號波形抽出程式之 媒體。 【先前技術】 在作爲醫療用測定器的脈搏檢測裝置，有連續地檢測 從生物信號所得之波形的脈搏者。在這種脈搏檢測裝置， 因爲使用從脈搏檢測裝置所輸出之脈搏波形在短時間進行 正確的診斷，所以需要從所檢測之很多個脈搏波形抽出並 顯示代表性波形（以下稱爲代表波形）。 在特開2004 - 1 36 1 07號公報（以下稱爲專利文獻1)所 揭示之動脈硬化檢查裝置，爲了不會顯示已混入由心律不 整或身體活動等所引起的雜訊之波形，並抽出且顯示代表 波形，而採用從逐次所檢測的波形中選擇並顯示前驅波之 尖峰的銳利度接近平均波形的手法。 專利文獻1 ··特開2004 — 1 36 1 07號公報 【發明内容】 【本發明欲解決之課題】 可是，在該專利文獻1所揭示之手法，將波形的特徵 量限定爲前驅波之尖峰的銳利度，因爲失去銳利度以外之 特徵量，所以會有發生，就診斷目的而言所抽出之代表波 200816953 形變成不充分的波形之情況的問題。 又，在其他的手法，雖然亦有使用傅立葉轉換等，從 所檢測之波形算出並顯示平均波形之手法，但是利用此手 法所顯示的波形不是實際上從生物所檢測之波形。因而， 具有發生失去本來之特徵量的情況之問題。 本發明係鑑於這種問題而開發者，其目的在於提供一 種醫療用測定器、生物信號波形抽出方法以及已記錄生物 信號波形抽出程式的媒體，用以從檢測之由生物信號所得 的波形之中，抽出並輸出因應於診斷目的之波形。 【解決課題之方式】 爲了達成該目的，根據本發明之一種形態，醫療用測 定器具備有：檢測部，係檢測從生物信號所得之複數個波 形；算出部，係自該複數個波形分別算出第1特徵量和第 2特徵量；檢索部，係根據從該複數個波形算出之複數個 第1特徵量所得的第1指標和第1特徵量之關係、及從該 複數個波形算出之複數個第2特徵量所得的第2指標和第 2特徵量之關係，從該複數個波形之中檢索代表波形；以 及輸出處理部，係進行用以輸出代表波形之處理。 較佳爲從生物信號所得之波形係脈搏波形；在第1特 徵量及第2特徵量中包含有AI(Augmentation Index)値、脈 搏週期、基線變動率、銳利度、以及ET(Ejection Time)値 之至少一個。 較佳爲檢索部包含有：係數設定部，係對在一個波形 之該第1特徵量及該第2特徵量的各個設定加權係數；及 200816953 決定部，係對於第1指標和第1特徵量之關係、及第2指 標和第2特徵量的關係，分別考慮加權係數，並決定代表 波形。 較佳爲檢索部包含有：平均算出部，係對該複數個波 形算出第1特徵量之平均値，並作爲第1指標，及對該複 數個波形算出第2特徵量之平均値，並作爲第2指標；差 分算出部，係對該複數個波形，分別算出作爲第1指標和 第1特徵量之差分的第1差分、及作爲第2指標和第2特 徵量之差分的第2差分；正規化處理部，係將所算出之第 1差分及第2差分進行正規化；以及決定部，係根據已進 行正規化之第1差分及第2差分來決定代表波形。 更佳爲檢索部又包含有：係數決定部，係對一個波形 之第1特徵量及第2特徵量分別設定加權係數；及加權處 理部，係對已進行正規化之第1差分及第2差分，分別乘 以所設定之加權係數，決定部根據已進行加權之第1差分 及第2差分決定代表波形。 更佳爲檢索部又包含有加法處理部，係對該複數個波 形，分別將已進行加權之第1差分及第2差分柑加；決定 部將已相加之該差分爲最小的波形決定爲代表波形。 較佳爲在該檢索部之檢索處理包含有：第1階段之檢 索處理，係根據對從該複數個波形之第1特徵量及第2特 徵量的各個和第1階段之指標的關係，從該複數個波形之 中將對象波形分類；及第2階段之檢索處理，係根據對從 該對象波形之第1特徵量及第2特徵量的各個和第2階段 200816953 之指標的關係，從該對象波形之中檢索代表波形。 較佳爲該輸出處理部係進行用以和代表波形一起顯示 在檢索部檢測代表波形之處理所使用的値或特定代表波形 之資訊的處·理。 較佳爲第1指標及第2指標都是該複數個波形所對應 的複數個特徵量之平均値、中央値、頻率最大値、最大値、 最小値以及任意的臨限値之任一個。 較佳爲又具備有設定第1指標及該第2指標之指標設 定部。 若根據本發明之其他的形態，生物信號波形抽出方法 係從生物信號所得的波形之中抽出代表波形的方法，其具 備有：取得步驟，係取得連續之複數個波形；劃分步驟， 係從連續之複數個波形劃分單位波形；算出步驟，係從複 數個單位波形之各個，算出第1特徵量和第2特徵量；抽 出步驟，係根據從複數個波形算出之複數個第1特徵量所 得的第1指標和第1特徵量之關係、及從複數個波形算出 之複數個第2特徵量所得的第2指標和第2特徵量之關係， 自該複數個單位波形之中抽出代表波形；以及輸出步驟， 係輸出代表波形。 若根據本發明之另外的形態，已記錄生物信號波形抽 出程式的媒體係一種電腦可讀取的記錄媒體，係令電腦執 行從生物信號所得的波形之中抽出代表波形的處理之程 式，而該程式執行如下的步驟：取得步驟，係取得連續之 複數個波形；劃分步驟，係從該連續之複數個波形中劃分 200816953 單位波形；算出步驟，係從複數個單位波形之各個，算出 第1特徵量和第2特徵量；抽出步驟，係根據從複數個波 形算出之複數個第1特徵量所得的第1指標和第1特徵量 之關係、及從複數個波形算出之複數個第2特徵量所得的 第2指標和第2特徵量之關係，自該複數個單位波形之中 抽出代表波形；以及輸出步驟，係輸出該代表波形。 【發明之效果】 在本發明之醫療用測定器，因爲使用多種特徵量從測 得之生物信號中獲得之波形抽出作爲代表的波形，所以可 抽出安定之波形。 此外，藉由考慮各特徵量之重要度，而可抽出和診斷 目的一致的波形，並可提供更有益於診斷之波形。 【實施方式】 以下，一面參照圖面，一面說明本發明之實施形態。 在以下之說明，對相同的零件及構成元件賦與相同之符 號。該等之名稱及功能亦相同。 在本實施形態，本發明之醫療用測定器係被採用於檢 測從生物信號所得之波形的脈搏之脈搏檢測裝置。可是， 本發明之醫療用測定器未限定於脈搏檢測裝置，只要是用 以檢測係從生物信號所得之波形的裝置，可被其他所有的 裝置採用。 . 參照第1圖，本實施形態之脈搏檢測裝置作爲一個具 體例，大致包含有感測器單元1'顯示單元3以及固定座 單元7而構成。 200816953 顯示單元3包含有：操作部24，係設置成可從外部操 作，爲了輸入與脈搏分析等相關之各種資訊而操作；及顯 示部25 ’係由用以向外部輸出脈搏分析結果等之各種資訊 的 LED(Light Emitting Diode)或 LCD(Liquid Crystal Display) 等構成。 固疋座單兀7包含有：R〇M12(Read Only Memory)或 RAM(Random Access Memory)13，係記憶用以控制脈搏檢測 裝置之資料或程式；CPU(Central Processing Unit)ll，係爲 了集中地控制該脈搏檢測裝置而執行包含有運算的各種處 理；控制電路14，係用以接受來自加壓泵15、負壓泵16、 切換閥17以及CPU11的信號，並向加壓泵15、負壓泵16 以及切換閥1 7傳送；特性可變濾波器22，係可變更成至少 2種値；以及A/D轉換器23。 CPU11係向R0M12存取並讀出程式，再展開於RAM13 上並執行，進行該脈搏檢測裝置整體的控制。然後，CPU 1 1 自操作部24接受來自使用者之操作信號，並根據其操作信 號進行該脈搏檢測裝置整體的控制處理。即，CPU 1 1根據 從操作部24所輸入之操作信號，送出控制信號至控制電路 14、多工器20以及特性可變濾波器22。又，CPU11進行用 以將脈搏分析結果等顯示於顯示部25的控制。 加壓泵1 5係用以將後述之推壓腕帶（空氣袋）1 8的內壓 (以下稱爲「腕帶壓」）進行加壓的泵。負壓泵1 6係用以將 腕帶壓進行減壓的泵。切換閥1 7選擇這些加壓泵1 5和負 壓泵1 6的任一個並切換成和氣壓管5連接。然後，控制電 -10- 200816953 路1 4根據來自CPU 1 1的控制信號控制這些元件。 感測器單元1包含有：半導體壓力感測器1 9，係包含 有多個感測器元件；多工器20，係選擇性地引出多個感測 •器元件各自輸出的壓力信號；放大器21，係用以將從多工 器20輸出之壓力信號放大；以及推壓腕帶18,係爲了將半 導體壓力感測器1 9推壓至手腕上而進行加壓調整。 半導體壓力感測器1 9包含有在由單結晶矽等所構成 _ 之半導體晶片朝向一方向按照既定間隔排列的多個感測器 元件而構成。半導體壓力感測器1 9係利用推壓腕帶1 8的 壓力被推壓至量測中之被檢驗者的手腕等之量測部位。半 導體壓力感測器1 9在此狀態下，經由橈骨動脈來檢測被檢 驗者的脈搏。半導體壓力感測器1 9將藉由檢測脈搏而從感 測器元件所輸出之壓力信號按照各感測器元件之頻道而輸 入至多工器20。多個感測器元件例如排列40個。 多工器20選擇性地輸出各感測器元件所輸出之壓力 φ 信號。從多工器20所送出之壓力信號被放大器21放大， 再經由特性可變濾波器22選擇性地供給A/D轉換器23。 在本實施形態，至選擇最適合用於脈搏檢測之感測器 元件爲止，多工器20根據來自CPU11的控制信號，依序切 換並輸出從各感測器元件所輸出之多個壓力信號。又，在 選擇最適合用於脈搏檢測之感測器元件後，根據來自 CPU11的控制信號，被固定於該頻道。因此，此時，多工 器20選擇並輸出從所選擇之感測器元件輸出的壓力信號。 特性可變濾波器22係用以切斷既定値以上之信號的 -11- 200816953 低通濾波器，可變更成至少2種値。 A/D轉換器23將從半導體壓力感測器1 9所引出之係 類比信號的壓力信號轉換成數位資訊，並供給於CPU 1 1。 A/D轉換器23至利用CPU 11固定多工器20之頻道爲止， 經由多工器20同時取得半導體壓力感測器1 9所包含之各 感測器元件所輸出的壓力信號。然後，在利用CPU 11固定 多工器20之頻道後，A/D轉換器23取得從該感測器元件 所輸出的壓力信號。壓力信號之取樣的週期（以下稱爲「取 樣週期」）例如設爲2ms。 特性可變濾波器22在至固定多工器20之頻道爲止和 已固定後，變更截止頻率的値。至固定多工器20之頻道爲 止，切換多種壓力信號並進行取樣。因此，在特性可變濾 波器22，選擇比此時之取樣頻率更高的截止頻率之値。藉 此，可防止在A/D轉換後發生失真，並可適當地選擇最適 合的感測器元件。而，在已固定頻道後，在特性可變濾波 器22根據來自CPU 1 1之控制信號，選擇對某一個壓力信號 之取樣頻率（例如500Hz)的1/2以下之截止頻率的値。藉 此，可減少折疊雜訊（aliasing noise)，並可精確地進行脈搏 分析。此外，折疊雜訊意指在根據取樣定理將類比信號轉 換成數位信號的情況，由於折回現象，而出現於取樣頻率 之1/2以下的頻域，並具有取樣頻率之1/2以上之頻率成分 的雜訊。 在本實施形態，因爲在固定座單元7上具備有CPU11、 ROM 12以及RAM13，所以可進行顯示單元3的小型化。 -12- 200816953 此外，雖然分開地設置固定座單元7和顯示單元3， 但是亦可係將顯示單元3內建於固定座單元7之構造。又， 反之，亦可採用CPU1 1、ROM12以及RAM13設置於顯示單 元3之構造。又，亦可和PC(Personal Control)連接，並進 行各種控制。 在本實施形態的脈搏檢測裝置，從所檢測之多個脈搏 檢索、抽出並顯示作爲代表的脈搏。第2圖係表示用以在 脈搏檢測裝置，從所檢測之多個脈搏中檢索、抽出並顯示 作爲代表的脈搏並抽出、顯示之功能構造的具體例之方塊 圖。第2圖所示之各功能係CPU11向R0M12存取並讀出程 式，再展開於RAM 13上並執行，雖然主要形成於CPU11， 但是亦可至少一部分利用第1圖所示之裝置發揮的功能。 參照第2圖，在脈搏檢測裝置之該功能，構成爲包含： 脈搏量測値取得部101，係從半導體壓力感測器19所含之 感測器元件取得壓力信號（感測器信號）；脈搏劃分處理部 1 03，係對脈搏量測値取得部1 〇 1所取得之壓力信號進行處 理，並進行逐拍地劃分連續的脈搏之處理；特徵量算出部 105 ’係針對表示脈搏劃分處理部1〇3所劃分之1拍的脈搏 之各單位波形而算出多種特徵量；特徵量記憶部1 07，係逐 拍地記憶所算出之多種特徵量；代表波形檢索處理部109， 係使用針對各單位波形所算出之多種特徵量，進行從連續 的脈搏之中檢索並抽出作爲代表性波形（以下稱爲代表波 形）的脈搏波形之處理；以及代表波形顯示控制部1 1 1，係 進行產生並顯示以顯示部25顯示所抽出之代表波形所需 -13- 200816953 的信號之處理。 作爲特徵量算出部1 05所算出之特徵量的種類，例如 可列舉AI値、ET値、脈搏週期、上昇銳利度MSP以及基 線變動率等。在本實施形態，特徵量算出部105包含有AI 算出部1 050、ET算出部1051、脈搏週期算出部1 052、MSP 算出部1 053以及基線變動率算出部1 054,並算出這5種特 徵量。 在此，AI係表示脈搏所佔反射波的比例之習知指標， 主要係用以評估中樞血管之動脈管壁硬化之指標。AI値是 根據脈搏之最低位置（換言之，脈搏波形的開始位置）和第1 尖峰値的振幅差之位準a、及脈搏之最低位置和第2尖峰値 的振幅差之位準b之比率而得到，例如AI(%) = b/axl00。舉 例說明檢測到第3圖所示之脈搏波形的情況。在第3圖， 縱軸表示腕帶壓，橫軸表示時間經過。這些在以下表示波 形的圖亦相同。位準a表示心臟之心跳所引起的血液之前 驅波的壓力値，位準b表示對心跳之前驅波的反射波所引 起之壓力値。此反射波，對應於血管的硬化，其強度和出 現時相發生變化，血管愈硬化反射波的強度愈大，出現時 相變早（向左偏移）。即，AI値愈大，表示血管的硬化正在 惡化。此外，作爲決定位準a、b的方法，可列舉對脈搏波 形進行微分等之運算操作的方法。即，參照第3圖，對所 檢測的波形，將四次微分所得之微分曲線重疊，並使用和 其極點位置對應的振幅而可決定位準a、位準b。 又，ET意指從主動脈瓣開放開始點至關閉點爲止之時 -14- 200816953 間’和心收縮力、一次打出量、流出路阻力以及末梢阻力 相關的指標。ET値係由表示一個脈搏之波形的二次微分波 形之最大點至第2個極大點爲止的長度而得到。例如，在 檢測到第4圖所示之脈搏波形的情況，從二次微分波形之 最大點PA至第2個極大點PB爲止的長度c表示ET値。 二次微分波形之最大點PA係脈搏的上昇點，第2個極大點 PB表示主動脈瓣開始關閉的點。即，et値愈小，表示心 功能愈低落。200816953 IX. Description of the Invention: [Technical Field] The present invention relates to a medical measuring device, a biological signal waveform extracting method, and a medium for recording a biosignal waveform extracting program, and more particularly to continuously detecting a biological signal. A medical measuring device for waveforms, a biosignal waveform extracting method, and a medium for recording a biosignal waveform extracting program. [Prior Art] A pulse detecting device as a medical measuring device has a pulse that continuously detects a waveform obtained from a biological signal. In such a pulse detecting device, since the pulse waveform outputted from the pulse detecting device is used for accurate diagnosis in a short time, it is necessary to extract and display a representative waveform (hereinafter referred to as a representative waveform) from a plurality of detected pulse waveforms. The arteriosclerosis inspecting apparatus disclosed in Japanese Laid-Open Patent Publication No. 2004-1367-1 (hereinafter referred to as Patent Document 1) does not display the waveform of noise caused by arrhythmia or physical activity, and is extracted. And the representative waveform is displayed, and the method of selecting and displaying the sharpness of the spike of the precursor wave from the successively detected waveforms close to the average waveform is employed. [Problem to be Solved by the Invention] However, in the technique disclosed in Patent Document 1, the feature amount of the waveform is limited to the peak of the precursor wave. The sharpness, because the feature quantity other than the sharpness is lost, there is a problem that the representative wave 200816953 extracted for the purpose of diagnosis becomes an insufficient waveform. Further, in other methods, a Fourier transform or the like is used to calculate and display an average waveform from the detected waveform. However, the waveform displayed by this method is not a waveform actually detected by a living being. Therefore, there is a problem that a situation in which the original feature amount is lost occurs. The present invention has been made in view of such a problem, and an object thereof is to provide a medical measuring device, a biological signal waveform extracting method, and a medium for recording a biosignal waveform extracting program for detecting a waveform obtained from a biological signal. , extract and output the waveform for diagnostic purposes. [Means for Solving the Problems] In order to achieve the object, according to one aspect of the present invention, a medical measuring device includes: a detecting unit that detects a plurality of waveforms obtained from a biological signal; and a calculating unit that calculates each of the plurality of waveforms The first feature amount and the second feature amount; the search unit is a relationship between the first index and the first feature amount obtained from the plurality of first feature quantities calculated from the plurality of waveforms, and a complex number calculated from the plurality of waveforms The relationship between the second index and the second feature amount obtained by the second feature amount is used to retrieve a representative waveform from the plurality of waveforms, and the output processing unit performs processing for outputting a representative waveform. Preferably, the waveform is a pulse waveform obtained from a biological signal; and the first feature amount and the second feature amount include an AI (Augmentation Index), a pulse period, a baseline change rate, a sharpness, and an ET (Ejection Time). At least one of them. Preferably, the search unit includes a coefficient setting unit that sets a weighting coefficient for each of the first feature amount and the second feature amount of one waveform; and a 200816953 determining unit for the first index and the first feature amount. The relationship between the second index and the second feature amount is considered as a weighting coefficient, and the representative waveform is determined. Preferably, the search unit includes an average calculation unit that calculates an average value of the first feature amount for the plurality of waveforms, and calculates an average value of the second feature amount for the plurality of waveforms as the first index. The second index; the difference calculation unit calculates a first difference that is a difference between the first index and the first feature amount, and a second difference that is a difference between the second index and the second feature amount for the plurality of waveforms; The normalization processing unit normalizes the calculated first difference and the second difference, and the determination unit determines the representative waveform based on the first difference and the second difference that have been normalized. More preferably, the search unit further includes: a coefficient determining unit that sets a weighting coefficient for each of the first feature quantity and the second feature quantity of one waveform; and a weighting processing unit that is configured to normalize the first difference and the second The difference is multiplied by the set weighting coefficient, and the determining unit determines the representative waveform based on the first difference and the second difference that have been weighted. More preferably, the search unit further includes an addition processing unit that adds the first difference and the second difference that are weighted to the plurality of waveforms, and the determination unit determines the waveform in which the difference is minimized as Represents a waveform. Preferably, the search processing in the search unit includes the first stage search processing based on the relationship between each of the first feature quantity and the second feature quantity of the plurality of waveforms and the index of the first stage. Among the plurality of waveforms, the target waveform is classified; and the second stage of the search processing is based on the relationship between the first feature amount and the second feature amount from the target waveform and the index of the second stage 200816953. The representative waveform is retrieved among the object waveforms. Preferably, the output processing unit performs processing for displaying information on the 値 or the specific representative waveform used in the process of detecting the representative waveform by the search unit together with the representative waveform. Preferably, the first index and the second index are any one of an average 値, a central 値, a maximum frequency 値, a maximum 値, a minimum 値, and an arbitrary threshold 复 of the plurality of feature quantities corresponding to the plurality of waveforms. Preferably, the indicator setting unit that sets the first index and the second indicator is further provided. According to another aspect of the present invention, the biosignal waveform extraction method is a method of extracting a representative waveform from a waveform obtained from a biosignal, and includes: obtaining a step of acquiring a plurality of consecutive waveforms; and dividing the step from continuous The plurality of waveforms divide the unit waveform; the calculating step calculates the first feature quantity and the second feature quantity from each of the plurality of unit waveforms; and the extracting step is based on the plurality of first feature quantities calculated from the plurality of waveforms a relationship between the first index and the first feature amount, and a relationship between the second index and the second feature amount obtained from the plurality of second feature quantities calculated from the plurality of waveforms, and extracting a representative waveform from the plurality of unit waveforms; In the output step, the output represents the waveform. According to another aspect of the present invention, the medium for recording the biosignal waveform extracting program is a computer readable recording medium for causing the computer to execute a program for extracting a representative waveform from the waveform obtained by the biological signal, and The program performs the following steps: the obtaining step is to obtain a continuous plurality of waveforms; the dividing step is to divide the 200816953 unit waveform from the continuous plurality of waveforms; and the calculating step is to calculate the first characteristic from each of the plurality of unit waveforms. The amount and the second feature amount; the extraction step is a relationship between the first index and the first feature amount obtained from the plurality of first feature quantities calculated from the plurality of waveforms, and a plurality of second feature quantities calculated from the plurality of waveforms The relationship between the obtained second index and the second feature amount is obtained by extracting a representative waveform from the plurality of unit waveforms, and an output step of outputting the representative waveform. [Effects of the Invention] In the medical measuring device of the present invention, since a representative waveform is extracted from a waveform obtained from the measured biological signal using a plurality of characteristic amounts, a stable waveform can be extracted. In addition, by considering the importance of each feature quantity, it is possible to extract waveforms that are consistent with the purpose of diagnosis, and to provide waveforms that are more beneficial for diagnosis. [Embodiment] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description, the same components and constituent elements are assigned the same symbols. The names and functions of these are also the same. In the present embodiment, the medical measuring device of the present invention is a pulse detecting device for detecting a pulse of a waveform obtained from a biological signal. However, the medical measuring device of the present invention is not limited to the pulse detecting device, and any device that can detect the waveform obtained from the biological signal can be used by all other devices. Referring to Fig. 1, the pulse detecting device of the present embodiment is configured as a specific example, and includes a sensor unit 1' display unit 3 and a fixed base unit 7. 200816953 The display unit 3 includes an operation unit 24 that is operable to be externally operated to input various information related to pulse analysis and the like, and a display unit 25' that is used to output pulse analysis results to the outside. Information LED (Light Emitting Diode) or LCD (Liquid Crystal Display). The solid housing unit 7 includes: R〇M12 (Read Only Memory) or RAM (Random Access Memory) 13 for storing data or programs for controlling the pulse detecting device; CPU (Central Processing Unit) ll for concentration The pulse detecting device is controlled to execute various processes including calculations; the control circuit 14 is configured to receive signals from the pressurizing pump 15, the negative pressure pump 16, the switching valve 17, and the CPU 11, and to the pressurizing pump 15, negative The pressure pump 16 and the switching valve 17 are transmitted; the characteristic variable filter 22 can be changed to at least two types of ports; and the A/D converter 23. The CPU 11 accesses and reads the program to the ROM 12, expands it on the RAM 13, and executes it, and controls the entire pulse detecting device. Then, the CPU 1 1 receives an operation signal from the user from the operation unit 24, and performs control processing of the entire pulse detecting device based on the operation signal. That is, the CPU 1 1 sends a control signal to the control circuit 14, the multiplexer 20, and the characteristic variable filter 22 based on the operation signal input from the operation unit 24. Further, the CPU 11 performs control for displaying the pulse analysis result or the like on the display unit 25. The pressure pump 15 is a pump for pressurizing an internal pressure (hereinafter referred to as "wrist band pressure") of a wristband (air bag) 18 to be described later. The vacuum pump 16 is a pump for decompressing the wrist strap. The switching valve 17 selects either one of the pressurizing pump 15 and the negative pressure pump 16 and switches to be connected to the pneumatic tube 5. Then, the control unit -10- 200816953 way 14 controls these elements in accordance with control signals from the CPU 11. The sensor unit 1 includes: a semiconductor pressure sensor 197, comprising a plurality of sensor elements; and a multiplexer 20 for selectively extracting a pressure signal outputted by each of the plurality of sensing device elements; 21, for amplifying the pressure signal output from the multiplexer 20; and pushing the wristband 18 for pressure adjustment in order to push the semiconductor pressure sensor 19 to the wrist. The semiconductor pressure sensor 19 includes a plurality of sensor elements arranged in a predetermined interval in a semiconductor wafer composed of a single crystal germanium or the like. The semiconductor pressure sensor 19 is pressed by the pressure of the pressing wrist band 18 to the measurement portion of the wrist or the like of the subject to be measured. In this state, the semiconductor pressure sensor 19 detects the pulse of the subject via the radial artery. The semiconductor pressure sensor 19 inputs the pressure signal output from the sensor element by detecting the pulse to the multiplexer 20 in accordance with the channel of each sensor element. For example, a plurality of sensor elements are arranged in 40 pieces. The multiplexer 20 selectively outputs a pressure φ signal output from each of the sensor elements. The pressure signal sent from the multiplexer 20 is amplified by the amplifier 21, and is selectively supplied to the A/D converter 23 via the characteristic variable filter 22. In the present embodiment, the multiplexer 20 sequentially switches and outputs a plurality of pressure signals output from the respective sensor elements in accordance with a control signal from the CPU 11 until the sensor element most suitable for pulse detection is selected. Further, after selecting the sensor element most suitable for pulse detection, it is fixed to the channel based on the control signal from the CPU 11. Therefore, at this time, the multiplexer 20 selects and outputs a pressure signal output from the selected sensor element. The characteristic variable filter 22 is a low-pass filter of -11-200816953 for cutting a signal of a predetermined value or more, and can be changed to at least two types of turns. The A/D converter 23 converts the pressure signal of the analog signal drawn from the semiconductor pressure sensor 19 into digital information and supplies it to the CPU 11. The A/D converter 23 simultaneously acquires the pressure signal output from each of the sensor elements included in the semiconductor pressure sensor 19 via the multiplexer 20 until the CPU 11 fixes the channel of the multiplexer 20. Then, after the channel of the multiplexer 20 is fixed by the CPU 11, the A/D converter 23 obtains the pressure signal output from the sensor element. The period of sampling of the pressure signal (hereinafter referred to as "sampling period") is set to, for example, 2 ms. The characteristic variable filter 22 changes the threshold of the cutoff frequency until it is fixed to the channel of the fixed multiplexer 20. After the channel to the fixed multiplexer 20, various pressure signals are switched and sampled. Therefore, in the characteristic variable filter 22, the threshold of the cutoff frequency higher than the sampling frequency at this time is selected. Thereby, distortion can be prevented after the A/D conversion, and the most suitable sensor element can be appropriately selected. On the other hand, after the channel has been fixed, the characteristic variable filter 22 selects 値 of a cutoff frequency of 1/2 or less of the sampling frequency (e.g., 500 Hz) of a certain pressure signal based on the control signal from the CPU 11. As a result, aliasing noise can be reduced and pulse analysis can be performed accurately. In addition, the folding noise means that when the analog signal is converted into a digital signal according to the sampling theorem, it appears in the frequency domain below 1/2 of the sampling frequency due to the folding back phenomenon, and has a frequency of 1/2 or more of the sampling frequency. The noise of the ingredients. In the present embodiment, since the CPU 11, the ROM 12, and the RAM 13 are provided in the fixed base unit 7, the size of the display unit 3 can be reduced. -12- 200816953 Further, although the mount unit 7 and the display unit 3 are separately provided, the configuration in which the display unit 3 is built in the mount unit 7 may be employed. Further, conversely, a configuration in which the CPU 1, the ROM 12, and the RAM 13 are provided in the display unit 3 can be employed. Also, it can be connected to a PC (Personal Control) and perform various controls. In the pulse wave detecting apparatus of the present embodiment, the representative pulse is retrieved and extracted from the detected plurality of pulses. Fig. 2 is a block diagram showing a specific example of a functional configuration for searching, extracting, and displaying a representative pulse from the pulse detecting device, and extracting and displaying the pulse. Each of the functions shown in FIG. 2 is a CPU 11 that accesses and reads a program from the ROM 12, expands it on the RAM 13, and executes it. Although it is mainly formed in the CPU 11, it is also possible to use at least a part of the functions of the device shown in FIG. . Referring to Fig. 2, the function of the pulse detecting device is configured to include: a pulse rate measuring unit 101 for obtaining a pressure signal (sensor signal) from a sensor element included in the semiconductor pressure sensor 19; The pulse division processing unit 203 processes the pressure signal acquired by the pulse rate measurement acquisition unit 1 〇1, and performs a process of dividing the continuous pulse beat by beat; the feature amount calculation unit 105' is directed to the pulse division processing. The plurality of feature quantities are calculated by the unit waveforms of the pulse of one beat divided by the unit 1〇3, and the feature amount storage unit 107 records the calculated plurality of feature amounts on a beat-by-shot basis; the representative waveform search processing unit 109 uses a plurality of feature quantities calculated by the unit waveforms, a process of searching for and extracting a pulse waveform as a representative waveform (hereinafter referred to as a representative waveform) from a continuous pulse; and a representative waveform display control unit 1 1 1 The processing of displaying the signal of -13 - 200816953 required for the extracted representative waveform by the display portion 25 is displayed. The types of the feature amounts calculated by the feature amount calculation unit 506 include, for example, AI値, ET値, pulse period, rising sharpness MSP, and base line variation rate. In the present embodiment, the feature amount calculation unit 105 includes an AI calculation unit 1 050, an ET calculation unit 1051, a pulse period calculation unit 1 052, an MSP calculation unit 1 053, and a baseline change rate calculation unit 1 054, and calculates these five characteristics. the amount. Here, AI is a conventional indicator indicating the ratio of the reflected wave of the pulse, and is mainly used to evaluate the arterial wall hardening of the central blood vessel. AI値 is based on the ratio of the lowest position of the pulse (in other words, the start position of the pulse waveform) to the level a of the amplitude difference of the first peak 、, and the level of the lowest position of the pulse and the level b of the amplitude difference of the second peak 値. Get, for example, AI(%) = b/axl00. The case where the pulse waveform shown in Fig. 3 is detected will be described by way of example. In Fig. 3, the vertical axis represents the wristband pressure, and the horizontal axis represents the passage of time. These figures are also the same in the following waveforms. The level a represents the pressure 之前 of the blood before the heartbeat of the heart, and the level b represents the pressure 引 caused by the reflected wave of the drive wave before the heartbeat. This reflected wave corresponds to the hardening of the blood vessel, and its intensity and appearance phase change. The stronger the intensity of the blood vessel hardening reflected wave, the earlier the phase changes (shifts to the left). That is, the larger the AI is, the more the hardening of the blood vessel is worsening. Further, as a method of determining the levels a and b, a method of performing arithmetic operations such as differentiation of the pulse waveform can be cited. That is, referring to Fig. 3, the differential waveform obtained by the four differentials is superimposed on the detected waveform, and the level a and the level b can be determined using the amplitude corresponding to the position of the pole. In addition, ET means an index related to cardiac contractility, primary output, outflow resistance, and distal resistance from the start of the aortic valve opening to the closing point -14-200816953. The ET system is obtained by expressing the length from the maximum point of the second differential waveform of the waveform of one pulse to the second maximum point. For example, in the case where the pulse waveform shown in Fig. 4 is detected, the length c from the maximum point PA of the second differential waveform to the second maximum point PB indicates ET 値. The maximum point of the second differential waveform is the rising point of the pulse, and the second maximum point PB indicates the point at which the aortic valve begins to close. That is, the smaller the et, the lower the heart function.
I 又’脈搏週期意指一個劃分之脈搏波形的週期，作爲 檢測身體活動、雜訊等之影響所引起的波形形狀之時間成 分的變形或心律不整之指標。作爲決定脈搏波形之週期的 方法之一例，可列舉如下的方法。參照第5圖，檢測到原 波形之最大値時，參照從和下一個上昇方向之零交越點 (Eero cross point)對應的極小點（PB點）至和前一個上昇方 向之零交越點對應的極小點（PA點）爲止之原波形，在其間 存在最大點（PP點），且只要確認在從PA點至PB點爲止之 間PA點爲最小値，就將PA點確定爲「上昇點」，換言之， 一個脈搏之「脈搏起點」。而且，將從PA點至PB點爲止 的長度D決定爲一個脈搏之脈搏波形的週期。 又，上昇銳利度MSP係顯著地表示推壓腕帶18之推 壓力並評估波形失真之指標。在半導體壓力感測器1 9之輸 出變化的上昇點爲尖銳的情況下，表示推壓腕帶1 8之推壓 力爲適當，而上昇點接近平坦的變化時，推壓腕帶18之推 壓力爲不適當，表示量測部位之皮下的動脈要加壓至必要 -15- 200816953 値以上。上昇銳利度MSP係將半導體壓力感測器1 9之輸 出變化的上昇點之尖銳利度進行數値化的値。例如，參照 第6圖，將從在以脈搏波形的上昇位置爲中心所設定之既 定的劃分點（TDIA)之振幅上昇最大振幅的10%之2點的時 間間隔定義爲Ta，將從該2點間之端點至脈搏波形的尖峰 値爲止之時間定義爲Tb時，可由Ta和Tb的比率得到上昇 銳利度MSP，例如MSP = Ta/Tb。 基線變動率表示相鄰之脈搏波形的基線之變動，作爲 檢測身體活動、雜訊等之影響所引起的波形形狀之振幅成 分的變形之指標。可由脈搏之最大値及「脈搏起點」和下 一脈搏的「脈搏起點」之間的差之比率得到基線變動率。 例如，在檢測到第6圖所示之脈搏波形的情況，若設脈搏 起點S 1和脈搏最大點S2之壓力差爲A、設脈搏起點S 1和 下一脈搏的脈搏起點S 3之壓力差爲B，則基線變動率 (%) = B/Axl 00。此外，可利用和在脈搏週期之說明所述的方 法一樣之方法來決定「脈搏起點」。 在特徵量記憶部1 07，記憶對各脈搏所算出之該5種特 徵量。此外，特徵量算出部105所算出之特徵量的種類未 限定爲該5種，亦可係其中之至少2種。又，亦可算出其 他的特徵量。作爲其他的特徵量，例如，可列舉成爲評估 反射波之時間成分或動脈管壁的硬化度之指標的前進波最 大點和反射波最大點之時間間隔，或表示前進波上昇點和 反射波上昇點之時間間隔的TR(Traveling time to Reflected wave)等之習知指標。此外，可採用有益於使用脈搏波形來 -16- 200816953 進行診斷之所有特徵量。 參照第7圖’在代表波形檢索處理部1〇9包含有：特 徵量讀出部1 09 1 ’係對各脈搏波形從特徵量記憶部1 〇7讀 出所記憶之特徵量；平均算出部1 092，係對各種特徵量算 出平均値並設定爲指標；差分算出部1〇93，係對各特徵量 算出與平均値的差分來作爲和該指標之關係；正規化處理 部1 094 ’係因應於特徵量的種類將所算出之差分値進行正 規化；係數決定部1095，係決定與在抽出代表波形時考慮 之特徵量的種類相關的加權係數；加權處理部1 〇 9 6，係對 各脈搏波形使用所決定之加權係數將已正規化的該差分値 進行加權；加法處理部1 097，係算出已加權之該差分値之 各脈搏波形的合計；以及代表波形決定部1 098，係自各脈 搏波形之差分値的該合計來決定代表波形。 在此，加權意指在使用如上述所示之多種特徵量來抽 出代表波形時所使用的特徵量之重要度，例如藉由將加權 係數設爲1，而對於和在抽出代表波形時對應之特徵量直 接使用該値；藉由將加權係數設爲0.5，而對於和在抽出代 表波形時對應之特徵量使用該値的50% ;藉由將加權係數 設爲0，而不使用和在抽出代表波形時對應之特徵量。因 而，在此藉由設定加權係數的組合，而亦可僅使用所算出 之多個特徵量之中的某一個特徵量（例如AI値）來抽出代表 波形。又，平均地使用所算出之多個特徵量的全部亦可抽 出代表波形。又，根據某目的’特別重視所算出的多個特 徵量之中的特定特徵量，並使其他的特徵量之重要度變 -17- 1095 200816953 小，亦可抽出代表波形。這種加權係數係在係數決定部 中所決定。作爲在係數決定部10 9 5決定加權係數之戈 可列舉如下的方法。 作爲在係數決定部1 095決定加權係數的第1方法 提，在操作部24具備有特定的按鈕。或在顯示部25 設定畫面。醫生等之使用者配合診斷目的而進行該特 鈕之操作或進行按照該設定畫面的操作，以輸入和各 量對應之加權（數字或重要度（小、中、大）選擇等）。係 定部1 095接受那些操作之操作信號，並對各特徵量決 權係數。 作爲第2方法，可列舉從在本量測之前輸入，並 於RAM 13等的既定區域之被檢驗者的病歷或診斷目的 醫生等的使用者的輸入資訊，自動地設定特徵量之加 數的方法。或，可列舉從年齢、性別等之關於被檢驗 輸入資訊，自動設定合乎年代、性別之發病率高的症 特徵量的加權係數之方法。在自動地設定加權係數的 下，預先將該輸入資訊和對各特徵量之加權係數的關 憶於ROM12等之記憶區域，而由係數決定部1 095因 需要參照該記憶區域，讀出對應之加權係數，藉此決 第8圖係表示在本實施形態的脈搏檢測裝置，在 脈搏後顯示代表波形之處理的流程圖。第8圖之流程 示的處理，係CPU11向ROM12存取並讀出程式，再展 RAM 1 3上並執行，控制第2圖及第7圖所示之各功戧 此而實現。 ▼法， 之前 顯示 定按 特徵 數決 定加 記憶 等之 權係 者的 狀之 情況 係記 應於 定。 :量測 丨圖所 :開於 i，藉 -18- 200816953 參照第8圖，首先，在脈搏量測値取得部1 〇丨取得來 自半導體壓力感測器19的壓力信號，並量測脈搏（步驟 S 1)。在步驟s1所量測之脈搏在脈搏劃分處理部丨〇 3被劃 分成各個脈搏之脈搏（步驟S3)，在特徵量算出部1〇5，對各 脈搏波形以上述的方法算出該5種特徵量（步驟s5)。該步 驟S1〜S5的處理，亦可和脈搏之量測平行地進行。在此情 況’至脈搏量測結束爲止，對各脈搏波形進行在步驟S 5之 特徵量的算出。然後，量測結束時，在代表波形檢索處理 部109 ’根據關於在步驟S5所算出之各脈搏波形的該5種 特徵量檢索代表波形，並抽出符合的脈搏波形（步驟S 7)。 代表波形顯示控制部1 1 1係產生用於使在步驟S7中作爲代 表波形而被抽出之脈搏波形作爲量測結果而和其他的計算 値一起在顯示部25顯示的顯示信號，並進行顯示（步驟S 9)。 作爲在該步驟S 3執行的脈搏劃分處理，在第9圖之流 程圖表示一例。此外，在以後，將此處理當作係已固定多 工器20之頻道後的分析處理來說明。 參照第9圖，首先，檢測從半導體壓力感測器1 9已送 出壓力信號時（S301)，多工器20選擇與根據來自CPU11之 控制信號所固定的頻道符合之感測器元件的感測器信號， 並輸入放大器2 1。 然後，所輸入之壓力信號在放大器21，將壓力信號放 大至既定之頻率爲止（S3 03)，在特性可變濾波器22進行類 比濾波器處理（S 305)。 此時，特性可變濾波器22阻斷取樣頻率之丨/2以上的 -19- 200816953 信號成分。若取樣頻率係500Hz時，例如阻斷頻率超過 2 5 0 Η z之信號成分。即，在此，除去符合規格範圍外之脈 搏波形的信號成分。 已通過特性可變濾波器22之壓力信號，在A/D轉換器 23被轉換成數位信號（S3 07)，並爲了除去雜訊等而進行用 以抽出既定範圍之頻率的數位濾波器處理（S3 09)。然後， A/D轉換器23向CPU11傳輸已數位化之壓力信號。 脈搏劃分處理部103對於從A/D轉換器23所接收之壓 力信號，藉由取各資料的差分而將從壓力信號所得之脈搏 波形進行Ν次微分（S3 11)。然後，脈搏劃分處理部1〇3根 據步驟S 3 1 1之微分結果未劃分脈搏波形，而抽出一個脈搏 的脈搏波形（S313)。 第10圖係表示在步驟S7在代表波形檢索處理部109 所執行之代表波形檢索處理的一例之流程圖。參照第1 0 圖，在該步驟S7，對於特徵量讀出部1 〇9 1從特徵量記憶部 107所讀出之特徵量，在平均算出部1092，對各種特徵量 算出平均値（步驟S701)。接著，在差分算出部1 093，對各 特徵量算出與該平均値的差分（步驟S703)。在正規化處理 部1 094，對在步驟S703所算出之差分進行正規化處理（步 驟S 705)。在此之正規化處理係一般的正規化處理，在本發 明中未限定爲特定之處理。在步驟S7 05已進行正規化的差 分，在加權處理部1 096，使用在係數決定部1 095所決定之 加權係數進行加權（步驟S707)。具體而言，對在步驟S705 已進行正規化的差分，因應於特徵量之種類乘以利用係數 -20- 200816953 決定部1 095對各種特徵量所決定的加權係數，藉此進行加 權。在加算處理部1 097，加上已進行加權處理之一個脈搏 的各波形之該5種特徵量的差分，並算出各波形之差分的 和（步驟S709)。然後，在代表波形決定部1 098，將對於該 和爲最小的波形，即對於該5種特徵量之全部整體上最接 近平均値的波形決定爲代表波形，並予以抽出（步驟S7 11)。 第1 1圖係說明在該步驟S9根據來自代表波形顯示控 制部1 1 1之顯示信號所顯示在顯示畫面之顯示內容的圖。 參照第1 1圖，在顯示畫面，作爲代表波形，顯示在該處理 所抽出之一個脈搏的脈搏波形200，及將血壓値或AI的平 均値作爲脈搏量測之結果3 0 0，顯示對於有效的全波形所得 之血壓的平均値及AI値之平均値。此外，較佳爲在顯示畫 面中，作爲表示關於抽出代表波形之處理的顯示，包含有 所抽出之代表波形表示屬於有效的波形之中的第幾個波形 之顯示20 1、或表示在抽出代表波形時所使用的特徵量在該 波形上的特徵性位置或値之記號203 : 205、207、或該特徵 量的顯示209。在第11圖之例子，在抽出代表波形時，使 用該5種特徵量之中的AI値、ET値以及脈搏週期，記號 2 03表示在算出AI値時所使用之上述的位準a和位準b之 位置，記號205表示相當於ET之上述長度c，記號207表 示屬脈搏週期之上述脈搏起點。此外’顯不209表示那些 特徵量之具體數値。如此，藉由在顯不畫面顯示代表波形 及和抽出代表波形之處理相關的資訊，而弄清楚該代表波 形係如何抽出者、或該代表波形位於整體的波形之中的哪 -21 - 200816953 個位置（第幾個），並可提供有益於診斷之資訊。 此外’在第1 1圖之例子，作爲脈搏量測之結果300， 雖然採用顯示針對有效的全波形所得之血壓的平均値及AI 値的平均値，但是亦可作成顯示針對代表波形所得之血壓 値及AI値。藉由如此地顯示，而從顯示畫面可得到代表波 形與該代表波形相關之數値。 在以上之實施形態，作爲該指標，使用各種特徵量的 φ 平均値’而作爲和該指標之關係，針對各特徵量算出和平 均値的差分，對所有特徵量將整體上最接近平均値的波形 決定爲代表波形。依此方式，可將整體上接近平均且安定 之波形作爲代表波形。 該指標未限定爲平均値，可使用其他的値。和採用平 均値作爲指標的情況相同，若目的在於將安定之波形作爲 代表波形時，作爲其他的指標，可使用中央値或頻率最大 # 更具體而言，在指標採用中央値的情況，代表波形檢 索處理部1 09將從特徵量記憶部1 07所讀出之特徵量進行 正規化後’對於一個脈搏之波形添加該多種已進行正規化 的特徵量，並按照大小順序排列，將大小順序位於中央之 波形決疋爲代表波形。 又’在採用頻率最大値作爲指標的情況，代表波形檢 索處理部1 09將從特徵量記憶部1 07所讀出之特徵量進行 正規化後’對於一個脈搏之波形添加該多種已進行正規化 的特徵量’並畫成圖形等，確認其分布，而將位於特徵量 -22- 200816953 總和存在最多之範圍中的波形決定爲代表波形。 又’作爲指標之其他的具體例，亦可將關於該多種特 徵量中的任意種類之特徵量的任意臨限値設爲指標。例 如’可將AI値爲1 〇%以上之臨限値設爲指標。在此情況， 代表波形檢索處理部1 09係將所讀出之特徵量爲臨限値以 上的波形、臨限値以下之波形、或從臨限値位於既定範圍 中的波形決定爲代表波形。藉由使用這種指標，選擇特徵 性波形，並可抽出有益於特定之診斷的特徵波形。 又，作爲指標之其他的具體例，可使用最小値或最大 値。例如，可將基線變動率之最小値或AI値之最大値使用 作爲指標，在此情況，代表波形檢索處理部1 09將所讀出 之特徵量滿足該條件的波形決定爲代表波形。藉由使用這 種指標，可選擇安定之波形或選擇特徵性波形。 此外，在代表波形檢索處理部1 09，亦可決定將任一 個値用作指標。作爲決定指標的方法，可列舉和上述之決 定加權係數的方法相同之方法。即，作爲第1方法亦可根 據操作者之操作的操作信號決定，亦可從被檢驗者的病歷 或診斷目的等之醫生等的使用者的輸入資訊、或年齡、性 別等之與被檢驗者相關的輸入資訊自動地進行設定。 [變形例] 在變形例之脈搏檢測裝置，在代表波形檢索處理部1 09 以2階段進行代表波形的檢索處理。 第1 2圖係在變形例之脈搏檢測裝置’表示在步驟S7 在代表波形檢索處理部1 09所執行之代表波形檢索處理的 -23- 200816953 一例之流程圖。 參照第1 2圖，在變形例之脈搏檢測裝置，作爲第1階 段的檢索，如上述所示，將在各種特徵量之平均値作爲第 1階段的指標，並進行該步驟S701〜S709之處理，在步驟 S712中將在代表波形決定部1 098對於一個脈搏的波形之 特徵量的差分總和位於從平均値算起的既定範圍內之波 形，作爲下一第2階段之檢索的對象波形來分類。 此外，在變形例，代表波形決定部1 09 8例如將AI値 之最大値作爲第2階段的指標，並對在步驟S7 1 2所分類之 對象波形進行第2階段之檢索，再從該對象波形中將AI値 爲最大的波形決定爲代表波形（步驟S7 13)。 如此，在變形例的脈搏檢測裝置，以2階段或2階段 以上之多階段從多個脈搏波形中檢索代表波形並予以抽 出。在多個階段的檢索中，預先從多個脈搏波形中將與指 標的關係之條件一致者作爲對象波形進行分類，並將在和 其他指標的關係之條件一致者抽出爲代表波形，藉此，可 將例如雜訊小、且具有突出之AI値的脈搏波形等之更合乎 目的之波形抽出爲代表波形。因而，可抽出更有益於診斷 之代表波形。 此外，在上述的脈搏檢測裝置之屬抽出代表_波形之方 法的生物信號波形抽出方法，未限定爲脈搏的分析，亦可 採用針對因心臟收縮而產生的第1波形和因舒張而產生的 第2波形所合成之其他的所有生物信號波形，來抽出代表 波形的處理。 -24- 200816953 又，亦可提供用以實現在本實施形態之脈搏檢測裝置 的生物信號波形抽出方法之程式。這種程式利用附屬於電 腦之軟碟、CD — R〇M(Compact Disk — Read Only Memory)、 ROM、RAM以及記憶卡等的電腦可讀取之記錄媒體來進行 記錄，亦可提供作爲程式產品。或者，由在電腦所內建之 硬碟等的記錄媒體來進行記錄，亦可提供程式。又，利用 經由網路之下載，亦可提供程式。 此外，本發明之程式，亦可在作爲電腦的作業系統（〇S) 之一部分所提供的程式模組中，按照既定的排列在既定之 時序呼叫必要的模組並執行處理。在此情況，程式本身未 含有該模組，而和OS協同動作並執行處理。在本發明之程 式亦可包含有未含有這種模組的程式。 又，本發明的程式，亦可被裝入例如一般之脈搏量測 所需的程式等之其他程式之一部分而被提供。在此情況， 程式本身亦未含有該其他程式所含之模組，而和其他的程 式協同動作並執行處理。在本發明之程式亦可包含有被裝 入這種其他程式之程式。 所提供之程式產品被安裝於硬碟等的程式儲存部並執 行。此外，程式產品包含有程式本身和已記錄程式之記錄 媒體。 這次所揭示之實施形態在所有的事項都是舉例表示， 而不是用以限制的。本發明之範圍不是如上述的說明，而 利用專利申請範圍表示，包含有和專利申請範圍相當的意 義及在範圍內之所有的變更。 -25- 200816953 【圖式簡單說明】 第1圖係表示脈搏檢測裝置之裝置構造的具體例之 圖。 第2圖係表示脈搏檢測裝置之功能構造的具體例之方 塊圖。 第3圖係表示脈搏波形之具體例的圖。 第4圖係表示脈搏波形之具體例的圖。 第5圖係表示脈搏波形之具體例的圖。 第6圖係表示脈搏波形之具體例的圖。 第7圖係表示代表波形檢索處理部1 09所含的功能構 造之一的具體例之方塊圖。 第8圖係表示在脈搏檢測裝置之處理的流程圖。 第9圖係表示在步驟S 3所執行之脈搏劃分處理的一例 之流程圖。 第10圖係表示在步驟S7所執行之代表波形檢索處理 的一例之流程圖。 第11圖係說明在顯示畫面之顯示內容的圖。 第1 2圖係在變形例之脈搏檢測裝置，表示在步驟S7 所執行之代表波形檢索處理的一例之流程圖。 【主要元件符號說明】 1 感測器單元 3 顯示單元 5 氣壓管 7 固定座單元 -26- 200816953The I-pulse period means a period of a divided pulse waveform as an index of deformation or arrhythmia of a time component of a waveform shape caused by an influence of physical activity, noise, or the like. As an example of a method of determining the period of the pulse waveform, the following method can be mentioned. Referring to Fig. 5, when the maximum chirp of the original waveform is detected, the minimum point (PB point) corresponding to the Eero cross point from the next rising direction is referred to the zero crossing point with the previous rising direction. The original waveform up to the corresponding minimum point (PA point) has a maximum point (PP point) therebetween, and the PA point is determined to be "rise" as long as it is confirmed that the PA point is the smallest between the PA point and the PB point. Point, in other words, a pulse start of the pulse. Further, the length D from the PA point to the PB point is determined as the period of the pulse waveform of one pulse. Further, the ascending sharpness MSP system significantly indicates the pushing force of the wristband 18 and the evaluation of the waveform distortion. In the case where the rising point of the change in the output of the semiconductor pressure sensor 19 is sharp, it means that the pushing force of the pressing wrist band 18 is appropriate, and when the rising point is close to a flat change, the pushing force of the wrist band 18 is pushed. Inappropriate, the subcutaneous arteries at the measurement site are pressurized to the necessary -15-200816953 値 or more. The ascending sharpness MSP is a digital enthalpy that sharpens the sharpness of the rising point of the output change of the semiconductor pressure sensor 19. For example, referring to Fig. 6, the time interval from the point where the amplitude of the predetermined division point (TDIA) set at the rising position of the pulse waveform is increased by 10% of the maximum amplitude is defined as Ta, and When the time from the end point of the point to the peak 値 of the pulse waveform is defined as Tb, the rising sharpness MSP can be obtained from the ratio of Ta and Tb, for example, MSP = Ta/Tb. The baseline change rate indicates the change in the baseline of the adjacent pulse waveform as an indicator of the deformation of the amplitude component of the waveform shape caused by the influence of physical activity, noise, and the like. The baseline rate of change can be obtained from the ratio of the maximum pulse rate and the difference between the "pulse start point" and the "pulse start point" of the next pulse. For example, in the case where the pulse waveform shown in Fig. 6 is detected, if the pressure difference between the pulse start point S1 and the pulse maximum point S2 is A, the pressure difference between the pulse start point S1 and the pulse start point S3 of the next pulse is set. For B, the baseline rate of change (%) = B/Axl 00. Further, the "pulse origin" can be determined in the same manner as described in the description of the pulse period. The feature amount storage unit 107 stores the five kinds of feature quantities calculated for each pulse. In addition, the type of the feature amount calculated by the feature amount calculation unit 105 is not limited to the above five types, and at least two of them may be used. Also, other feature quantities can be calculated. The other characteristic quantity includes, for example, a time interval between the maximum value of the forward wave and the maximum point of the reflected wave, which is an index for evaluating the time component of the reflected wave or the degree of hardening of the arterial tube wall, or the rising point of the forward wave and the rising of the reflected wave. A conventional indicator such as TR (Traveling time to Reflected wave). In addition, all the feature quantities that are useful for diagnosis using the pulse waveform -16-200816953 can be used. Referring to Fig. 7 'in the representative waveform search processing unit 1 to 9, the feature amount reading unit 1 09 1 ' reads out the stored feature amount from the feature amount memory unit 1 〇 7 for each pulse waveform; the average calculation unit 1 092, the average 値 is calculated for each feature amount and set as an index; the difference calculation unit 1 〇 93 calculates the difference between the feature metrics and the average 値 as the relationship with the index; the normalization processing unit 1 094 ' The calculated difference 値 is normalized in accordance with the type of the feature amount; the coefficient determining unit 1095 determines the weighting coefficient related to the type of the feature amount considered when extracting the representative waveform; the weighting processing unit 1 〇 9 6 Each of the pulse waveforms weights the normalized difference 使用 using the determined weighting coefficient; the addition processing unit 1 097 calculates the total of the pulse waveforms of the weighted difference ;; and the representative waveform determining unit 1 098 The representative waveform is determined from the total of the differences 各 of the respective pulse waveforms. Here, the weighting means the degree of importance of the feature amount used when extracting the representative waveform using the plurality of feature amounts as described above, for example, by setting the weighting coefficient to 1, and corresponding to when extracting the representative waveform. The feature quantity directly uses the 値; by setting the weighting coefficient to 0.5, 50% of the 値 is used for the feature quantity corresponding to when the representative waveform is extracted; by setting the weighting coefficient to 0, without using and extracting The amount of features corresponding to the waveform. Therefore, by setting a combination of weighting coefficients, it is also possible to extract a representative waveform using only one of the calculated plurality of feature amounts (e.g., AI 値). Further, the representative waveform can be extracted by using all of the plurality of calculated feature amounts on average. Further, according to a certain purpose, the specific feature amount among the plurality of calculated feature amounts is particularly emphasized, and the importance degree of the other feature amounts is changed to -17 - 1095 200816953, and the representative waveform can be extracted. This weighting coefficient is determined by the coefficient determining unit. The method of determining the weighting coefficient by the coefficient determining unit 109 can be exemplified as follows. As the first method of determining the weighting coefficient by the coefficient determining unit 1095, the operation unit 24 is provided with a specific button. Or, the screen is set on the display unit 25. The user of the doctor or the like performs the operation of the button for the purpose of diagnosis or performs the operation according to the setting screen, and inputs the weight corresponding to each amount (number or importance (small, medium, large) selection, etc.). The system 1 095 accepts the operational signals of those operations and determines the coefficient of the coefficient for each feature. In the second method, the input amount of the feature amount is automatically set from the input information of the user such as the medical record of the examinee or the doctor of the diagnosis in the predetermined area such as the RAM 13 before the measurement. method. Or, a method of automatically setting a weighting coefficient for a symptom type having a high incidence rate of chronology and gender, such as age, gender, and the like, may be cited. When the weighting coefficient is automatically set, the input information and the weighting coefficient for each feature amount are previously recalled in the memory area of the ROM 12 or the like, and the coefficient determining unit 1 095 refers to the memory area as needed, and reads the corresponding information. The weighting coefficient is a flowchart showing a process of displaying a representative waveform after the pulse in the pulse detecting device of the present embodiment. The processing shown in the flowchart of Fig. 8 is realized by the CPU 11 accessing and reading the program to the ROM 12, re-executing the RAM 13 and executing it, and controlling the functions shown in Figs. 2 and 7. ▼ method, before the display of the number of features to determine the number of people who have the right to remember, etc. The system should be determined. :Measurement map: open at i, by -18- 200816953 Referring to Fig. 8, first, the pulse pressure measurement acquisition unit 1 〇丨 obtains the pressure signal from the semiconductor pressure sensor 19, and measures the pulse ( Step S1). The pulse measured in step s1 is divided into pulses of the respective pulses in the pulse division processing unit 3 (step S3), and the feature amount calculation unit 1〇5 calculates the five characteristics for each pulse waveform by the above-described method. Quantity (step s5). The processing of the steps S1 to S5 can also be performed in parallel with the measurement of the pulse. In this case, the calculation of the feature amount in step S5 is performed for each pulse waveform until the end of the pulse measurement. When the measurement is completed, the representative waveform search processing unit 109' searches for the representative waveform based on the five kinds of characteristic amounts of the respective pulse waveforms calculated in step S5, and extracts the corresponding pulse waveform (step S7). The representative waveform display control unit 1 1 1 generates and displays a display signal for displaying the pulse waveform extracted as the representative waveform in step S7 as a measurement result on the display unit 25 together with other calculations ( Step S9). As the pulse division processing executed in this step S3, an example of the flowchart in Fig. 9 is shown. Further, in the following, this processing will be described as an analysis process after the channel of the fixed multiplexer 20 is used. Referring to Fig. 9, first, when the pressure signal has been sent from the semiconductor pressure sensor 19 (S301), the multiplexer 20 selects the sensing of the sensor element in accordance with the channel fixed according to the control signal from the CPU 11. Signal, and input to amplifier 2 1. Then, the input pressure signal is amplified by the amplifier 21 to a predetermined frequency (S03), and the characteristic filter 22 performs analog filter processing (S305). At this time, the characteristic variable filter 22 blocks the -19-200816953 signal component of 取样/2 or more of the sampling frequency. If the sampling frequency is 500 Hz, for example, the signal component whose blocking frequency exceeds 250 Η z is blocked. That is, here, the signal component of the pulse waveform outside the specification range is removed. The pressure signal of the characteristic variable filter 22 has been converted into a digital signal (S307) by the A/D converter 23, and digital filter processing for extracting a frequency of a predetermined range is performed in order to remove noise or the like ( S3 09). Then, the A/D converter 23 transmits the digitalized pressure signal to the CPU 11. The pulse division processing unit 103 differentially differentiates the pulse waveform obtained from the pressure signal by taking the difference of each data with respect to the pressure signal received from the A/D converter 23 (S3 11). Then, the pulse division processing unit 1〇3 extracts the pulse waveform of one pulse based on the differential result of the step S 3 1 1 and extracts the pulse waveform of one pulse (S313). Fig. 10 is a flowchart showing an example of representative waveform search processing executed by the representative waveform search processing unit 109 in step S7. With reference to the tenth figure, in the step S7, the average amount of the feature amounts read by the feature amount reading unit 1 〇9 1 from the feature amount storage unit 107 is calculated by the average calculating unit 1092 for each feature amount (step S701). ). Next, the difference calculation unit 1 093 calculates a difference from the average 对 for each feature amount (step S703). The normalization processing unit 1 094 normalizes the difference calculated in step S703 (step S705). The normalization process here is a general normalization process, and is not limited to a specific process in the present invention. The difference normalized in step S7 05 is weighted by the weighting processing unit 1 096 using the weighting coefficient determined by the coefficient determining unit 1 095 (step S707). Specifically, the difference which has been normalized in step S705 is multiplied by the weighting coefficient determined by the coefficient -20-200816953 determining unit 1 095 for each feature amount in accordance with the type of the feature amount, thereby performing the weighting. The addition processing unit 1 097 adds the difference between the five kinds of feature amounts of the waveforms of one pulse subjected to the weighting process, and calculates the sum of the differences of the waveforms (step S709). Then, in the representative waveform determining unit 1 098, the waveform having the smallest sum for the sum, that is, the waveform closest to the average 値 of the entire five kinds of feature amounts as a whole is determined as a representative waveform, and is extracted (step S7 11). Fig. 1 is a view for explaining the display content displayed on the display screen based on the display signal from the representative waveform display control unit 111 in this step S9. Referring to Fig. 1, on the display screen, as a representative waveform, the pulse waveform 200 of one pulse extracted in the process is displayed, and the average 値 of the blood pressure 値 or AI is taken as the result of the pulse measurement 300, and the display is effective. The average value of the blood pressure obtained from the full waveform and the average 値 of the AI. Further, preferably, in the display screen, as a display indicating a process for extracting a representative waveform, the representative waveform including the extracted representative waveform indicates a display 20 1 of the first waveform among the valid waveforms, or indicates that the representative is extracted. The characteristic position of the feature used in the waveform is the characteristic position on the waveform or the mark 203: 205, 207, or the display 209 of the feature quantity. In the example of Fig. 11, when the representative waveform is extracted, AI 値, ET 値 and the pulse period among the five kinds of feature quantities are used, and the symbol 2 03 indicates the above-mentioned level a and bit used in calculating AI 値The position of the quasi-b, the symbol 205 indicates the length c corresponding to ET, and the symbol 207 indicates the pulse start point of the pulse period. In addition, 'not shown 209 indicates the specific number of those feature quantities. In this way, by displaying the information related to the process of representing the waveform and extracting the representative waveform, it is clarified which of the representative waveforms is extracted, or which representative waveform is located in the overall waveform - 21 - 200816953 Location (several) and can provide information useful for diagnosis. Further, in the example of Fig. 1, as the result 300 of the pulse measurement, although the average 値 of the blood pressure obtained for the effective full waveform and the average 値 of the AI 値 are used, it is also possible to display the blood pressure obtained for the representative waveform.値 and AI値. By thus displaying, the number of 代表 representing the waveform associated with the representative waveform can be obtained from the display screen. In the above embodiment, as the index, the φ average 値' of various feature amounts is used as the relationship with the index, and the difference between the average 値 is calculated for each feature amount, and the total eigenvalue is the closest to the average 所有. The waveform is determined to represent the waveform. In this way, a waveform that is nearly average and stable as a whole can be taken as a representative waveform. This indicator is not limited to an average 値, and other 値 can be used. The same as the case where the average 値 is used as the index. If the purpose is to use the stable waveform as the representative waveform, as the other indicator, the center 値 or the frequency maximum can be used. More specifically, the waveform is represented by the central 値 when the indicator is used. The search processing unit 109 normalizes the feature amount read from the feature amount storage unit 107, and adds the plurality of feature quantities that have been normalized to the waveform of one pulse, and arranges them in order of size, and the size order is located. The waveform of the center is representative of the waveform. In the case where the maximum frequency 値 is used as the index, the representative waveform search processing unit 109 normalizes the feature amount read from the feature amount storage unit 107, and adds the plurality of patterns to the pulse waveform. The feature quantity ' is drawn into a graph or the like to confirm its distribution, and the waveform in the range in which the sum of the feature quantity -22-200816953 is most present is determined as a representative waveform. Further, as another specific example of the index, any threshold 特征 of the feature amount of any of the plurality of characteristic amounts may be used as an index. For example, 'the threshold 可 can be set to 1 〇% or more. In this case, the representative waveform search processing unit 109 determines the waveform to be read as a waveform having a threshold value or more, a waveform below a threshold, or a waveform having a threshold value within a predetermined range as a representative waveform. By using such an indicator, a characteristic waveform is selected and a characteristic waveform that is useful for a particular diagnosis can be extracted. Further, as another specific example of the index, the minimum 最大 or the maximum 値 can be used. For example, the minimum value of the baseline change rate or the maximum value of AI 値 can be used as an index. In this case, the representative waveform search processing unit 119 determines the waveform in which the read feature amount satisfies the condition as a representative waveform. By using this indicator, you can choose a stable waveform or select a characteristic waveform. Further, in the representative waveform search processing unit 109, it is also possible to decide to use any one of them as an index. As a method of determining the index, the same method as the above method of determining the weighting coefficient can be cited. In other words, the first method may be determined based on an operation signal of an operator's operation, or may be input information from a user such as a doctor's medical record or a diagnosis purpose, or an age, a gender, or the like with the examinee. The relevant input information is automatically set. [Modification] In the pulse wave detecting device according to the modification, the representative waveform search processing unit 109 performs the search processing of the representative waveform in two stages. Fig. 1 is a flowchart showing an example of -23-200816953 in the representative waveform search processing executed by the representative waveform search processing unit 109 in step S7. With reference to Fig. 12, the pulse detecting device according to the modification is the first stage search, and as shown above, the average value of the various feature amounts is used as the index of the first stage, and the processing of steps S701 to S709 is performed. In step S712, the sum of the difference sums of the feature amounts of the waveforms of one pulse of the representative waveform determining unit 1 098 is within a predetermined range from the average chirp, and is classified as the target waveform of the next second stage search. . Further, in the modified example, the representative waveform determining unit 109 selects, for example, the maximum 値 of the AI値 as the index of the second stage, and performs the second-stage search on the waveform of the object classified in the step S71, and then proceeds from the object. The waveform in which the AI 値 is the largest in the waveform is determined to represent the waveform (step S7 13). As described above, in the pulse detecting apparatus according to the modification, the representative waveform is searched for from the plurality of pulse waveforms in two stages or two or more stages and extracted. In the search for a plurality of stages, the condition that matches the condition of the index is classified as the target waveform from among the plurality of pulse waveforms, and the condition that matches the relationship with the other index is extracted as the representative waveform. A more desirable waveform such as a pulse waveform having a small noise and having an outstanding AI 抽 can be extracted as a representative waveform. Thus, a representative waveform that is more beneficial for diagnosis can be extracted. Further, the method of extracting the biosignal waveform by the method of extracting the representative waveform described above is not limited to the analysis of the pulse, and the first waveform generated by the contraction of the heart and the first due to the relaxation may be employed. 2 All other biosignal waveforms synthesized by the waveform are used to extract the processing representing the waveform. Further, a program for realizing the biosignal waveform extraction method of the pulse wave detecting device of the present embodiment can be provided. This program is recorded by a computer-readable recording medium such as a floppy disk attached to a computer, a CD-ROM (Compact Disk - Read Only Memory), a ROM, a RAM, and a memory card, and can also be provided as a program product. . Alternatively, recording may be performed by a recording medium such as a hard disk built in a computer, and a program may be provided. Also, the program can be provided by downloading via the network. Further, the program of the present invention can also call a necessary module and execute processing at a predetermined timing in a program module provided as a part of a computer operating system (〇S) in accordance with a predetermined arrangement. In this case, the program itself does not contain the module, but cooperates with the OS and performs processing. The program of the present invention may also include a program that does not include such a module. Further, the program of the present invention can be provided by being incorporated in, for example, a part of another program such as a program required for general pulse measurement. In this case, the program itself does not contain the modules contained in the other programs, and cooperates with other programs to perform processing. The program of the present invention may also include a program loaded with such other programs. The supplied program product is installed in a program storage unit such as a hard disk and executed. In addition, the program product contains the program itself and the recording medium of the recorded program. The embodiments disclosed herein are illustrative of all matters, and are not intended to be limiting. The scope of the present invention is defined by the scope of the invention, and is intended to be -25-200816953 [Brief Description of the Drawings] Fig. 1 is a view showing a specific example of the structure of the apparatus of the pulse detecting device. Fig. 2 is a block diagram showing a specific example of the functional configuration of the pulse detecting device. Fig. 3 is a view showing a specific example of a pulse waveform. Fig. 4 is a view showing a specific example of a pulse waveform. Fig. 5 is a view showing a specific example of a pulse waveform. Fig. 6 is a view showing a specific example of a pulse waveform. Fig. 7 is a block diagram showing a specific example of one of the functional configurations included in the waveform search processing unit 109. Fig. 8 is a flow chart showing the processing of the pulse detecting means. Fig. 9 is a flow chart showing an example of the pulse division processing executed in step S3. Fig. 10 is a flow chart showing an example of representative waveform search processing executed in step S7. Fig. 11 is a view for explaining the display contents on the display screen. Fig. 1 is a flowchart showing an example of the representative waveform search processing executed in step S7 in the pulse rate detecting apparatus according to the modification. [Main component symbol description] 1 Sensor unit 3 Display unit 5 Air tube 7 Mounting unit -26- 200816953
11 CPU 12 ROM 13 RAM 14 控 制 電 路 15 加 壓 泵 16 負 壓 泵 17 切 換 閥 18 推 壓 腕 帶 19 半 導 體 壓 力 感 測 器 20 多 X 器 21 放 大 器 22 特 性 可 變 濾 波 器 23 A/D 轉 換 器 24 操 作 部 25 顯 示 部 101 脈 搏 量 測 値 取 得 部 103 脈 搏 劃 分 處 理 部 105 特 徵 里 算 出 部 107 特 徵 量 記 憶 部 109 代 表 波 形 檢 索 處 理 部 111 代 表 波 形 顯 示 控 制 部 105 0 A] :算出部 105 1 ET算出部 1052 脈 搏 週 期 算 出 部 -27- 200816953 1053 MSP算出部 1054 基線變動率算出部 1091 特徵量讀出部 1092 平均算出部 1093 差分算出部 1094 正規化處理部 1095 係數決定部 1096 加權處理部 1097 加法處理部 1098 代表波形決定部 -2811 CPU 12 ROM 13 RAM 14 Control circuit 15 Pressurizing pump 16 Negative pressure pump 17 Switching valve 18 Pushing wrist strap 19 Semiconductor pressure sensor 20 Multiple X 21 Amplifier 22 Characteristic variable filter 23 A/D converter 24 Operation unit 25 Display unit 101 Pulse rate measurement acquisition unit 103 Pulse division processing unit 105 Characteristic calculation unit 107 Characteristic quantity storage unit 109 Representative waveform search processing unit 111 Representative waveform display control unit 105 0 A] : Calculation unit 105 1 ET calculation Section 1052 Pulse Period Calculation Unit -27- 200816953 1053 MSP calculation unit 1054 Baseline variation rate calculation unit 1091 Characteristic quantity reading unit 1092 Average calculation unit 1093 Difference calculation unit 1094 Normalization processing unit 1095 Coefficient determination unit 1096 Weighting processing unit 1097 Addition processing Section 1098 represents waveform decision section-28