Biomedical Signals and Signal Processing PDF

1 Chapter one Introduction to Biomedical signals and biomedical signal processing 1.1 Introduction to biomedical signals 1.1.1 Definitions and models of biomedical signals Before everything else let's define what signal is. A signal is a simple valued represe...

1 Chapter one Introduction to Biomedical signals and biomedical signal processing 1.1 Introduction to biomedical signals 1.1.1 Definitions and models of biomedical signals Before everything else let's define what signal is. A signal is a simple valued representation of information as a function of an independent variable. Even though, signal is defined as something that can be acquired from sensors (output of sensors), it is not entirely true. For example the sequence of protein in the DNA strand is signal. Thus, a Signal is not an output of a sensor, rather it is a mathematical representation of information as a function of an independent variable (Kaniusas, 2012). As a result, biomedical signals are signals which are acquired from instrumentation devices. The registered biomedical signals are called biosignals. In the scope of biomedical signals and sensors, biosignals are descriptors of physiological phenomenons (Kaniusas, 2012). A simple model for signal generation upto registration for acoustic signal of the chest cavity is explained below. Such signals are used for assessment of the cardiorespiratory pathologies. The heart sounds are given by closure of heart valves whereas the lung sounds originate from air turbulence created inside the branched airways. Before reaching the registration, these signals experience propagation losses. This propagation loss is due to the tissue the signal has to pass (travel) while reaching for the sensor. The amplitude of the signal as well as the frequency or frequencies of it would be affected as a result. Since many signals do have different frequency spectrums, the effect may differ. For example, the propagation attenuation for lung sound is very high as compared to the heart sound (Kaniusas, 2012). After the propagation losses (which is modeled by impedance) there are coupling and conversion losses. This loss is as a result of amplification and conversion of the signal from nonelectrical signal to electrical signal. The figure shown below gives electrical modeling of  2 the biomedical signals for induced and permanent signal types. Figure 1.1: Figure explaining the modeling of the signal propagation in the body to the sensor and registration for permanent signal (a) and induced signal types (b)from “Biomedical signals and sensors I, By Eugenjis Kaniusas”. Historical aspects: the registration of biosignals is driven by patients and physicians' needs (Kaniusas, 2012). Apart from verbal account, the very first diagnosis is done through the following: Inspection (visual inspection): applied by Hippocrates. Palpation (touch and apply pressure): applied by Hippocrates. Percussion (striking the body by hand): invented by Leopold Auerugger by 1761. Auscultation (direct/indirect listening): through the invention of the stethoscope by Dr. Rene Theophite Hyacinthe Leanne in 1816, became more popular. The main problems with these direct diagnosis systems is that; the proof of biosignals analysis of biosignals comparison of biosignals circulation of biosignals was impossible due to the subjective nature of the diagnosis.  3 Figure 1.2: Figure showing inspection (a), palpation (b) and auscultation (c) from “Biomedical signals and sensors I, By Eugenjis Kaniusas”. Figure 1.3: Figure showing direct auscultation (left) and indirect auscultation (right) from “Biomedical signals and sensors I, By Eugenjis Kaniusas”. To objectify these issues, verbal description, using musical notes and technical tools, were implemented. N.B. high (height of note) gives quality and rhythm gives quantity for blood pressure signal. 1.1.2 Classifications of biosignals Biosignals can be classified based on existence, dynamic and origin.  4 Existence : based on existence, biosignals are clanifred as permanent and induced. Permanent biosignals exist without any artificial impact, trigger, excitation from outside the body and they are available at any time. Example: ECG. Induced brosignals require triggers from outside. They only exist for the duration of excitation. Example: Oximeter (pulse). Dynamic: based on dynamic biosignals are classified as static (quasi) and dynamic. Static biosignals are signals that may not alter or be affected by time very slowly. Example core temperature. Dynamic signals are extensively changed with time Example heart beat. Origin: based on origin signals are classified as: - electric biosignal - chemical biosignals - magnetic - thermal - mechanic - acoustic - optic - other.  5 Figure 1.4: Figure showing the difference among signal types based on existence (a), dynamic (b) and origin (c), from “Biomedical signals and sensors I, By Eugenjis Kaniusas” 1.2 Nature and challenges of biomedical signals Medical data are basically classified as alphanumeric (patient name, address, Id and so on, results of lab tests and so on), medical images from different imaging modalities archived into basic medical data communication like PACs and DICOM, and finally physiological signals. The medical data are acquired, archived and manipulated in hospitals (Tompkins, 1993).  6 Figure 1.5: Figure showing the different medical data types from “Biomedical digital signals processing, By Willis J. Tompkins” 1.2.1 Signal acquisition and physiological measurement The basic physiological measurement trends are not so much different from basic instrumentation schematic. The basic schematic diagram of physiologized measurement is shown below. Figure 1.6: Figure showing the measurement system (Devasahayam, 2012). The physiological process from the body will be converted to an equivalent electrical representation through a transducer. The transducer converts these non-electrical physical quantities into electrical signals. Most of the converted signals are usually in analog form. Thus, they need signal conditioning. The Signal conditioning includes, attenuating undesirable frequencies from the signal (filtering). And the signal has to be amplified for better signal recording and identification. Most of the time the instrumentation amplifier is  7 implemented for this task. This conditioned signal can then be recorded in its analog form or converted to digital form for better signal analysis and processing through analog to digital converter (ADC). The signal in digital form could be used for computer analysis and excellent feature extraction (Tompkins, 1993). In medical instrumentation which includes monitoring, measuring and analysis of the signal needs signal processing (usually digital signal processing) (Tompkins, 1993). Figure 1.7: Figure showing the basic medical instrumentation system (Tompkins, 1993). 1.2.2. Sources of biomedical signals Most physiological processes manifest themselves as signals. These reflect their nature and activities. Any disorder / disease in these physiological processes causes abnormalities. This is called a pathological process which affects the health and general well-being of the system (Rangayyan, 2002).  8 Figure 1.8: Figure Schematic representation of a generic physiological system with various types of possible inputs and outputs. The effect of a pathological process is depicted by the zigzag line across the system and the list of possible outputs from “Biomedical signal analysis By Rangayan M. Rangaraj.” Most classes of biomedical signals comprise those which are electrical in nature. These electrical signals inducing mechanical contraction of a single cell are called action potential. This is stimulated by electric current created by the movement of flow of Na+, K+, Cl- and other ions across the cell membrane. An action potential is the basic component of all bioelectrical signals. On the other hand resting potential (without any stimulus), occurs when there is no trigger. The cell membrane is semi-permeable, it allows some ions or molecules to flow in while blocking the others. In resting state the cell membrane allows K+ and Cl- ions to flow in while blocking Na+ ions (Rangayyan, 2002). - There is less concentration of Na+ inside the cell than outside - Outside of the cell there are more positive ions. - To balance the charge K+enters the cell, causing higher concentration inside the cell. - Charge balance can't be attained due to permeability of the membrane - A state of equilibrium is established with a certain potential difference Thus the potential difference at the resting potential is from -60mV to -100mV.  9 Figure 1.9: Figure Schematic representation of semi permeable membrane (left), resting potential (middle) and action potential (right) “Biomedical signal analysis By Rangayan M. Rangaraj.” Depolarization: when the cell is excited by ionic current, the permeability of the cell membrane changes and begins to allow Na+ ion to flow into the cell. While doing so, the K+ ions inside the cell began to exit the cell due to concentration gradient. But, the flow of K+ is not as fast as Na+. This makes an ionic current producing a positive charge inside the cell (higher potential). This situation of the cell is called action potential. This action potential is an equilibrium state caused by the stimulus. For most cells the action potential is around 20mV. This process is called depolarization. And the cells are depolarized cells. Repolarization: after a certain period of being in action potentials the cell began to rest to its resting potential. At this time the membrane becomes a barrier for Na+ and lets K+ ions enter into the cell. This creates much of Na+ ions to exit the cell rapidly through the ion channel and K+ ion enters through the cell by concentration gradient created by action potential. Na+-k+ pump is essential for resetting the balance of resting potential. Nerve and muscle cells depolarize rapidly with action potential of 1ms(millisecond). And for the heart the action potential duration is 150-300ms. An action potential is always the same for a specific cell. After an action potential there is a period the cell would not respond to the stimulus known as absolute refractory period (approx. lms for nerve cells). This is followed by a relative refractory period (in several ms) when another action  10 potential by strong stimuli is triggered in a normal situation. The response to the stimuli disregards the intensity and method of excitation as long as it is beyond the threshold. This is all or none phenomena. Propagation of an action potential: an action potential propagates through the length of the muscle unmyelinated fiber without decrease in amplitude. Current carrier by intracellular and extracellular fluids will depolarize the cell along. Myalinated nerve fibers are covered by myelin sheath. This sheath is interrupted by nodes of ranvier, where the fiber is exposed to interstitial fluids. Sites of excitations and changes in membrane permeability happen only at the nodes. And current flows from one node to another node by a process called saltatory conduction (Rangayyan, 2002). 1.2.3. Common biomedical signals Common biomedical signals are ECG, EEG, ENG, PCG EGG, ERG, COG...etc. In this course the base detailed view will be given to ECG, EEG and EMG. ECG (electrocardio-gram): this is a signal generated by an action potential originated in the heart ( specially at SA fiber). This is sensed through electrodes. This creates a PQRST - curve. This curve (especially the - R curve) gives better information for diagnosis of heart disease and arrhythmia. The figure below shows the action potential of the heart muscles as depicted. The whole action potential for a normal patient lasts for about 150 ms. The first action potential is the depolarization of the atrium and its amplitude is small as compared to the depolarization of the ventricle (left). And lastly, the ventricular repolarization is shown by T-wave. The repolarization of the atrium overlaps on the depolarization of the ventricle (Bruce & Bruce, 2001).  11 Figure 1.10: Figure Schematic representation of ECG signal “Biomedical signal analysis By Rangayan M. Rangaraj.” EEG (electroencephalogram): this is a signal from the brain cells. There are a number of electrodes involved for measuring the alpha, Beta, gamma and theta waves. They are very much used in accessing the sleeping action of the brain activities (Rangayyan, 2002).  12 Figure 1.11: Figure Schematic representation of 10-20 EEG electrode system placement , pg: nasopharyngeal, a: auricular, fp: refrontal, f: frontal, p: parital, c: central, o: occipital, t: temporal, cb: cerebellar, z: midline, odd number in the left and even numbers in the right, “Biomedical signal analysis By Rangayan M. Rangaraj.” Figure 1.12: From top to bottom: (a) delta rhythm; (b) theta rhythm; (c) alpha rhythm; (d) beta rhythm; (e) blocking of the alpha rhythm by eye opening; (f) 1 s time markers and 50 µV marker. Reproduced with permission from R. Cooper, J.W. Osselton, and J.C. Shaw, EEG Technology, 3rd Edition, 1980. C Butterworth Heinemann Publishers, a division of Reed Educational & Professional Publishing Ltd., Oxford, UK. EMG (electromyogram): this electrical based biosignal is formed due to the contraction of muscle cells. It is a triggered potential which will be transduced through on surface  13 (non-invasive) or invasive AgCl electrodes. It is usually the sum of the responses of a number of muscle fibers. The summation of the responses of muscle units is called multiple-unit EMG (MUEMG).. This is because most stimuli trigger a number of muscle cells and most importantly to move (for movement of) a single part of the skeleton needs the contraction of a number of muscle fibers. It is mostly used to detect the abnormality in the muscle cells (muscular dystrophy) and other likely disorders (Rangayyan, 2002). Figure 1.13: Schematic representation of a motor unit and model for the generation of EMG signals. Top panel: A motor unit includes an anterior horn cell or motor neuron (illustrated in a cross­section of the spinal cord), an axon, and several connected muscle fibers. Each system hi(t) shown represents a motor unit that is activated and generates a train of SMUAPs. The net EMG is the sum of several SMUAP trains, from “Biomedical signal analysis By Rangayan M. Rangaraj.” Electroneurogram (ENG): it is used to measure the conduction velocity of the nerve cell. Use a needle or on surface AgCl electrode. The method is through applying strong stimuli (100v) for 10-30 microseconds to some nerve cells. (this is an applied relaxed position to avoid any other muscular contraction). Then by measuring the action potential on a known and measured body length, the latency is measured. This latency is conduction time. Then through the measured length of the body the signal has to travel the velocity will be  14 acquired. Then by applying amplification of gains around 2000 and filter with bandwidth to 10Hz - 10KHz the signal can be ready for further processing or recording. This signal is affected by power lines greatly, as a result of its band width and small amplitude (10 microvolts) (Rangayyan, 2002). Nerve fibers 45-70 m/s Heart muscles 0.2-0.4ms Between atria and ventricle 0.05-0.04 m/s Most of the time any nerve disease may increase the latency. Figure 1.14: Nerve conduction velocity measurement via electrical stimulation of the ulnar nerve from “Biomedical signal analysis By Rangayan M. Rangaraj.” Elechogastrogram (EGG): the electrical activity of the stomach consists of rhythmic waves of depolarization and repolarization of its smooth muscle cells. The activity begins in the mid- corpus of the stomach with intervals of 20s in humans. An external electrode could sense these signals. With the subject in the auto supine position, a 5MHz ultrasound transducer array is used for localizing the stomach. Three active electrodes are placed at the abdomen with inter-electrode spacing of 3.5cm. A common electrode is placed 6cm away at the upper quadrant. Then the signal is then conditioned in a band width 0.02-0.3 Hz with 6 DB/octave and sampled at 2Hz. The surface of EGG is believed to reflect only the  15 overall electrical activity of the stomach. Accurate and reliable measures require implantable electrodes. Thus, diagnosis is not possible yet (Rangayyan, 2002). Phonocardiogram (PCG): the heart sound signal may be a traditional biomedical signal. The PCG is a vibration of sound related to the contractile activity of the cardihemic system (the heart and the blood together). The transducer is implemented to convert the vibration to electrical signal (accelerometers or microphones or pressure transducers are placed on the chest for this purpose). The normal heart sound provides an indication of the general state of the heart in terms of rhythms and contractile. Any cardiovascular disorder disease causes changes to the sound/additional sounds which are useful for diagnosis (Rangayyan, 2002). The carotid pulse: this is a pressure signal recorded over the carotid artery as it passes near the surface of the body at the neck; it is the extension of the pressure on the aorta that is felt on the neck due to its proximity (Rangayyan, 2002). Figure 1.15: Three ­channel simultaneous record of the PCG, ECG, and carotid pulse signals of a normal male adult from “Biomedical signal analysis By Rangayan M. Rangaraj.” Other biomedical signals: Signals from catheter tip sensors, Speech signal, Vibromyogram, Vibroarthrogram, Otoacoustic emission (OAE) signals and Bioacoustic signals. 1.2.4 Challenges and difficulties in biomedical signal acquisition and analysis There are many practical difficulties in biomedical signal acquisition despite advancements of instrumentation research (Rangayyan, 2002). Thus, from among many particular  16 attention must be given to the followings: Accessibility of variables to measurement: trade of between the information acquired as the risk and pain of the subject. Variability of signal source, as biomedical signals are affected by different factors. Inter relationships and interaction among physiological systems: Effect of instrumentation or procedure on the system: example heavy sensor Physiological artifacts and interference: example coughing. Energy limitations: most biomedical signals are at milli and micro levels at their source. Patient safety 1.3 Introduction to biomedical signal processing of the 1.3.1 Why are signals processed? Signal processing can be defined as the manipulation of a signal for the purpose of either extracting information from the signal, extracting information about the relationships of two signals or producing an alternative representation of the signal (Bruce & Bruce, 2001). According to the book by Bruce, Eugene N. motivation of signal processing categorized as follows: - To remove unwanted signal components that are corrupting the signal of interest. - To extract information by rendering it in a more obvious or more useful form. - To predict future values of the signal in order to anticipate the behavior of its source  17 Figure 1.16: Computer ­aided diagnosis and therapy based upon biomedical signal analysis (Rangayyan, 2002). 1.3.2 Objectives of biomedical signal analysis Some major objectives of biomedical signal processing are as follows: -Information extraction -Diagnosis -Monitoring -Therapy and control -Evaluation (Rangayyan, 2002) 1.3.3 Application of computer in medicine/biomedical signals Nowadays CAD (computer Aided decision/diagnosis) is used for the advancement of whole clinical decision making in diagnosis and therapy. This is due to the availability of better signal processing tools in digital systems (Rangayyan, 2002). From the advancement of computer technologies and evolution of ubiquitous computers, it has become more simple to make a computer itself as a medical instrument. This is by using complementary medical software. Usually these medical software are application software. The basic software architecture is shown below (Tompkins, 1993).  18 Figure 1.17: Software levels from “Biomedical digital signals processing, By Willis J. Tompkins” To transform any computer to biomedical instrumentation device, there exist two basic decisions to make these are; the choice of disk operating system and the high level language that should be implemented (Tompkins, 1993). Figure 1.18: Disk operating systems (left) and Languages (right).  19 References Bruce, E. N. (2001). Biomedical Signal Processing and Signal Modeling. Wiley. Devasahayam, S. R. (2012). Signals and Systems in Biomedical Engineering: Signal Processing and Physiological Systems Modeling. Springer US. Kaniusas, E. (2012). Biomedical Signals and Sensors I: Linking Physiological Phenomena and Biosignals. Springer. Rangayyan, R. M. (2002). Biomedical Signal Analysis: A Case-Study Approach. Wiley. Tompkins, W. J. (1993). Biomedical Digital Signal Processing (W. J. Tompkins, Ed.). Prentice Hall.