14-06-2013, 12:52 PM
Electro-CardioGram (ECG) Data Compression
Electro-CardioGram.pptx (Size: 236.28 KB / Downloads: 31)
Objectives of the proposed work
To have a lossless or a near lossless ECG Data Compression
To have an appreciable Compression Ratio (CR) that can reduce memory required to store ECG data
To minimize Percentage-Root-mean-Square Difference (PRD) to an extremely low value
To propose a suitable decoding and
dequantization to reconstruct the
original ECG signal
Why ECG data Compression ?
Electro-Cardiogram data is pretty much required
from the clinical point of use.
Many a times doctors would fix the ECG monitoring
device in patients body for a reasonably long duration
( may be a week) to check at which time heart would
function bit differently
Hence data collected for so many days would be really
long hence needed to be compressed , but keeping the
fact that critical information should not be lost..
Inferences fromLiterature Survey
Scalar quantization leads to lossy compression. Hence PRD is very high
Wavelet transforms provide low PRD value but Compression ratio is very less ( <10:1). Hence requires more storage space
Compression based on neural network provides a bad reconstruction of original signal
Sparse decomposition and compression technique provide a normal CR around 6.5:1, but PRD value is quite high
( >7%)