Detection of Arrhythmia
-
Upload
matthew-dunning -
Category
Healthcare
-
view
67 -
download
1
Transcript of Detection of Arrhythmia
ArrhythmiaPresenters:
Matthew Dunning
Rahib Zaman
Manmohan Singh
Brendan Wiggins
Presentation Topics
Introduction Methods Results Discussion Conclusion
Introduction
Cardiac dysrhythmia, known as arrhythmia, is a medical condition where the rhythm of the heart is irregular, faster or slower than average.
The average healthy human adult has a heart rate of 60-70 heart beats per minute
Three forms of arrhythmia Tachycardia – When the heart rate exceeds 90 heart beats per
minute Bradycardia – When the heart rate is less than 60 heart beats per
minute Irregular – Inconsistent heart rhythm
Significance of Problem
Each year in the United States, around 500,000 deaths occur from arrhythmia.
Arrhythmia In the atria results in inefficient flow of blood to the rest of the body. Can result in shortness of breath, blood clots and even a
stroke.
However, there can be a 15-20% decrease in the number of deaths if there is a correct and early diagnosis.
Objective
The objective of this research is to analyze Electrocardiogram (ECG) signals to determine any onsets of arrhythmia.
The primary questions is whether the algorithm can accurately detect sinus tachycardia and bradycardia, along with any irregular heart rhythms.
Study Population
The data was acquired from physionet’s online database;47 different ECG signals were obtained.
The study population had an age range from 23-89 years old, the average age of the patient was 63.
In the study, there were 21 males and 26 females. It is important that the people do not eat or drink anything
prior to the test as it could sway the results (ex caffeine). The data contains four different types of ECG’s: regular,
tachycardia, bradycardia and those who have an irregular heartbeat.
Study Population
Each ECG downloaded contained an array of voltages (in mV).
A typical ECG lasts about (30-40seconds); the length of the acquired test was chosen to be one minute.
The signals were sampled at 360Hz.
Methods
A modified Pan-Tompkins algorithm was used to analyze the ECG signals.
The original algorithm works by passing the signal through a low pass filter (to remove noise), a high pass filter (accentuate QRS peaks) and a derivative filter. It is then squared, passed through a moving average filter and then through a thresholding technique to detect R-peaks
Methods (Modified Algorithm)
Bandpass – reduce noise and baseline drift Derivative filter – identifies QRS complex Squaring operation – increases frequencies Moving Average – signal is smoothed to highlight the
QRS complex Thresholding – Detects two types of peaks; the QRS
complex and T waves. Uses a search back technique to detect each R peak
Detecting Irregularities
The algorithm will determine if a heartbeat is irregular1. Calculates time period differences between each peak2. Finds the difference of the two differences between peaks,
and compares to a tolerance level estimated to allow small number of premature contractions
3. If the QRS difference is greater than the tolerance level, then the program detects that segment as an irregularity.
4. The algorithm deems a ECG signal as irregular if it counts more than 8 irregularities
Results
51%
13%
13%
21%
2%ECG Data ResultsNormal
TachycardiaBradycardiaIrregularVtach/VFib
Normal ECGVo
ltage
(mV)
Index (N)
Normal ECG
Volta
ge(m
V)
Index (N)
Normal ECG 2Vo
ltage
(mV)
Index (N)
Normal ECG 2
Volta
ge(m
V)
Index (N)
Bradycardia Vo
ltage
(mV)
Index (N)
Bradycardia
Volta
ge(m
V)
Index (N)
Bradycardia 2Vo
ltage
(mV)
Index (N)
Bradycardia 2
Volta
ge(m
V)
Index (N)
TachycardiaVo
ltage
(mV)
Index (N)
Tachycardia
Volta
ge(m
V)
Index (N)
Tachycardia 2Vo
ltage
(mV)
Index (N)
Tachycardia 2
Index (N)
Volta
ge(m
V)
IrregularVo
ltage
(mV)
Index (N)
Irregular
Index (N)
Volta
ge(m
V)
Irregular 2Vo
ltage
(mV)
Index (N)
Irregular 2
Index (N)
Volta
ge(m
V)
Ventricular Tachycardia/FibrillationVo
ltage
(mV)
Index (N)
Irregular
Index (N)
Volta
ge(m
V)
Ventricular Tachycardia/Fibrillation
Discussion
Out of the 47 patients, 43 had a correct heart rate calculated by the algorithm (a 91.48% success rate).
Better than original algorithm (a 72.3% success rate) Problem with original algorithm is that it filtered the
signal so much that some of the peaks were reduced below the threshold which caused inaccurate calculation of heartbeats/minute.
It was important to make modifications so that the sampling and cut-off frequencies kept the QRS peaks intact.
Future
The algorithm can be modified for future use to include detection of life threatening heart rhythms (ventricular fibrillation)
As a result there is no P wave, T wave and the QRS is elongated and occurs rapidly without a refractory period.
The algorithm can be modified to detect such occurrences by detecting absence of p waves. By detecting absence of p waves and measuring if the BPM is extremely high over small periods of time.
Conclusion
The algorithm was successful in the primary objective of determining arrhythmic heart rhythm from the given ECG data.
The algorithm correctly identified sinus tachycardia and sinus bradycardia, while had a 91.48% overall success rate of identifying normal and arrhythmic heart rhythms.
References[1] ‘Arrhythmia’, American Heart Association, 23-Oct-2014. [Online]. Available:
http://www.heart.org/HEARTORG/Conditions/Arrhythmia/Arrhythmia_UCM_002013_SubHomePage.jsp. [Accessed: 20-Nov-2014].[2] ‘Arrhythmia: A Patient Guide’, Health Central, 05-Sep-2001. [Online]. Available: http://www.healthcentral.com/heart-disease/patient-guide-
44628-6_1.html. [Accessed: 20-Nov-2014].[3] ‘Types of Arrhythmias’, Cleveland Clinic, Nov-2012. [Online]. Available: https://my.clevelandclinic.org/services/heart/disorders/arrhythmia/
types. [Accessed: 20-Nov-2014].[4] M. J. Janse and M. R. Rosen, ‘History of Arrhythmias’, Basis and Treatment of Cardiac Arrhythmias, 2006.[5] ‘What Is An Electrocardiogram (ECG)?’, The Internet Journal of Advanced Nursing Practice, vol. 4, 2000.[6] A. Davies and A. Scott, ‘Arrhythmias’, Starting to Read ECGs, 2015.[7] ‘What Is an Electrocardiogram?’, National Heart, Lung, and Blood Institution, 01-Oct-2010. [Online]. Available:
http://www.nhlbi.nih.gov/health/health-topics/topics/ekg. [Accessed: 20-Nov-2014].[8] ECG Database http://www.physionet.org/physiobank/database/mitdb/. [Accessed: 20-Nov-2014]
[9] H. Sedghamiz, 'Complete Pan Tompkins Implementation ECG QRS detector - File Exchange - MATLAB Central', Mathworks.com, 2014. [Online]. Available: http://www.mathworks.com/matlabcentral/fileexchange/45840-complete-pan-tompkins-implementation-ecg-qrs-detector. [Accessed: 08- Dec- 2014].
[10] C. Pavlatos, A. Dimopoulos, G. Manis and G. Papakonstantinou, Hardware Implementation of Pan & Tompkins QRS Detection Algorithm, 1st ed. Zografou, Athens: National Technical University of Athens, 2014, pp. 1-2 [Online]. Available: http://mule.cslab.ece.ntua.gr/docs/c8.pdf. [Accessed: 08- Dec- 2014]
[11] V. Afonso, ECG QRS Detection, 1st ed. 2014 [Online]. Available: http://www.masys.url.tw/AU/2014SP/BMSD-D/Text/BMSD-text-ECG_QRS_Detection.pdf. [Accessed: 08- Dec- 2014] [12] Pan.J, Tompkins. W.J,"A Real-Time QRS Detection Algorithm" Transactions On Biomedical Engineering, Vol. BME-32, No. 3, March 1985.