TW201322959A - Apparatus and method of personalized ECG - Google Patents

Abstract

An apparatus and a method of a personalized ECG mainly include a detecting unit, a capturing unit, a database, a recognizing unit and a determination unit. The method is capable of detecting, capturing and recognizing a personalized ECG template in real time and performing a normal/abnormal determination of a personalized ECG signal toe generate a determined result. A database for a personalized ECG data is established by the determined result. The database can record a data history of an individual patient's ECG and expand into other database for integrity and reference.

Description

Personalized electrocardiogram processing device and method

A personalized electrocardiogram analysis technique, in particular, relates to a personalized ECG processing apparatus and method.

With the changes in people's lifestyles, such as eating more, less movement, smoking, life stress, stress and other factors, cardiovascular disease is increasing and rejuvenating. For heart disease patients, the reconstruction of lifestyle will be the key to improving survival. Among them, cardiac rehabilitation exercise is the main key to increase myocardial tolerance. At present, many medical centers and foreign professional organizations in China regard the heart rehabilitation exercise as one of the treatments after heart disease. The aforementioned cardiac rehabilitation exercise will achieve the purpose of monitoring the current electrocardiographic signal of the patient through the exercise electrocardiogram.

However, when current exercise patients use exercise ECG, the ECG monitoring system can only record the current ECG and make an easy interpretation in the case of rehabilitation exercise. Then, through the professional diagnosis of the heart rehabilitation doctor, it is judged whether the rehabilitation exercise has a cardiac rehabilitation effect on the patient. The aforementioned monitoring state does not immediately detect, interpret, and alert the heart rhythm before it begins to mutate in a safe rehabilitation environment.

In addition, for the current 24-hour Holter monitoring, there are many algorithms to analyze abnormal ECG waveforms afterwards, rather than analyzing and processing them in a timely manner. In addition, the current products of many ECG instrument manufacturers only detect the stability of the ECG waveform and make basic and simple judgments. There is no real-time waveform comparison function and more stringent normal abnormal rule function.

In view of the above problems, the present disclosure provides a personalized ECG processing apparatus and method for increasing the immediacy, rigor and reliability of electrocardiogram analysis.

A personalized ECG processing device of the present disclosure mainly includes a detecting unit, a capturing unit, a data base, an identifying unit and a determining unit. The detection unit receives and detects the complete personal electrocardiogram signal to search for the QRS wave signal. The capturing unit is coupled to the detecting unit for receiving the at least two QRS wave signals to calculate a heartbeat interval, and then obtaining a personalized ECG template corresponding to the complete ECG signal according to the heartbeat interval. The database is used to store a plurality of preset ECG templates. The identification unit is coupled to the detection unit and the data base for receiving the personalized ECG template and the plurality of preset ECG templates, and simultaneously identifying the personalized ECG template and the preset ECG template in a parallel processing manner to generate the identification result. The determining unit is coupled to the identification unit for receiving the identification result, and compares the identification result with a plurality of personalized positive abnormality determining mechanisms in a parallel processing manner to generate a personalized personalized judgment result and establish an exclusive personalization. A database of ECG data patterns.

A personalized ECG processing method of the present disclosure includes the following steps. Detect a complete personal ECG signal to generate a QRS wave signal. Receive and detect at least two QRS wave signals to calculate the heartbeat interval. According to the heartbeat interval, a personalized ECG template corresponding to the complete personal electrocardiogram signal is obtained. A plurality of preset ECG templates are obtained from the database, and at the same time, a parallel processing method is used, for example, Single Timing Single-Input-Multiple-Output Dynamic Time Warping (Single Timing Single-Input-Multiple-Output Dynamic Time Warping, ST-SIMO DTW), which identifies the personalized ECG template and the preset ECG template to generate identification results. At the same time, the identification result is parallelized with a plurality of personalized positive abnormality judgment mechanisms, for example, Single Timing Single-Input-Multiple-Output Decision Rule (Single Timing Single-Input-Multiple-Output Decision Rule, ST-SIMO DR) ), produce a judgment result and analyze whether the personalized ECG template is normal or abnormal. In this way, the immediacy, rigor and reliability of personalized ECG analysis can be effectively increased.

The features and implementations of the present disclosure are described in detail below with reference to the drawings.

Please refer to "Figure 1" for a block diagram of the personalized ECG processing device disclosed herein. The personalized ECG processing device 100 mainly includes a detecting unit 110, a capturing unit 120, a data library 130, an identifying unit 140, a determining unit 150, storage units 160, 170, and a processing unit 180.

The detecting unit 110 receives and detects the complete personal electrocardiogram signal to search for the QRS wave signal, as shown in "Fig. 2". For example, the user (patient) body surface can be connected by an electrocardiograph patch to detect a complete personal electrocardiogram signal. After the detecting unit 110 receives the complete personal electrocardiogram signal, the detecting unit 110 uses the Pan-Tompkins algorithm to search for the position of the QR wave signal of the Q wave, the R wave, and the S wave in the complete personal electrocardiogram signal. In turn, the QRS wave signal is detected and output. The main function of the detecting unit 110 is that the personal electrocardiogram signal is transmitted through a bandpass filtering program to strengthen the QRS wave signal, suppress P waves, T waves and other noises which may cause misjudgment, and successively pass through a series of processing procedures of differential, square and integral. After that, each QRS wave signal can generate a corresponding clock type waveform; in addition, the QRS wave signal detection correct probability is increased by an elastically adjustable threshold.

The capturing unit 120 is coupled to the detecting unit 110 for receiving the at least two QRS wave signals to calculate a heartbeat interval to obtain a personalized ECG template corresponding to the complete personal electrocardiogram signal. In this embodiment, the capturing unit 120 can obtain the heartbeat interval by using the interval time generated by the R waves in the two QRS wave signals, for example, an R-R Interval. Then, the capturing unit 120 estimates the generation time points of the P wave and the T wave according to the calculated heartbeat interval, and then extracts the personalized electrocardiogram template with the PQRST wave corresponding to the complete personal electrocardiogram signal, such as "Fig. 2" Shown.

Figure 2 is a waveform diagram of the complete personal ECG signal of this disclosure. Wherein, P1 and P2 represent P waves; Q1 and Q2 represent Q waves; R1 and R2 represent R waves; S1 and S2 represent S waves; T1 and T2 represent T waves; and RR represents a heartbeat interval, that is, an R-R interval.

The database 130 is used to store a plurality of preset electrocardiogram templates, and the preset electrocardiogram templates may be various electrocardiogram templates before the heart rhythm variation. These preset ECG templates can be created by the user in advance and stored in the database 130.

The identification unit 140 is coupled to the detection unit 120 and the data library 130 for receiving the personalized electrocardiogram template and the plurality of preset electrocardiogram templates, and simultaneously identifying the personalized electrocardiogram template and the preset electrocardiogram template in a parallel processing manner to generate Identify the results. In this embodiment, the identification unit 140 uses an example of a Single Timing Single-Input-Multiple-Output Dynamic Time Warping (ST-SIMO DTW) comparison method. The personalized ECG sample is compared with a plurality of preset ECG samples at the same time, that is, processed in a parallelized identification manner, thereby accelerating the recognition speed.

Since the length of the time series generated by the personalized ECG template and the preset ECG template may not be the same, the present embodiment uses the dynamic time warping algorithm to map points to similar points for similarity comparison. .

In addition, further operations of the identification unit 140 are explained below. Please refer to FIG. 3, which is a schematic diagram of the operation of the identification unit 140 of the present disclosure. Among them, the template 1, the template 2, ..., the template N respectively represent different personalized ECG templates; time 1, time 2, ..., time N are the identification time of each humanized electrocardiogram template; number 1, number 2, ..., The number M indicates a different preset ECG template. Where N and M are positive integers greater than one.

Since the heartbeat will continue to be generated, the capturing unit 120 will sequentially extract the personalized electrocardiogram template of the template 1, the template 2, ..., the template N. Then, the identification unit 140 performs the similarity recognition on the currently obtained personalized ECG template with the plurality of preset ECG templates of the database 130. That is, as shown in "Fig. 3", at time 1, the identification unit 140 recognizes the personalized ECG template of the template 1 and the preset ECG template numbered from 1 to M; at time 2, The identification unit 140 identifies the personalized ECG template of the template 2 and the preset ECG template numbered from 1 to M. When the time is N, the identification unit 140 will customize the personalized ECG template of the template N. The default ECG template numbered from 1 to M is recognized. In this way, by the parallelization processing method, the identification time of N times can be effectively shortened, and the personalized ECG signal processing is more immediacy.

The determining unit 150 is coupled to the identifying unit 140 for receiving the identification result, and comparing the identification result with a plurality of personalized positive abnormality determining mechanisms to generate a determination result. In this embodiment, the determining unit 150 uses a single-input single-input-multiple-output decision rule (ST-SIMO DR) to determine the identification. The result is simultaneously compared with a plurality of personalized positive abnormality judgment mechanisms, that is, processed in a parallel manner to generate a judgment result.

The basic personalization positive abnormality determining mechanism is, for example, whether a P wave is formed in the personalized electrocardiogram template, the width of the QRS wave signal is narrow (<120 ms) or wide (>120 ms), and the relationship between the P wave and the QRS wave signal. Whether the rhythm of the heart rhythm and a variety of specific abnormal conditions are basic judgment mechanisms can be expanded into a personalized positive abnormality judgment mechanism, as shown in Table 1. That is, the personal positive abnormality determining unit 150 in the processing apparatus 100 is formed.

In Table 1, a plurality of judgment mechanisms for judging the abnormality are further enumerated. For example, the width of the QRS wave signal is wide (>120ms) and the non-P wave is generated simultaneously, the QRS wave signal and the T wave are opposite directions, the QRS wave signal exhibits a strange shape, and the rhythm pattern is irregular, but the disclosure is not limited thereto. A reference pattern of abnormal conditions is also included.

In addition, the determining unit 150 may further determine any abnormality according to any combination of abnormal condition determining mechanisms. The foregoing abnormal situation determination combination may include: two or more abnormal reference patterns appear in the adjacent ten QRS wave signals, and any two different shapes of abnormal reference patterns appear in the personalized electrocardiogram template, and the heart rhythm is greater than 160 per minute. The lower (HR>160/min) and the width of the non-P wave and the QRS wave signal are narrow (<120ms), the rhythm rule is changed into irregular, the irregularity of the heart rhythm is changed into a rule, and the personalized ECG template is There are three or more consecutive identical abnormal reference patterns, no QRS wave signals in the personalized ECG template (presenting a multi-wave baseline condition), and a heart rhythm greater than 100 beats per minute (HR > 100/min) and heart rhythm The rhythm is irregularly generated and the P wave is not generated simultaneously. The average of the first 6-8 RR intervals with no waveform generation (one line) and the width greater than 1.5 times in the personalized ECG template, and the P wave in the personalized ECG template After that, the QRS wave signal or the like does not appear, but the disclosure is not limited thereto, and the abnormal situation determination combination may further include other abnormal condition determination rules.

In the process of making the judgment, the judging unit 150 compares the identification result including the P wave formation in the personalized electrocardiogram template, the width of the QRS wave signal is narrow (<120 ms), the QR wave signal is immediately adjacent to a P wave, and the heart rhythm is regular. And no other abnormalities (as listed in Table 1) will result in a normal judgment. On the other hand, when the determining unit 150 compares the result of the determination that the recognition result does not conform to the normal situation, the specific abnormal situation is further analyzed through the parallel positive personal abnormality judgment mechanism to generate a judgment result of the specific abnormal situation, and the safety rigor of the processing device is achieved. The purpose.

In addition, the operation of the determination unit 150 is explained as follows. As shown in Figure 4, it is a schematic diagram of the operation of the judgment unit of this disclosure. Among them, the results 1, the results 2, ..., the result N respectively represent different identification results; the time 1, the time 2, ..., the time N are the judgment moments of the respective identification results; the judgment 1, the judgment 2, ..., the judgment R respectively represent Different personalization positive abnormal judgment mechanisms. Wherein N and R are positive integers greater than one.

As can be seen from "Fig. 4", at time 1, the determination unit 150 judges the identification result of the result 1 and the individualized positive abnormality determining mechanism (judgment 1 to judgment R); at time 2, judges The unit 150 judges the identification of the result 2 and the individualized positive abnormality determining mechanism; when the time is N, the determining unit 150 judges the identification result of the result N and the individualized positive abnormality determining mechanism. In this way, by the normal abnormal rule determination method of the parallelization process, the N-time determination time can be effectively shortened, and the personalized positive abnormality judgment mechanism is used, so that the ECG judgment is personalized, and the signal processing speed is more immediate. Sex. Therefore, the rigor and reliability of personal ECG signal detection analysis can be increased by the above two stages of identification and judgment.

The storage unit 160 is coupled to the determining unit 150 for storing the judgment result of the personalized normal situation. The storage unit 170 is coupled to the determining unit 150 for storing the judgment result of the personalized abnormality. The processing unit 180 is coupled to the storage unit 170 for reading the determination result of the abnormal condition in the storage unit 170. Moreover, the processing unit 180 is further configured to output the judgment result of the foregoing personalized abnormal situation to the back end device, for example, the diagnosis and treatment system of the remote medical center or the remote rehabilitation center. In this way, the patient can be sure that the rehabilitation exercise can be performed under the safety electrocardiogram monitoring, and can be discovered immediately before the heart condition causes a problem.

In addition, the determining unit 150 is further coupled to the database 130, and transmits the determination result (ie, the personalized ECG template of the abnormal situation) to the database 130, and stores or serves as a preset ECG template to expand the preset ECG template data of the database 130. . In this way, the accuracy of the personalized electrocardiogram processing device 100 for identifying the personalized electrocardiogram template and the personalization of the electrocardiogram processing device can be increased.

As illustrated by the above embodiments, a personalized ECG processing method can be summarized. Please refer to "Figure 5" for a flow chart of the personalized ECG processing method disclosed herein. In step S510, the complete personal electrocardiogram signal is detected to search for the QRS wave signal. In step S520, at least two QRS wave signals are received and calculated to calculate a heartbeat interval. In step S530, a personalized ECG template corresponding to the complete personal electrocardiogram signal is obtained according to the heartbeat interval. In step S540, a plurality of preset electrocardiogram templates are obtained from the database, and the parallelized dynamic time warping processing method is used to identify the personalized electrocardiogram template and the preset electrocardiogram template to generate the identification result. In step S550, for example, the normal abnormal rule processing manner is parallelized, and the identification result and the plurality of personalized positive abnormality determining mechanisms are simultaneously determined to generate a determination result. In step S560, when the determination result is normal, the determination result is stored in the first storage unit. In step S570, when the determination result is an abnormal condition, the determination result is stored in the second storage unit and output. In step S580, when the determination result is abnormal, the determination result is stored in the database to expand the preset ECG template data of the database.

The personalization positive abnormality judgment mechanism mentioned above includes whether there is a P wave in the personalized electrocardiogram template, the width of the QRS wave signal, the relationship between the P wave and its corresponding QRS wave signal and the regularity of the heart rhythm, and various specific abnormalities for further analysis. . In addition, the foregoing identification processing method can use a parallel dynamic time warping algorithm of a single input multiple output at a single time, and simultaneously recognize the similarity between the personalized ECG template and the preset ECG template in a parallel processing manner, and the foregoing The judgment processing method can be exemplified by using a parallel normal abnormality judgment mechanism of a single input multiple output at a single moment, and at the same time, the identification result is a normal situation or an abnormal situation by the identification result and the personalized positive abnormality judgment mechanism.

The present disclosure is disclosed in the foregoing embodiments, and is not intended to limit the disclosure. Any subject matter of the present invention can be modified and retouched without departing from the spirit and scope of the disclosure. The scope of patent protection shall be subject to the definition of the scope of the patent application attached to this specification.

100. . . Personalized ECG processing device

110. . . Detection unit

120. . . Capture unit

130. . . database

140. . . Identification unit

150. . . Judging unit

160, 170. . . Storage unit

180. . . Processing unit

P1, P2. . . P wave

Q1, Q2. . . Q wave

R1, R2. . . R wave

S1, S2. . . S wave

T1, T2. . . T wave

RR. . . Heartbeat interval

Sample 1~Model N Personalized ECG Sample Board

Time 1~Time N The identification time of each humanized ECG template and the judgment time of each identification result

No. 1~No. M Preset ECG template

Judgment 1~Judgement R Personalization Positive Abnormal Judgment Mechanism

Figure 1 is a block diagram of a personalized ECG processing apparatus of the present disclosure.

Figure 2 is a waveform diagram of the complete personal electrocardiogram signal disclosed herein.

Figure 3 is a schematic diagram of the operation of the identification unit of the present disclosure.

Figure 4 is a schematic diagram of the operation of the judging unit of the present disclosure.

Figure 5 is a flow chart of the personalized ECG processing method disclosed herein.

100. . . Electrocardiogram processing device

110. . . Detection unit

120. . . Capture unit

130. . . database

140. . . Identification unit

150. . . Judging unit

160, 170. . . Storage unit

180. . . Processing unit

Claims (14)

A personalized ECG processing device includes: a detecting unit for receiving and detecting a complete personal electrocardiogram signal to generate a QRS wave signal; a capturing unit coupled to the detecting unit for receiving at least two QRS wave signals to calculate a heartbeat interval, and then according to the heartbeat interval to obtain a humanized electrocardiogram template corresponding to the complete personal electrocardiogram signal; a database for storing a plurality of preset ECG templates; an identification unit coupled The capturing unit and the data library are configured to receive the personalized electrocardiogram template and the preset electrocardiogram templates, and simultaneously identify the personalized electrocardiogram template and the preset electrocardiogram templates in a first parallel processing manner to generate a recognition result; and a determination unit coupled to the identification unit for receiving the identification result, and determining the identification result in a second parallelization manner simultaneously with a plurality of personalized positive abnormality determination mechanisms to generate A judgment result. The personalized electrocardiogram processing device of claim 1, further comprising: a first storage unit coupled to the determining unit for storing the result of the normal situation; a second storage unit coupled to the a determining unit, configured to store the abnormal result; and a processing unit coupled to the second storage unit for reading and processing the determination result. The personalized electrocardiogram processing device of claim 1, wherein the determining unit is coupled to the database, and storing the judgment result of the abnormal condition in the database to expand the preset electrocardiogram of the database. Template. The personalized electrocardiogram processing device of claim 1, wherein the personalized positive abnormality determining mechanism of the determining unit comprises: whether there is a P wave in the personalized ECG template, a width of the QRS wave signal, The relationship between the P wave and the corresponding QRS wave signal and whether a heart rhythm is regular; wherein the width of the QRS wave signal is wide and the P wave is not generated simultaneously, the QRS wave signal and the T wave are opposite directions, and the QRS wave signal is presented. The singular shape and a rhythm pattern present a reference pattern that is irregular to the positive abnormality determining mechanism. The personalized electrocardiogram processing apparatus according to claim 4, wherein the personalized positive abnormality determining mechanism of the determining unit further comprises the following and a combination thereof: two or two of the ten adjacent QRS wave signals appear More than one abnormal reference pattern; any two different shapes of abnormal reference patterns appear in the personalized electrocardiogram template; the heart rhythm is greater than 160 times per minute and no such P wave is narrower than the width of the QRS wave signal; The rule of the rhythm law is transformed into an irregularity; the irregularity of the rhythm law is converted into a rule; the personalized electrocardiogram template has three or more consecutive abnormal reference patterns; the personalized ECG template does not have the QRS wave signal and has Multi-wave baseline condition, no such P wave, the heart rhythm is greater than 100 times per minute and the irregularity of the rhythm pattern is generated simultaneously; the width of the waveform generated in the personalized ECG sample is greater than the first 6 to 8 heartbeat intervals 1.5 times the average; and the QRS wave signal does not appear after the P wave appears in the personalized ECG template. The personalized electrocardiogram processing apparatus according to claim 1, wherein the first parallelization processing mode of the identification unit is a single-input multiple-output parallel dynamic time warping algorithm at a single moment to simultaneously identify the personalization. The similarity between the ECG template and the preset ECG templates. The personalized electrocardiogram processing apparatus according to claim 1, wherein the second parallelization processing mode of the judging unit is a parallel personal normal abnormality judging mechanism of a single input multiple output at a single moment to simultaneously determine the identification result. . A personalized ECG processing method includes: detecting a complete personal electrocardiogram signal to generate a QRS wave signal; receiving and detecting at least two QRS wave signals to calculate a heartbeat interval; and obtaining a corresponding response according to the heartbeat interval A humanized electrocardiogram template of a complete personal electrocardiogram signal; a plurality of preset electrocardiogram templates are obtained from a database, and the personalized electrocardiogram template and the preset electrocardiogram templates are simultaneously identified by a first parallelization processing method to generate an identification The result is determined by simultaneously determining the identification result and the plurality of personalized positive abnormality determining mechanisms by a second parallelization processing manner to generate a determination result. The method for processing a personalized electrocardiogram according to claim 8 further includes: storing the determination result in a first storage unit when the determination result is normal; and when the judgment result is an abnormal situation, The judgment result is stored in a second storage unit and output. The method for processing a personalized electrocardiogram according to claim 9 further includes: when the judgment result is an abnormal situation, storing the judgment result in the database to expand the preset ECG template of the database; data. The personalized ECG processing method of claim 8, wherein the personalized positive abnormality determining mechanism comprises: whether there is a P wave in the personalized ECG template, a width of the QRS wave signal, and the P wave corresponding thereto The relationship between the QRS wave signal and whether a heart rhythm is regular; wherein the width of the QRS wave signal is wide and the P wave is not generated simultaneously, the QRS wave signal and the T wave are opposite directions, the QRS wave signal exhibits a strange shape, and The rhythm pattern presents a reference pattern that is irregular to the positive abnormality judgment mechanism. The personalized electrocardiogram processing method according to claim 11, wherein the personalized positive abnormality determining mechanism further comprises the following and a combination thereof: two or more abnormal reference occurrences in ten adjacent QRS wave signals a pattern; any two different shapes of the abnormal reference pattern appear in the personalized electrocardiogram template; the heart rhythm is greater than 160 times per minute and no such P wave is narrower than the width of the QRS wave signal; the rhythm rule Turning into irregularity; the irregularity of the rhythm pattern is transformed into a rule; the personalized electrocardiogram template has three or more consecutive abnormal reference patterns; the personalized ECG template does not have the QRS wave signal and has a multi-wave baseline Condition, no such P wave, the heart rhythm is greater than 100 per minute and the irregularity of the rhythm pattern is generated simultaneously; the width of the waveform generated in the personalized ECG sample is greater than the average of the first 6 to 8 heartbeat intervals. Times; and the QRS wave signal does not appear after the P wave appears in the personalized ECG template. The personalized electrocardiogram processing method of claim 8, wherein the step of simultaneously identifying the personalized electrocardiogram template and the preset electrocardiogram templates in the first parallelization processing manner comprises: using a single input at a single moment A parallel dynamic time warping algorithm for multiple outputs to simultaneously identify the similarity between the personalized ECG template and the preset ECG templates. The method for processing a personalized electrocardiogram according to claim 8 , wherein the step of simultaneously determining the identification result and the personalized positive abnormality determining mechanism by using the second parallelization processing method comprises: using a single input of a single moment Parallel normal abnormality judgment mechanism of multiple outputs to simultaneously judge the identification result and the personalized positive abnormality judgment mechanism.