An Algorithm of ST Segment Classification and Detection Zhao Shen 1 , Chao Hu 2 , Jingsheng Liao 2 Max Q.H Meng 3 ,Fellow IEEE1.Institute of Automation, Northwestern Polytechnical University , Xi’an Shaanxi Province, China 3.Department of Electronic and Engineering 2.Shenzhen Institutes of Advanced Technology Shenzhen, Guangdong Province, China * Chinese University of Hong Kong [email protected]Hong Kong , China{chao.hu & js.liao}@siat.ac.cn [email protected]This work is supported by the grants from National Sc. & Tech. Pillar Program (2008BAI65B21), the Guangdong/CAS Cooperation Project (2009B091300160), Shenzhen Sc. & Tech. Research Funds, and the Knowledge Innovation Eng. Funds of CAS. Abstract- In electrocardiogram real-time monitoring, the ST segment detection is important, it has close relationship with myocardial ischemia and myocardial infarction. In this paper, ST detection is divided into two parts, firstly using wavelet and morphology method to calculate the offset direction, waveform and summarizing features of ST, eventually divide the ST in 15 types; next analyzing the ST waveform change tendency in about 30 minutes and discovering the rhythm of ischemia or infarction to help doctor. At last, the first part has been confirmed by MIT- BIH Arrhythmia database and European ST-T database and t he second is proved by two experts’ conclusio n. Index Terms - ST segment, wavelet, R wave, T wave I.I NTRODUCTIONIn ECG monitoring, ST segment means the change of electric potential in the period which from the end of ventricular depolarization to the origin of repolarization. In normal conditions, ST segment shows horizon level, but in some heart disease conditions, the ST segment will be affected and drifted in different direction and shown in various forms, as is shown in figure1.Because ST segment will show in various form for different kinds heart disease attack, so that the change of ST segment is a significant indicator of various heart disease in ECG clinical care. During analysis the ST segment, the main contents include the following two steps: one is determining the type of ST; the other is analysis the change trend of ST. The first step of ST type classification is calculate ST offset level, it include J+X, R+X, regional search, and T wave methods. The J+X method is firstly determine the end point of QRS, named J point, and then select the point which several milliseconds after J point as J+X point, using the value of this point as the offset level, the R+X method is the same with the above method, just use R point replace J point because R point is easily determined than J point. Regional research method means select several points as a template and uses this template to matching other part of ST segment and selects t he max value or min value as the offset level. The T wave method is using the original point of T wave as the offset level. Fig1.Different Kinds of ST Waveform The shape recognition of ST segment is another important. Because ST segment will show various shape for different heart disease, this feature not can help to analysis the reason of ST change, but will help people to determine the different heart disease. In this field, many people have achieved great success. Mao [1] use other methods to eliminate the noise and analysis the pattern of ST, has achieved good result, but the progress is too complex to use in real time monitor, Shi [2] use wavelet transform method to get the key points of ST, and use straight line to fitting ST segment curve, and finally divide ST segment into five pattern, because the complex of ST itself and noise disturbance, the accuracy of this result is relatively low. Liu [3] and others use neural network and other algorithm to analysis the ST segment and achieve good result but it must with lots of data and it can’t recognize the new pattern which have not been trained in the past. Other people al so have done much work in this fiel d but most of these methods are too complex to using in real time monitoring. In this paper, a new method about ST segment analysis is proposed. Firstly, using wavelet transform to eliminate the disturbance and determine the key points of ECG, calculating the ST offset, the curve type, and the concave-convex factor or straight slope. After determine the ST pattern, for help doctor monitor and diagnose, the ST change trend will be analysis by using the former method, nearly 30 minutes ECG will be 559
6
Embed
An Algorithm of ST Segment Classification and Detection
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
7/22/2019 An Algorithm of ST Segment Classification and Detection
In part two, the ST waveform has been recognized in 15types, next we will use this conclusion to continuesmonitoring and diagnose myocardial ischemia and infarction.
A. Myocardial Ischemia and Infarction
Myocardial ischemia is one of the most common heart
diseases. The main reason is coronary atherosclerosis
decreased blood flow to the heart, that make oxygen supply
decrease and myocardial metabolism appears abnormal, in thiscondition the heart can not work normally. If this continues, it
means the patient get myocardial ischemia.
Myocardial infarction means myocardial vascular
necrosis, the immediate reason is coronary artery disease
which makes blood fell sharply or interrupted, thus heart will
appears serious and sustainability of ischemia and cause
infarction. The basic reason is plaques and thrombosis which
caused by coronary atherosclerotic limit the blood flow and
make coronary artery disease.
B. The Relationship of ST with Ischemia and Infarction
ST can rapidly and accurately reflects the myocardial
ischemia attack interval, duration and severity, the modern
standard is if ST fit for the following: ○1 at JX point and
nearby, ST shows in horizon or downward type;○2 drop over
1mm or 0.5mV and continue over 1 minute; ○3 the attack
interval is over 1 minute, this we can conclude the myocardial
ischemia has happened. And myocardial infarction can also be
reflected by ST change, when myocardial infarction has
happened, ST will shows in convex type, and be elevated until
exceed T wave, sometimes because of T wave inversion, ST
will become wilder than ever. For the relationship between ST
and myocardial ischemia or infarction is analysed, the
diagnose result is concluded by experts and shown in table2.
C. Analysis the Process of ST Change
In real time monitoring, ST change means the patient has
been attacked by myocardial ischemia or infarction. In table2,
experts have given a reference diagnose result, based on this
we will analysis the rhythm of ST change process, it will help
the doctor to detect the heart disease pathogenesis regularity,
the work mainly include three part: at the beginning the attack
time of ischemia or infarction is detected, the next step is
duration time, finally is the severity. According these test
result doctor will conclude the ischemia or infarction belong to
acute or continually disease, and common or severe. During
the time from 0:00~30:00, firstly we detect the ST waveform
type at 0:00 o’clock, and in every 5 minute we detect ST type
and compare with the type 5 minutes ago, at the same time if
there happens the break out of disease and T wave change,record and give the diagnose result. The ST change process
test result is shown in table 3.
“N means this ST segment with great noise”
100
105
E0104
E0105
E0515
E0601
E0614
E0704
0:00 5:00 10:00 15:00 20:00 25:00 30:00
Lower
Curve
Convex
ST↑ ST↑ ST↓
ST↑
T wave
Notch
ST↑ No
Change
Lower
Curve
Concave
ST
Noise
No
Change
No
Change
ST
NoiseST↑ ST↓
Higher
Curve
Concave
No
Change
ST
Noise
No
Change
No
Change
No
ChangeST↓
Higher
Straight
Upward
No
Change
No
ChangeST↓
ST↑
Burst MI
ST ↓
RevertST↓
Higher
Curve
Concave
ST↑
T↑
MI
ST↑ ST↑ ST↑ ST↑
ST↑
T↑
MI
Lower
Straight
Down
No
Change
ST
Noise
ST
Noise
No
Change
No
Change
No
Change
Normal
Straight
Horizon
No
Change
No
Change
No
Change
No
Change
No
Change No
Change
Lower
Straight
Horizon
No
Change
No
Change
No
Change
ST
Great
Noise
ST
Noise
No
Change
Table 3. ST Change Process
“MI means myocardial infarction”
In table 3 we select two examples to explain the change
process of ST segment, both of two diagnose are consistent
with the experts conclusion, first example in E0105, at the
original time, the patient has signs of myocardial infarction for
ST begin to raise, at about 5:00 o’clock, myocardial infarction
happen and ST is higher than T wave, after the time ST begin
to drop and the symptom begin to gradually release, and
before 30:00 o’clock, ST raise again and the myocardial
infarction happen, this means that the myocardial infarction is
period occur, the period is about every 25~30 minutes, and
563
7/22/2019 An Algorithm of ST Segment Classification and Detection