Category Archives: Biomedical DSP

Neural network signal filters are more adaptive filters than Wiener filters where resulting signal is a product of minimisation mean-square error, Wiener filter is linear filter which is adapted for some specific environment. But linear adaptive filters are limited when noise has Gaussian process pattern. So if significant Gaussian noise is present then there is need to use non-linear filters, which can deal with such noise. Also non-linear filters do not need mathematical analysis in order to calculate reference frame signal, what is not always ease to do. Non-linear adaptive filters may deal with this in two ways: improving learning efficiency and a broadening of application areas.

A neural network is a massively parallel distributed processor that has a natural propensity for storing experiential knowledge and making it available for use. It resembles the brain in two respects:

• Knowledge is acquired by the network through a learning process.

• Interconnection strengths known as synaptic weights are used to store the knowledge.

Continue reading →

Electro-cardiographic examination may be performed not only from patients who is laying or sitting down peacefully, but also under stress conditions as some deceases may be detected when patient isn’t in the rest. For this ECG should be measured when patient is actively moving. We know that ECH signal is very weak among various artefacts. Motion artefacts are ones who is hard to filter out using regular methods. Various researches show that motion artefacts may be extracted by using accelerometers. Accelerometer reads motion pattern which is simultaneously analysed and filtered out from ECG using adaptive filtering algorithms including Least Mean Squares (LMS), Recursive Least Squares (RLS)

As we mentioned – motion artefacts disturb the ECG signal so that it is almost impossible to recognize ECG pattern. And it is really hard to eliminate it and extract valuable information because artefact spectrum overlaps with ECG spectrum. And worse – noise spectrum changes all the time as movement may be not the same. So it is logical to measure the motion pattern with separate sensor – single or multiple axis accelerometer and use adaptive filtering technique to remove noise caused by motion.

Continue reading →

Let’s make a filter, which filters off the 60Hz frequency from ECG signal. As we know American power supply is 60Hz. This is common noise in biomedical signals, while they are powered from industrial power supply. This type of noise can be defined easily and can be filtered as parameters of noise are known.
Here is one example of how to implement FIR filter using mathematical tools, like Matlab. This can be done by using microcontroller, like ARM or even ARM, because the frequencies are up to 1 KHz.
Initial conditions:

f0=60Hz – pover supply frequency;
fs=500Hz – sampling rate;
frequencies who define complex zeros:

we get w0=0.754;
Positions of complex zeros:

Zeros and poles in z plane

System Function

From it we can calculate filter coefficients:

And filter coefficients:

       

Also we know that:

And here we get filter characteristics:

Ve have band stop filter at 60Hz and its jam at 60Hz is -300dB. Bellow is filter structure:

Now using this filter we can filter ECG signal:

As you can see this is simple FIR filter. In other words there is nothing more than average function which doesn’t need much of resources. The other benefit of FIR filter is that we can get linear Phase characteristics, there fore filtered signal isn’t distorted because of different spectra frequency delay. As result wee see that ECG become more smooth. FIR filter