Processing of biosignals using DSP techniques 

Biosignals are nothing but signals that can be continuously monitored and measured. The signal is a parameter that contains information about the time and space of the object. Biosignals are of great importance cause they can be used for diagnosis. Any change in the obtained signal indicates an abnormality in the organ and thus aids in the diagnosis of diseases.

Why do we need to process biosignals?

Processing is mainly done to improve the quality of the signal and extraction of essential clinical parameters from the signal. This can lead to an accurate diagnosis. For example, if we consider the ECG (Electrocardiogram), the R-R interval is of great interest considering the calculation of heart rate whereas the T wave is of interest for the diagnosis of myocardial infarction. Hence, it is essential to extract the parameter required for the specified task/analysis.

Another factor is that the signals acquired in real-time are incorporated with noise. Noise is nothing but unwanted information. This can disrupt the nature of the signal and can heavily affect the diagnosis. The noise can occur due to improper placement of sensors, fault in the recording machine, or the movement of the patient (aberration). Hence it is essential to remove noise for better analysis.

Hence biosignal processing can be defined as the process of preprocessing acquired signals in order to extract meaningful information for the diagnosis of clinically significant events. The following are the steps that are done in biosignal processing.

Stages of biosignal processing

Signal acquisition– in this stage, the signal is acquired from the human body, This is mostly done through electrodes and sensors.

Amplification– the signal that is obtained is of low amplitude (microvolts) and hence has to be amplified before processing. This is done with help of analog amplifiers,

Filtering– in this step, the unwanted noise that is incorporated with the signal is eliminated. This is done with the help of filters like low pass, high pass, band pass, and band stop. We can choose a suitable filter by knowing the frequency of noise that affects the signal generally. We can use a high pass filter with a cutoff frequency higher than the noise frequency to eliminate them.

Analog to Digital conversion– in this stage, the real-time analog signal is converted to a digital signal so that the different digital processing techniques can be applied to them. We convert them to digital format because processing in the digital domain is quick and accurate. The conversion is done by a technique called sampling that converts continuous signals to discrete signals. Mostly a sampling frequency of 50Hz is considered for biosignals, especially ECG for heart rate analysis.

Digital Signal Processing for biosignals

The application of certain techniques to improve the accuracy and extract essential information is called as DSP. This method uses filters, windows, and many more tools to achieve the purpose. Most of these methods are programmable in software like MATLAB where the preprocessed signal can be imported for further processing using DSP techniques. It is important to note that Fourier transform is applied at every operation in order to plot and analyse the signal in frequency domain also.

Filters- Mostly notch and band pass filters are used for the elimination of noise and aberrations. The IIR (Infinite Impulse Response) is used for notch filter and the Butterworth is used for band pass. Chebyshev filter is an alternative to Butterworth, but this filter has ripples in its pass band, hence this filter wont be used widely.

Windows- There are four different types of windows namely, hamming, hanning, rectangular, and kaiser. A window function is a type of weighted function which has significance in the given interval and is zero outside the interval. Generally, the windows are convoluted with that of the signal. Out of the above specified windows, hamming is the best since it is accurate and it removes ripples. It has a narrow main lobe and the attenuation for the side lobes are very less.

Heart rate calculation using ECG

One of the applications for working on ECG is to calculate the heart rate. This is done using the R-R interval. Hence we need to extract the R-R interval from the ECG. This is done by the following algorithm.

1. the filtered and preprocessed signal is differentiated.
2. the differentiated signal is rectified by squaring it.
3. the rectified signal is smoothened using smoothening.
4. the peak is calculated from the smoothened signal using maximum function.
5. the threshold (for R) value is set as half of the peak.
6. the count value is initialized and is incremented if the amplitude exceeds the threshold.
7. the heart rate is calculated as the product of the time period and the count value after approximation.

Digital signal processing of EEG

One of the main tasks is to isolate the several frequency components of the EEG namely the alpha, beta, gamma and delta. This can be done by deploying band pass filters with suitable band pass attenuations and frequencies corresponding to that of the band. The table below represents the frequency range for the different bands of EEG.

Spectrum analysis techniques can be applied to analyze the signal in the frequency domain and calculate the spectral energy. This can be done using filters and FFT (Fast Fourier Transform). The autocorrelation, maximum cross-entropy, and maximum likelihood can be calculated from the signal for accurate diagnosis.

Digital Signal Processing for EMG

Spectrum analysis and derivation of spectrum-related components using fast Fourier transform is one major application. This can also be applied with machine learning and deep learning algorithms to classify and diagnose diseases associated with that the muscular system.

Hence these are some of the methods by which digital signal processing is applied for the processing of several biological signals.

Happy reading!!

Click to read about cancer-treating drug

Why do prices always end in 9?