Automatic Detection of Heart Disease Using Spatial Velocity
Computer aided automatic ECG signal processing has drawn significant attention of the researchers over the
past decade or so. Identifying the features of ECG signals such as peaks, the intervals and the segment has been one of the
major area of research in one dimensional ECG signal processing. An accurate and proper estimation and detection of this
features can help in better diagnosis of heart related diseases such as arrhythmia and so on. Even though there have been
several previous studies that have proposed different techniques for efficiently detecting ECG properties, very few works
have suggested simultaneous feature extraction from all twelve lead of data and integrated this features into a decision
making process . In this work we have proposed a novel ECG signal processing and feature estimation technique by
combining time domain windowing technique with spatial velocity and FFT. First we filter the signals with rectangular
window and separate out the improper frequency band .We follow it by a spatial velocity plus temporal domain filtering
which provides the high peaks of the R-Peak locations based on this reference points we calculate the rest of the other peaks,
their amplitude and onset and offset using time domain windowing technique. We further propose a novel T-wave alternan
detection system by using FFT and temporal analysis system. We evaluate our technique using piezonets, normal sinus
rhythm as well as T wave alternate public data base. Results shows that our system can detect the peaks with an accuracy of
91.5333% and precision of 0.0546 recall of 1.2582.
Keywords— T wave alternan, spatial velocity.