ECG compression algorithm based on lifting discrete wavelet transform

مقدم أطروحة جامعية

Hassun, Sumayyah Dari Awad

مشرف أطروحة جامعية

Abd Allah, Hadil Nasrat

الجامعة

الجامعة التكنولوجية

الكلية

-

القسم الأكاديمي

قسم الهندسة الكهربائية

دولة الجامعة

العراق

الدرجة العلمية

ماجستير

تاريخ الدرجة العلمية

2010

الملخص الإنجليزي

Remote monitoring of heart patients becomes increasingly important in the present days.

ECG (Electrocardiogram) is a very well-known method for diagnosing heart problems.

ECG data are very huge to be transferred through communication lines, unless being accurately compressed, without losing any part of data.

For example, a 3 channel, 24 hour ambulatory ECG typically has storage requirement of over 50 MB.

Therefore we need to reduce the data volume to decrease storage cost or make ECG signal suitable and ready for transmission through common Communication channels such as phone line or mobile channel.

So, an effective data compression method is need.

A compression method, in which both the Wavelet Transform (WT) and the Run Length Coding (RLC) methods are combined, has been implemented to compress ECG data in one and two dimensional forms.

The Fast Lifting Wavelet Transform (FLWT) is used to divide the 1-D and 2-D ECG data into sub bands to compact the energy of signals in much less samples than in time domain.

The sub band division has been repeated three times for higher compressibility.

Energy Packing Efficiency (EPE) has been used to threshold the resultant wavelet coefficients prior to compression using the Run Length Coding method (RLC) to get the compressed form of data.

Three different wavelets have been used in Wavelet Transform, namely, Haar, Cohen–Daubechies- Feauveau (CDF 2.2) and Daubechies 4.

The PRD are 0.814 %, 0.8167 % and 0.808% for compression ratio of 5, 4.8 and 5.2 for Haar, CDF2.2 and Db4 wavelets when compressing 1-D ECG using the 1D ECG signal algorithm.

The 2D ECG signal algorithm gives PRD equals to 0.907 %, 0.93 % and 0.87 % for Haar, CDF2.2 and Db4 wavelets respectively for compression ratio of 7, 6.875 and 7.1.

Db4 has better performance than the other two wavelets.

Db4 gives the highest compression ratio CR = 5.2 and 7.1 for 1D and 2D respectively with the lowest error PRD= 0.8 % and 0.87 % for 1D and 2D respectively.

Algorithms for FLWT, thresholding of wavelet coefficients and RLC for each filter in both one and two dimensional forms have been designed and implemented using M- file MATLAB R2009a.

Algorithms show fast and easy coding and decoding stages.

The ECG data used in the present simulation is available at the MIT–BIH arrhythmia database in Different signal shapes.

A comparison between the 1-D and 2–D forms of data has been carried out.

It was clear that 2-D arrangement of ECG data approximately doubled the compression ratio with slight increase in percent root mean square difference (PRD) which represents the distortion measure in the transferred data.

For example when comparing compression ratio after applying 1D and 2D algorithm to signal of record 14046 it is equal CR = 5.3 and CR = 7.599 respectively which means that it increased by 2.3 with little increment in the error value equal to 0.09 %.

التخصصات الرئيسية

الهندسة الكهربائية

الموضوعات

• القلب
• الأمراض
• الخواص الكهربائية
• التشخيص
• تخطيط القلب

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-305125