BMET3802/9802 Biomedical Instrumentation PDF

BMET3802/9802 Biomedical Instrumentation Week 07: ECG Signal and acquisition Dr Sandhya Clement Lecturer...

BMET3802/9802 Biomedical Instrumentation Week 07: ECG Signal and acquisition Dr Sandhya Clement Lecturer School of Biomedical Engineering The University of Sydney  I would like to acknowledge the Traditional Owners of Australia and recognise their continuing connection to land, water and culture. I am currently on the land of the Cardigal people of the Eora Nation, and I pay my respects to their Elders, past, present and emerging. I further acknowledge the Traditional Owners of the country on which you are on and pay respects to their Elders, past, present and future. The University of Sydney The University of Sydney Page 3 Please note that most of the Lecture content are prepared and most of the Figures in this week’s lecture slides have been taken from the following textbooks: Webster, J. G. (Ed.). (2009). Medical instrumentation: application and design. John Wiley & Sons. Bronzino&Peterson: https://www-taylorfrancis- com.ezproxy.library.sydney.edu.au/books/mono/10.1201/9781351228671/me dical-devices-human-engineering-joseph-bronzino-donald-peterson The University of Sydney Page 4 This lecture: Electrocardiogram (ECG) – ECG – Introduction and basics – ECG Vector – Leads in ECG system – Arrythmias – ECG Acquisition System – Electrodes – Amplifiers – Filters – ECG classification – Week 7 Lab sneak peak – Temperature sensor – Arduino – ADC and DAC The University of Sydney Page 5 ElectroCardiogram (ECG) Basics The University of Sydney Page 6  Electrocardio gram (ECG/EKG): Basics – Heart is a four chambered pump for the circulatory system. – Pumping function is done by ventricles. – Atria are the antechambers to store blood during the time ventricle is pumping. – The filling/resting phase of heart is diastole. – The pumping phase is called systole. – A well coordinated series of electrical events that take place within the heart to facilitate rhythmic contract of atria and ventricles. – These electric activities are initiated by a coordinated series of events within the specialized conduction system of heart. The University of Sydney Page 7 Heart Conduction System – The specialized conduction system is small compared to the whole heart. – The wall of the left ventricle is 2.5 to 3 times as thick as the right ventricle – The intraventricular septum is as thick as the left ventricle wall. – Considering heart as a bioelectric source, the source strength at each instant is directly related the active muscle mass at that moment. – Active muscle mass---Myocardial cells – So, active free walls of ventricle, atria and https://commons.wikimedia.org/wiki/File:2018_Conduction_System_of_Heart.jpg interventricular septum are the major contributors of action potentials recorded from the heart. The University of Sydney Page 8 Electrical Behaviour of Cardiac Cell – Heart tissues: SA and AV node, purkinje and ventricular and atrial tissues. – Cells of each tissue differ anatomically. – All these cells are electrically excitable and exhibit different characteristic action potential. – In ECG – P Wave: Atrial depolarisation – QRS complex: Ventricular depolarisation – T Wave: Ventricular repolarisation – U wave: not often recorded, it may be due to the slow https://www.thevinemedicalcenter.com/heart-health/ repolarisation of papillary muscles. – P-R and S-T are at zero potential Depolarisation: contraction of the any muscle associated with the – P-R interval is due to conduction delay in AV node electrical change. Repolarisation: Phase of recovery/relaxation. – S-T segment is related to average duration of plateu regions of individual ventricular cells. The University of Sydney Page 9 Short Video Courtesy: https://youtu.be/RYZ4daFwMa8?si=UjZQy8Ed7NOcYk_E The University of Sydney Page 10 Origin of ECG – Depolarisation initiated at the SA node spreads as a wave front across the two atria, and also at a higher speed along the three inter-nodal tracts to the AV node → P Wave on ECG – There is a delay at the AV node, then the wave front travels at high speed down the His bundle in the intraventricular septum, dividing the two ventricles. – The His bundle branches into the left and right bundles, which in turn connect to the Purkinje system which conducts the wave fronts along much of the endocardial surfaces of the two ventricles. The University of Sydney Page 11 Ventricular Depolarisation & Repolarisation – The wave fronts then spread more slowly through the normal myocardium. – They spread from the inside to the outside of the ventricles. Wave fronts travel with higher velocity in the direction of the fibre orientation. – The morphology of the resulting ECG recorded on the chest surface depends on the orientation of the heart, the active recording electrode and the reference electrode The University of Sydney Page 12  ECG Vector – Electro cardiographers developed a simple model to represent the electrical activity of the heart. – The effects of summing all the electrical activity in the heart can be represented by an electrical dipole whose magnitude and direction is constantly changing. – The scalar magnitude of the ECG is then the dot product of the dipole and the electrode orientation. Approximate dipole field of the heart at the peak of the R wave. The dipole consists of the points of equal positive and negative charge separated from one another and denoted by the dipole moment vector M. The University of Sydney Page 13 ECG Recording : Limb Leads: Einthoven triangle – In order to measure ECG waveform, a differential recording between two points on the body are made. – Each differential recording is known as lead. – Einthoven defined three leads as RA-Right Arm – Lead I (00angle)= VLA –VRA LA-Left Arm – Lead II (60 angle)= VLL –VRA 0 LL-Left Leg – Lead III (1200angle) = VLL –VLA RL-Right Leg – Body assumed to be resistive at ECG frequencies, and the four limbs can be considered as wire Einthoven’s law: attached to torso. Lead II=Lead I + Lead III – This way of recording is called bipolar recording. – The right leg electrode acts to reduce interference The University of Sydney Page 14 Wilson Central Terminal – The remaining lead positions use a common reference, known as Wilson's central terminal – This consists of the right and left wrists joined to the left ankle, each through a suitably large resistor, e.g. 5 k ohms The University of Sydney Page 15 Precordial leads: Wilson terminal concept – These are 6 electrodes placed for the physicians to look at the ECG in transverse plane. – These electrodes are placed at various anatomically defined positions in the chest wall. – The potential between the electrode and the Wilson’s central terminal is the ECG for that lead. The University of Sydney Page 16 Augmented Leads: Modified Wilson terminal(Goldberger’s terminal) – Unipolar leads. – These leads are derived from the three leads I, II and III. – These leads use Wilson central terminal as their negative pole. – Lead augmented vector right (aVR) has the positive electrode on the right arm. – Lead augmented vector left (aVL) has the positive electrode on the left arm. – Lead augmented vector feet (aVF) has the positive aVR-Augmented electrode on the left feet. Vector Right aVR= VRA – (VLA +VLL)/2 aVL-Augmented Vector Left aVL= VLA – (VRA +VLL)/2 aVF-Augmented aVF=VLL – (VRA +VLA)/2 Vector Feet. The University of Sydney Page 17 Why “Augmented?” – The three active electrodes: the right and left wrists and the left ankle are part of Wilson central terminal – In practice this means that when one limb is an active electrode, it is shunted by the resistance that is part of the Wilson's central terminal circuit. – To avoid this shunting, the active limb is connected by a resistor of half the value of the others, to the non- inverting input of the amplifier → The limb is not connected to the Wilson's central terminal. – This is known as the augmented lead system The University of Sydney Page 18 12-LEAD ECG configuration – 12-lead ECG provides spatial information about the heart's electrical activity in 3 approximately orthogonal directions: – Right ⇔ Left – Superior ⇔ Inferior – Anterior ⇔ Posterior – Bipolar limb leads (frontal plane): Lead I: RA (-) to LA (+) (Right Left, or lateral) Lead II: RA (-) to LL (+) (Superior Inferior) Lead III: LA (-) to LL (+) (Superior Inferior) – Augmented unipolar limb leads (frontal plane): Lead aVR: RA (+) to [LA & LL] (-) (Rightward) Lead aVL: LA (+) to [RA & LL] (-) (Leftward) Lead aVF: LL (+) to [RA & LA] (-) (Inferior) – Unipolar (+) chest leads (horizontal plane): Leads V1, V2, V3: (Posterior Anterior) Leads V4, V5, V6:(Right Left, or lateral) The University of Sydney Page 19 Arrhythmias The University of Sydney Page 20 Arrhythmia (abnormal rhythm) – Sinus tachycardia (fast beat, normal with exercise) The University of Sydney Page 21 Arrhythmia – Bradycardia (slow beat) : Slow arrhythmias are mostly due to failure of a wave front to travel through the AV node or adjacent conductive fibres. This is known as heart block. The University of Sydney Page 22 Arrhythmia – Sinus arrhythmia (normal in children) The University of Sydney Page 23 Arrhythmia – 1st degree heart block – prolonged PR interval The University of Sydney Page 24 Arrhythmia – Dropped beat – missing QRS complex – In second degree block, there is complete block every few cycles for just one beat The University of Sydney Page 25 Arrhythmia – 2nd degree heart block – missing QRS complexes The University of Sydney Page 26 Arrhythmia The University of Sydney Page 27 Arrhythmia – Premature beat The University of Sydney Page 28 Arryhthmia – Sino-Atrial nodal block, ventricular beat (no P waves, ventricles act as pacemaker, A-V nodal rhythm) The University of Sydney Page 29 Arrhythmia – Atrial paroxysmal tachycardia Paroxysmal: sudden occurrence or acute exacerbation of symptom. The University of Sydney Page 30 Arrhythmia – Ventricular paroxysmal tachycardia The University of Sydney Page 31 Arrhythmia – Atrial flutter – 2:1 and 3:1 A-V block The University of Sydney Page 32 Arrhythmia – Atrial fibrillation The University of Sydney Page 33 ECG Acquisition System The University of Sydney Page 34 ECG Clinical Acquisition system The University of Sydney Page 35 ECG electrodes – Many types of electrodes available – Most widely used: Wet electrodes – typically, with Ag/AgCl – Electrode paste or jelly between the electrode and skin – A local solution is formed at the skin- electrode interface → used to sense the potential of the body – Dry electrodes – Metal electrodes – No wet paste – Usually have small multiple pointed tip to penetrate skins for better contact impedance The University of Sydney Page 36 Source: google image  The series model in Figure 10 needs to be modified to account for the fact that the impedance does not increase to infinity as the frequency tends to zero. This is done by adding a parallel resistance R2 (as shown in Figure 11) which accounts for the ECG electrodes electrochemical processes taking place at the electrode−electrolyte interface. The values of R1, R2 and C depend on the electrode area, surface condition, current density and the type and concentration of electrode paste used. (Typical values are R1 = 2kW, R2=10kW and – Equivalent circuit of electrode-skin interface C=10mF.) C Electrode Body electrolytes R1 E R2 – Typical values: Figure 11: Equivalent circuit of the Ag−AgCl interface – R1 = 2 Ifk Movement artefact the electrode is moved with respect to the elctrolyte, this mechanically disturbs the – R2 = 10potential k until equilibrium can be re−established. If a pair of electrodes are in contact with distribution of charge at the interface and results in a momentary change of the half−cell – C = 10appears an electrolyte and one moves while the other remains stationary, a potential difference µF between the two during this motion. This potential is referred to as moverment artefact and can be a serious cause of interference in the measurement of ECG (or any other biopotential). The University of Sydney Overall equivalent circuit Page 37 Using the simple model of the electrode−electrolyte interface of Figure 11 as well as the Motion artefact – Inevitable when using electrodes. – If the electrode is moved with respect to the electrolyte, this mechanically disturbs the distribution of charge at the interface. – If a pair of electrodes are in contact with an electrolyte and one moves while the other remains stationary, a potential difference appears between the two during this motion. – This potential is referred to as motion artefact and can be a serious cause of interference in the measurement of ECG (or any other biopotential). The University of Sydney Page 38 Example of ECG with movement artefact Still OK to see Completely distorted The University of Sydney Source: google image Page 39 ECG amplifiers – ECG signal is differential signal: – Typical peak-peak voltage: 1mV – Need amplifier with high Common Mode Rejection Ratio (CMRR) – To reject common signal from electrical field interference – Source of interference: – AC main power line: the body continuously acts like an antenna, picking up 50 Hz noise – This noise can be as large as 20 mV The University of Sydney Page 40 ECG amplifier – Solution: instrumentation amplifier (recall lecture 3) The University of Sydney Page 41 Driven right leg circuit – WHY is DRL necessary? : – In modern ECG recording systems, the patient is often not grounded. – Instead, the right leg electrode is connected as shown to the output of an auxiliary op−amp. – The common−mode voltage on the body is sensed by two averaging resistors Ra, inverted and fed back to the right leg through R0. – This circuit actually drives a very small amount of current (less than 1 μA) into the right leg to equal the displacement currents flowing in the body. – The body therefore becomes a summing junction in a feedback loop and the negative feedback from this circuit drives the common−mode voltage to a low value. The University of Sydney Page 42  https://www.youtube.com/watch?v=WpSSg3f72as The University of Sydney Page 43  https://www.youtube.com/watch?v=WpSSg3f72as The University of Sydney Page 44 The University of Sydney Page 45 Driven right leg (cont.) – The circuit also helps to increase the patient’s safety. – If an abnormally high voltage should appear between the patient and ground due to electrical leakage or other means, the auxiliary op−amp in the right leg circuit saturates. – This effectively ungrounds the patient since the amplifier can no longer drive the right leg. – The resistance R0 between the patient and ground is usually several MΩ and is therefore large enough to protect the patient. – With a 5 MΩ resistor, for examples, and a supply voltage of 10 V, the amplifier will saturate at a current of approximately 2 μA. The University of Sydney Page 46  ECG Signal filtering The most common filters used are: – (a) low pass filters, used to reduce high frequency noise, such as that caused by voluntary muscle movement in the arms and chest; – (b) high pass filters, used to remove low frequency drifts due to movement of electrodes; – (c) notch filters, used to remove 50 or 60 Hz mains frequency induced in the leads; – and (d) band pass filters used to extract QRS complexes. – High pass filters are also used to differentiate the ECG in order to enhance higher frequency components, such as the QRS complex. A band pass filter with centre frequency of 17 Hz and Q of 3.3 is deal for maximising the QRS complex in a healthy person. – Lower centre frequencies are required in some disease states. The University of Sydney Page 47  ECG Signal Filtering – Careful Consideration (a): True ECG (b) Low pass filtered (25 Hz cut-off): high frequency distortion (amplitude of QRS) (c ) High pass filtered (1 Hz cut-off): Low frequency distortion (P and T waves) The University of Sydney Page 48  Amplifier saturation and cut-off Arrows: cut-off due to amplifier saturation (a) original signal, (b) clipping of peaks due to positive saturation in the amplifier, (c) clipping of negative peaks due to cut-off in the amplifier. The University of Sydney Page 49 Fetal ECG Measurement – Fetal ECG signal is usually week (50µV or less). – Hence hard to detect the heartbeat of the fetus by attaching electrodes to the abdomen of the mother during labor due to excess artefacts. – QRS complex of mother is much stronger than the fetus and it will be difficult to determine fetal heartrate electronically. – The circuit shown uses three electrodes; one on mother’s chest, one on the upper part of the fundus of the uterus and one on the lower part of the uterus. – ECG of the mother from the top electrodes and fetus+mother from bottom two. – Threshold detector turns off the analog switch when mother’s QRS is detected. Hence only fetal ECG is collected. The University of Sydney Page 50 ECG Classification The University of Sydney Page 51 ECG Classification (diagnosis) – ECG machines, heart monitors, pacemakers, defibrillators can automatically classify (diagnose) heart rhythms – Steps: ✓ Signal acquisition ✓ Filtering – QRS complex identification – Feature (metric) extraction – classification The University of Sydney Page 52  QRS Complex Identification Detection of ventricular beats. Software or hardware senses if either a positive or negative amplitude threshold is exceeded. Logic circuits or software determine a ventricular beat has occurred if either or both threshold senses are detected within a certain time period The University of Sydney Page 53  QRS Complex Identification A time threshold prevents multiple detection of a single beat. The thresholds are adaptive; being set to a fraction of the previous positive and negative sensed peaks, and decaying towards either zero or the noise level. The University of Sydney Page 54 QRS Identification - Summary – Adaptive thresholding - +/- thresholds – Sense event when threshold exceeded – Time window of sensed events – Events within window exclude each other (avoids multiple detections of same QRS) The University of Sydney Page 55 ECG Classification (diagnosis) – ECG machines, heart monitors, pacemakers, defibrillators can automatically classify (diagnose) heart rhythms – Steps: ✓ Signal acquisition ✓ Filtering ✓ QRS complex identification – Feature (metric) extraction – Classification The University of Sydney Page 56 Feature (metric) Extraction – Measurements extracted from the ECG. – Time, length, area, shape, angle, order of events, rate, frequency. – Features should be orthogonal, i.e. not contain the same information, uncorrelated. The University of Sydney Page 57 Commonly used features – The main feature extracted from the ECG is the heart rate, or the interval between QRS complexes. This is usually measured between the peaks of the R waves, or the peaks of the Q waves, if they are the dominant wave in the ventricular complex. – Other intervals measured are PR and RT and the duration of the QRS complex and the P and T waves. – Amplitudes of the P, Q, R, S and T waves are measured, including their sign, as is the integrated area of the QRS complex – Deviation of the ST segment from the zero axis is also measured. The University of Sydney Page 58 Features The difference in amplitude of the positive and negative peaks of the QRS complex is a commonly used feature. The areas of the curves under these peaks may also be used. The University of Sydney Page 59 Features The ratio of the areas above and below the x-axis found within the analysis window on a detected QRS complex may be used as a feature. The University of Sydney Page 60 Features The order of the various waves is noted as is the presence or absence of delta waves (small amplitude waves just before the ventricular complex). A note is made of the relative counts of each type of wave. The order of the peaks in the QRS complex is described by an angle. The University of Sydney Page 61 Features Some commercial analysers also measure slopes of the P, Q, R, S and T waves and the slope of the ST segment. The University of Sydney Page 62 Classifier – Types – Rules-based (semantic) – Statistical (Bayesian) – Fuzzy logic – Neural network – Hierarchical – combination of above The University of Sydney Page 63 Rules-based (semantic) Classifier – Feature – interval between beats – Rules – If interval < 375 ms for 6 out of 8 successive beats then ventricular fibrillation – If 375 < interval < 500 ms for 7 out of 10 successive beats then ventricular tachycardia – If 500 < interval < 660 ms for 8 out of 10 successive beats then sinus tachycardia – If 660 < interval < 1200 ms for 4 out of 6 successive beats then normal rhythm – If 1200 < interval for 4 out of 5 successive beats then bradycardia The University of Sydney Page 64  Statistical Classifier – Features – QRS peak height – QRS width – Classes – normal sinus rhythm (NSR) – ventricular tachycardia (VT) The University of Sydney Page 65 Classifier Training, Testing and Use – Training data – sample of rhythms classified manually by expert – Test data – different sample of rhythms classified automatically by classifier and manually by expert; results compared – Use – automatic classification, sampled cross check by expert in case of population variability The University of Sydney Page 66 This Week Lab The University of Sydney Page 67 Familiarising Embedded systems in BME application In the Week 7experiment, we will detect the temperature changes using the following apparatus, namely: 1. Temperature Sensor 2. Servo motor 3. Arduino Uno. Temp sensor/servo 3. Computer. motor The University of Sydney Page 68 Thermistor circuit The University of Sydney Page 69 Introduction to Arduino – Arduino is a low-cost microcontroller board that can do amazing things. – Arduino consists of a physically programmable board and a software (IDE) that runs on your computer. – The IDE is used to write and upload the computer code (simplified C++) to the physical board. – We are using ABX00033 in our lab. – In your lab 1, you will be able to read the signal from the sensor using Arduino as well as control a servo motor using Adrino. – Later in Week 8, you will be developing GUI. The University of Sydney Page 70 End Don’t forget to prepare for tomorrow’s lab The University of Sydney Page 71  BMET3802/9802 Biomedical Instrumentation Week 08: EEG and EMG Dr Sandhya Clement Lecturer School of Biomedical Engineering The University of Sydney  I would like to acknowledge the Traditional Owners of Australia and recognise their continuing connection to land, water and culture. I am currently on the land of the Cardigal people of the Eora Nation, and I pay my respects to their Elders, past, present and emerging. I further acknowledge the Traditional Owners of the country on which you are on and pay respects to their Elders, past, present and future. The University of Sydney Please note that most of the Lecture content are prepared and most of the Figures in this week’s lecture slides have been taken from the following textbooks and research papers Webster, J. G. (Ed.). (2009). Medical instrumentation: application and design. John Wiley & Sons. Li, Bohao, Tianshuo Cheng, and Zexuan Guo. "A review of EEG acquisition, processing and application." Journal of Physics: Conference Series. Vol. 1907. No. 1. IOP Publishing, 2021. Jamal, M. Z. (2012). Signal acquisition using surface EMG and circuit design considerations for robotic prosthesis. Computational Intelligence in Electromyography Analysis-A Perspective on Current Applications and Future Challenges, 18, 427-448. Li, G. (2011). Electromyography pattern-recognition-based control of powered multifunctional upper-limb prostheses. Advances in applied electromyography, 6, 99-116. Bronzino&Peterson: https://www-taylorfrancis- com.ezproxy.library.sydney.edu.au/books/mono/10.1201/9781351228671/medical-devices-human-engineering- joseph-bronzino-donald-peterson The University of Sydney Page 3 This lecture: Electroencephalogram (EEG) and Electromyogram (EMG) – EEG During Lecture, I will be completing – Introduction and basics ECG classification (last week lecture, – EEG Electrodes last section) before starting this – EEG acquisition content! – EEG Applications – EMG – Introduction and basics – EMG Electrodes – EMG acquisition – EMG Applications – Muscle Simulator – Introduction – Electrodes The University of Sydney Page 4 Electroencephalogram (EEG) The University of Sydney Page 5  Electroencephalogram (EEG): Basics – Recording of electrical activity associated with the brain. – Frequency of the EEG waveform ranges from 0.1-80 Hz – Amplitude of the signal is of 0.001-1 mV – The electrical activity of the brain is recorded by three different types of electrodes – Scalp electrode – Cortical – Depth electrodes – Electrocorticogram (ECoG): The electrode is placed on the exposed surface (cortex) of the brain – Depth recording: Using the insulated needle electrode to pick up the signal from the neural tissue of the brain. The University of Sydney Page 6 EEG Basics – The characteristics of EEG waveform is highly dependent on the degree of activity of the cerebral cortex. – For example, the waves change markedly between states of wakefulness and sleep. – Much of the time, the brain waves are irregular, and no general pattern can be observed. – Yet at other times distinct patterns do occur. Some of these are characteristic of specific abnormalities of the brain, such as epilepsy. – Others occur in normal persons and may be classified as belonging to one of four wave groups (alpha, beta, theta, and delta) The University of Sydney Page 7 Resting Rhythms of Brain – Alpha Wave: Waves at frequency of 7.5-12 Hz when the person is awake in a quiet, resting state. – Recorded mostly from the occipital region. – At times from parietal and frontal region. – Voltage is ~20-200 µV. – When person’s attention directed to specific activity, this alpha waves are replaced by high frequency low amplitude signals. – The effect on the waveform when opening the eyes in bright light and closing again as shown below The University of Sydney Page 8 Resting Rhythms of Brain – Beta Wave: Waves at frequency of 12-30 Hz when the person is active. – It goes up to 50Hz during intense mental activity – Frequently recorded from parietal and frontal region. – Two major types Beta I and Beta II – Beta I Wave : has twice the frequency of alpha and affected by mental activity by the same way as alpha. – Beta II Wave: Appear during intense activation of central nervous system and during tension. The University of Sydney Page 9 Resting Rhythms of Brain – Theta Wave: Waves have frequency of 4-7.5 Hz. – Occur mainly at the parietal and temporal regions in kids. – Also, occur during emotional stress in some adults, during emotional stress and disappointment. – Delta Wave: All waves in the EEG below 4 Hz. – These waves occurs only once every 2 or 3 sec. – Mostly occur in deep sleep, in infants. – Also, in the case of serious organic brain diseases. – Occurs only within the cortex, independent of activities in lower region of brains. The University of Sydney Page 10 EEG Waveform Summary: The University of Sydney Page 11 SLEEP PATTERNS – EEG pattern changes as a human goes to sleep The University of Sydney Page 12 The Clinical EEG acquisition – International federation 10-20 system is often used to place the EEG electrodes – This system uses certain anatomical landmarks to standardize the placement of electrode. – The representation of EEG channel is known as montage. https://youtu.be/YfjRhoC2V0E The University of Sydney Page 13 The Clinical EEG acquisition – Bipolar montage: each channel measures the difference between two adjacent electrodes (Figure C). – Referential montage: each channel measures the difference between one electrode and reference electrode (Figure A). – Average reference montage: each channel measures the difference between one electrode and average of all the other electrodes (Figure B). Zhang, Z., Ren, Y., Sabor, N., Pan, J., Luo, X., Li, Y.,... & Wang, G. (2020). DWT-Net: Seizure Detection System with Structured EEG Montage and Multiple Feature Extractor in Convolution Neural Network. Journal of Sensors, 2020. Page 14 The University of Sydney Non-ideal factors of EEG Measurement – The main kind of interferences with EEG are – Thermal Noise: generated by the random thermal motion of charge carriers inside an electrical conductor, which happens regardless of any applied voltage. – Flicker Noise: known as 1/f noise, is a signal with a frequency spectrum that falls off steadily into the higher frequencies. – Power Line interference: generated by power Interference reduction circuit line, so the frequency of it is around 50/60 Hz. – Electrode offset voltage: generated from charge accumulation between the metal and electrode gel caused by chemical interaction. It may saturate the amplifier. The University of Sydney Page 15 EEG acquisition System: Saptono, Debyo, Bambang Wahyudi, and Benny Irawan. "Design of EEG Signal Acquisition System Using Arduino MEGA1280 and EEGAnalyzer." MATEC Web of Conferences. Vol. 75. EDP Sciences, 2016. The University of Sydney Page 16 EEG Application: Brain disease detection a) non-invasion brain edema monitor. b) EEG and spectrum of different spaces in normal people's brains. c) EEG and spectrum of different spaces in patients' brains. Comparing the EEG and spectrum between normal people and brain edema patients, it can be concluded that EEG from normal people is clear around 5Hz while patients' around 10Hz The University of Sydney Page 17 EEG application: Parkinson’s disease detection The framework which uses deep learning, combining the feature extraction of instantaneous frequency and power spectrum entropy with LSTM neural network model to judge the EEG signals of Parkinson's disease then diagnose whether there is Parkinson's disease with the people. The University of Sydney Page 18 Electromyogram (EMG) The University of Sydney Page 19 Importance of Muscle signals An estimated 1.71 billion people worldwide living with a musculoskeletal disorder Rehabilitation services are the primary therapy Example muscle-related disorders/diseases: Muscular Rheumatism Limb Loss Muscle spasticity Muscular dystrophy ; atrophy Myositis Myopathy (1)Cieza, A.; Causey, K.; Kamenov, K.; Hanson, S. W.; Chatterji, S.; Vos, T. Global estimates of the need for rehabilitation based on the Global Burden of Disease study 2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet 2020, 396, 2006–2017. (2)Hertling, D.; Kessler, R.; Shimandle, S. A. Dimensions Of Critical Care Nursing, 4th ed.; Management of Common Musculoskeletal Disorders 5; Lippincott Williams \& Wilkins: Philadelphia, 1990; Vol. 9; p 279. The University of Sydney Page 20 Electromyogram (EMG) – EMG signal represents the electrical activity associated with the muscle. – This is different from nerve conduction study (ENG) – The device used for recording the EMG signal is called Electromyograph. – Applications Clinical neuro-physiology studies Study of the body organ movements (Kinesiology) Controlling artificial limbs by real EMG signals Animation producing Data fusion The University of Sydney Page 21 Single Motor Unit (SMU) – Whenever the muscles of the body are to be recruited for a certain activity, the brain sends excitation signals through the Central Nervous System (CNS). – Motor unit is the single junction point at which motor nerve fiber and the bundle of muscle fibers meet. – When an SMU get activated, it generate a Motor Unit Action Potential (MUAP/MUP) – The activation from the Central Nervous System is repeated continuously for as long as the muscle is required to generate force. – This continued activation produces motor unit action potential trains. – The trains from concurrently active motor units superimpose to produce the resultant EMG. The University of Sydney Page 22 Cellular Physiology of the Muscle Fiber – Change in electrical potential of the extracellular fluid, which surrounds muscle fibres. – It occurs due to the influx and efflux of ions across innumerable muscle fibres. Cellular physiology of a muscle fiber. The exchange of ions between the extracellular fluid and the muscle fiber occur in a wave-like motion, producing an action potential. This triggers several intracellular signaling pathways, including the release of calcium from the sarcoplasmic reticulum into the intracellular environment. The binding of calcium to troponin results in actin-myosin cross-bridging and an ensuing muscle fiber contraction. (2021, Jett van der Wallen, unpublished) The University of Sydney Page 23 Motor Unit Potential – Theoretical waveform of an MUAP measured is using a surface electrode. – The action potential from the innervation zone (IZ) is propagated bilaterally along the muscle fibers. – The direction of the waveform will reverse depending on whether the surface electrode is proximal or distal to IZ (Figure a). – The normal MUAP is triphasic, consisting of larger first- and second-phase peaks and a smaller third phase peak (Figure b) Nishihara, K., & Isho, T. (2012). Location of electrodes in surface EMG. In EMG Methods for Evaluating Muscle and Nerve Function. IntechOpen. The University of Sydney Page 24 EMG Signal – Theoretical EMG signal from action potentials propagated along muscle fibers. – The action potentials propagated along muscle fibers are attenuated according to the distance between the muscle fibers and the surface electrodes and are superimposed in surface EMG Nishihara, K., & Isho, T. (2012). Location of electrodes in surface EMG. In EMG Methods for Evaluating Muscle and Nerve Function. IntechOpen. Page 25 The University of Sydney The University of Sydney Page 26 EMG Signal Processing The University of Sydney Page 27 Electrodes in EMG Bar Electrode (eg: forDry – Two main types of electrodes Gelled EMG Electrode EMG Electrode) – Surface electrodes: Placed on the surface of the skin Gelled EMG Electrodes Dry EMG Electrodes – Inserted electrode: Inserted to the muscle Needle EMG Electrode Fine Wire Electrode Needle Electrode Fine Wire Electrode The University of Sydney Page 28 Electrode Placement SMES – surface myoelectric sensor eIMES – epimysial implantable myoelectric sensor iIMES – intramuscular implantable myoelectric sensor TMES – transcutaneous myoelectric sensor (2021, Jett van der Wallen, unpublished) The University of Sydney Page 29 Electrode Configuration: – Monopolar Configuration – Single electrode in the skin with respect to a reference electrode is used. – It is simple – But not recommended – This captures all electrical signal in the vicinity of the detecting surface The University of Sydney Page 30 Electrode Configuration: – Bipolar Configuration – Acquire EMG signal using two EMG detecting surfaces with the help of a reference electrode – Two detecting electrodes are placed 1-2 cm apart. – This is a most commonly used configuration. – The differential amplifier suppress the signal common to both the electrodes and amplifies the difference. Multipolar (more than two detecting electrodes) techniques together with muti differential amplifier stages are often employed in modern systems to reduce the cross talk and noise interference. The University of Sydney Page 31 EMG Acquisition System Also known as EMG or electromyographic Records and transmits myoelectric signals, by interfacing with muscle tissue Components: 1. Electrodes 2. Connection to analog circuit 3. Analog circuit 4. Connection to external device The University of Sydney Page 32 EMG Acquisition System: Basic Circuit The University of Sydney Page 33 Surface EMG Acquisition example Shobaki, M. M., Malik, N. A., Khan, S., Nurashikin, A., Haider, S., Larbani, S.,... & Tasnim, R. (2013, December). High quality acquisition of surface electromyography–conditioning circuit design. In IOP conference series: materials science and engineering (Vol. 53, No. 1, p. 012027). IOP Publishing. The University of Sydney Page 34  Non ideal factors for EMG Muscle crosstalk 1. Reduce interelectrode spacing 2. Reduce vertical electrode-muscle distance (i.e. use IMES, TMES) 3. Muscle-nerve interfaces (e.g. TMR, RPNI) 4. Increase N electrodes & machine-based approaches Motion artefacts 5. Electrode fixation 6. Electrode geometric design 7. Wireless transmission of power & data 8. Analog circuits (e.g. high pass filters) The University of Sydney Page 35  Myoelectric Sensor Applications 1. Assistive Devices Supernumerary limbs & exoskeletons Functional electrical stimulation (to be discussed later) Personal electronic devices Virtual surgery 2. Diagnostics Musculoskeletal health & disease Biomechanics The University of Sydney Page 36 EMG Application: Neuropathic Muscle disease Diagnostics Sadikoglu, F., Kavalcioglu, C., & Dagman, B. (2017). Electromyogram (EMG) signal detection, classification of EMG signals and diagnosis of neuropathy muscle disease. Procedia computer science, 120, 422-429. The University of Sydney Page 37 EMG Application: Myoelectric Prostheses Li, Guanglin. "Electromyography pattern-recognition-based control of powered multifunctional upper-limb prostheses." Advances in applied electromyography 6 (2011): 99-116. commercially available upper-limb myoelectric prostheses use a pair of muscles (usually an agonist/antagonist pair) to control one degree of freedom (DOF): one EMG signal from a flexor muscle and one from an extensor muscle The University of Sydney Page 38 EMG Application: Pattern Recognition based prosthesis control Using a pattern classification technique, the distinguishing characteristics of EMG patterns can be used to identify a variety of different intended movements. Once a pattern has been classified, a command is sent to a prosthesis controller to implement the movement The University of Sydney Page 39 EMG Application: Neuro control prosthesis Targeted muscle reinnervation (TMR) Brain computer interface (BCI) Peripheral nerve interface (PNI) Three emerging neural-machine interface techniques for control of neuroprostheses. The University of Sydney Page 40 Muscle Simulators The University of Sydney Page 41  Muscle Stimulators Utilise the same principles as myoelectric sensors, with a few differences. Mainly: 1. Larger electrodes 2. Stimulating analog circuit 3. More power delivery required The University of Sydney Page 42 1. Larger electrodes – Larger to prevent “burn” – dissipate electricity/heat – If surface electrodes, commonly utilise “TENS” electrodes – Transcutaneous electrical nerve stimulation – Functional Electrical Stimulation: Commonly in conjunction with myoelectric sensors (e.g. SMES) – Could also used IMES or TMES for higher muscle specificity The University of Sydney Page 43 Functional Electrical Stimulation (FES) – Applies electrical stimulation to paralysed/weak muscles (e.g. spinal cord injury) – Improve mobility or locomotion Power Muscle Stimulator Myoelectric Sensor Artificial muscle contraction The University of Sydney Feedback Page 44 2. Stimulating analog circuit – For those who are interested – We will only be looking at myoelectric sensing circuits – You need to know: – Shape: sine/square – Frequency: 12-50 Hz – Pulse amplitude: 25-120 V – Pulse duration: 0-300 µs – Current amplitude: 16-20 mA Bronzino and Peterson, Medical Devices and Human Engineering, 2014, 1st Ed. The University of Sydney Page 45 3. Power Supply (for implantable systems) – Transmitting and Receiving coil – Higher compensations – Variations in distance of different users – Skin movement – Coil-to-coil position changes during daily usage – Vast differences in power consumption – Stimulator electrodes: 20-30V – Circuit: 5V – Hence voltage regulators required – Taxes external power transmitter – Increases implant internal power dissipation The University of Sydney Page 46 This Week Lab The University of Sydney Page 47 Heartbeat Rate Measurement (Week 7 & Week 8) -Continuation to Week 7 Lab -More emphasis on Python Programming Last week, we used Arduino and temperature sensors to develop a temperature measurement device and plot these measurements in the IDE. However, we are not able to process or save the data. So, this week, we will learn how to cope with serial data in Python and then develop a simple GUI to control the device read and plot data and save them as csv files The University of Sydney Page 48 Week-9 The University of Sydney Page 49 Week 9: – Lecture: Introduction to Biomedical Imaging – Lab: Introduction to Medical Image processing using Python The University of Sydney Page 50 End Don’t forget to prepare for tomorrow’s lab The University of Sydney Page 51  BMET3802/9802 Biomedical Instrumentation Week 09: Introduction to Biomedical Imaging Dr Sandhya Clement Lecturer School of Biomedical Engineering The University of Sydney  I would like to acknowledge the Traditional Owners of Australia and recognise their continuing connection to land, water and culture. I am currently on the land of the Cardigalpeople of the Eora Nation, and I pay my respects to their Elders, past, present and emerging. I further acknowledge the Traditional Owners of the country on which you are on and pay respects to their Elders, past, present and future. The University of Sydney This lecture – An overview of biomedical imaging and image analysis – Part 1: Basics of biomedical imaging – Part 2: Introduction to Image acquisition systems and processing – Part 3:A sneak peak on the importance of imaging: Application examples The University of Sydney Page 3 Week 9 Lab: Introduction to Image Processing – Objectives In this Week 9 laboratory, you will: – Learn to setup your code file. – Learn one way to load an image. – Learn one way to display an image. – Gain a better understanding about the matrix structure that stores image pixels. – Access pixel values from different parts of the image. – Manipulate the pixel values directly to change the appearance of the image. – Access the image metadata. The University of Sydney Page 4 Biomedical imaging Where do we use it? Where does it come from? The University of Sydney Page 5 Motivation: Does biomedical imaging matter? Watson, Crick, and Wilkins Ernst Ruska Electron microscopy Nobel Prize 1962 Ruska, Binnig, Rohrer Nobel Prize 1986 Wilhelm Röntgen, Rosalind Franklin, “Photo 51” “left hand of Albert von Kölliker” X-ray diffraction of DNA Röntgen, Nobel Prize 1901 Computed Tomography Cormack, Hounsfield Nobel Prize 1979 Magnetic Resonance Imaging Lauterbur, Mansfield Nobel Prize 2003 The University of Sydney Page 6 What is biomedical imaging? Localised measurement of biophysical effect within a living system The University of Sydney Page 7 What is biomedical imaging? Localised measurement of biophysical effect within a living system – Localised: positioning (spatial) information is important – Measurement: varying ranges/values of information – Biophysical effect: physical effect of biological phenomenon – Living system: a system that is biological (is or was alive) – Within: can be inside – Implies minimal perturbation to the system The University of Sydney Page 8 Clinical context – imaging departments Leica Biosystems Whole slide Scanner GE BrightSpeed CT Imaging Suite 3D ultrasound via the GE Voluson E8 system: Dr Wolfgang Moroder The University of Sydney Page 9 Scientific contexts – research labs – Studying biology – Testing new therapies Stained Chromosomes – Examining disease Cell cycle (multi-nucleated) mechanisms Cortical neuron – Investigating drug interactions – Discovering functions and processes The University of Sydney Page 10  From my research… (a) Nano drug(VP) uptake by the PANC-1 cells (a) (b) Treatment progress (nano drug is PLGA- VP-PFOB, RDT is X-ray+nanodrug) Clement et al., Biomedicines, 2021 Clement et al., International journal of molecular sciences, 2021 The University of Sydney Page 11 Modern day context – from the home Nokia 7650 iPhone Braun et al., “Telemedical wound care using a new generation of mobile telephones: a Gunter et al., “Feasibility of an image-based mobile health protocol feasibility study”, Archives of dermatology, 141(2): 254-258, 2005. for postoperative wound monitoring”. Journal of the American College of Surgeons, 226(3): 277-286, 2018. The University of Sydney Page 12 What is an image (as a concept)? – In our context – A representation – In visual form – Of some object or subject more definitions! https://www.merriam-webster.com/dictionary/image The University of Sydney Page 13 What is an image? (Data perspective) – A matrix How do – Rows → height – Columns → width we create – Values in matrix determine what this? image looks like The University of Sydney Page 14 Acquisition: a diagrammatic view The 3 S Subject Signal Sensor (visible light) The University of Sydney Page 15 Acquisition: a diagrammatic view The 3 S The 4 S Subject Signal Sensor (visible light) Storage The University of Sydney Page 16 Subject – Tissues – Cell – Bones – Population – Muscles – Nuclei – Organs – Chromosome – Vessels – Genes – Activity – Cytoplasm – Brain – A person – Metabolic – Whole body The University of Sydney Page 17 Signals and sensors – Sensors: detect some form of physical phenomenon – Radiation – ‘Standard cameras’ – visible light – CT and x-rays – x-ray radiation – PET – photons emitted from positron annihilation – Acoustic energy – ultrasound – Electromagnetic waves and fields – MRI, MEG, Electrical Impedance The University of Sydney Page 18 The key challenge How do we transform the measured biophysical signal to localised pixel or voxel data? The University of Sydney Page 19 The key challenge How do we transform the measured biophysical signal to localised pixel or voxel data? The University of Sydney Page 20 Image data (overview) – Visual data – The actual content of the image – The “picture” – Metadata – Information about the image Time of capture Scanner used Image structure The University of Sydney Page 21 What is an image? (Data perspective) – A matrix – Rows → height – Columns → width – Values in matrix determine what image looks like The University of Sydney Page 22 3D Data: Pixels and voxels The University of Sydney Page 23 Multi-channel images The University of Sydney Page 24 Video data: frames … The University of Sydney Page 25 Intensity – The magnitude of the pixel/voxel value Increasing Intensity – Affects its appearance, usually through brightness or colour – High intensity → bright – Low intensity → dark The University of Sydney Page 26 Intensity – The magnitude of the pixel/voxel value Increasing Intensity – Affects its appearance, usually through brightness or colour – High intensity → bright – Low intensity → dark The University of Sydney Page 27 Spot the differences Version 1 original Version 2 The University of Sydney Page 28 Resolution – Resolution – The level of detail captured by the image – Not just about the size – Pixel resolution – Spatial resolution – Spectral resolution – Temporal resolution – Radiometric/colour/contrast resolution The University of Sydney Page 29 Pixel/voxel resolution – Total number of pixels in an image or voxels in a volume – Can be written, broken down into each axis/dimension 4000 x 2000 (width x height) = 8 000 000 pixels = 8 Megapixels The University of Sydney Page 30 Spatial resolution – The size of a pixel or voxel in ‘real’ space – The real dimensions of a pixel – The smallest size that can be represented by the pixels in the image The University of Sydney Page 31 Spectral resolution – The wavelengths (or frequencies) of electromagnetic radiation that can be distinguished Lipid + Lactate total choline total creatine Öz et al., “Clinical Proton MR Spectroscopy in Central Nervous System Disorders”, Radiology 270(3): 658-679, 2014. The University of Sydney Page 32 Temporal resolution – The time taken to acquire an image – The time between images – Between frames in the context of videos and dynamic images, i.e., the sampling period – Image capture is not instantaneous – The signal (light or radiation) takes time to travel from the body to the detector https://www.youtube.com/watch?v=_UJFCnni9as The University of Sydney Page 33 Radiometric/colour/contrast resolution – Level of variability that can be encoded by an individual pixel or voxel – The range of intensity values that are possible – Often expressed in bits (powers of 2) – 1 bit: 21 (2) possible values – 2 bits: 22 (4) possible values – 8 bits: 28 (256) possible values – RGB 83 bits: (23)3 (16 million+) values The University of Sydney Page 34 Image Acquisition Systems X-Ray CT PET Microscopy The University of Sydney Page 35 X-rays – Wavelength: ~10pm-10nm (10 000pm) – Ionising radiation – Wavelengths >100 or 200 pm are absorbed completely by the body – Not ideal for medical applications (body absorbs all radiation, nothing to detect) – Can we measure how much radiation is absorbed by each body part? The University of Sydney Page 36 Examples High Lower Object Detector X-ray source or Image High Lower Sensor Object Medium High The University of Sydney Page 37 Size/thickness also plays a role! High Lower Object X-ray source Detector Image High Object Medium The University of Sydney Page 38 X-rays and human tissues – Tissue density affects how much x-ray radiation is absorbed – Calcium (bone) absorbs quite a lot – Adipose tissue (fat), muscles and other soft tissues absorb less – Water absorbs even less – Air absorbs very little – Attenuation: The ‘stopping power’ of tissue – We will get a look at a table of values in a bit The University of Sydney Page 39 It’s not that simple – Rayleigh/Coherent Scatter – Compton Scatter electron incident x-ray incident x-ray – Interaction with electron – Interaction with whole atom – Ionisation more matter = more scatter The University of Sydney scatter reduces contrast! Page 40 Scatter and contrast Barnes GT. Contrast and scatter in x-ray imaging. Radiographics. 11(2):307-23, 1991. The University of Sydney Page 41  Hounsfield units (CT number) – pixel values in CT https://en.wikipedia.org/wiki/Hounsfield_scale F Fortin, 2020: https://radiopaedia.org/cases/hounsfield-scale-diagram The University of Sydney Page 42 The University of Sydney Page 43 Tomographic image reconstruction (3D volumes) The University of Sydney Page 44 Scanners don’t cover the whole body scan happens here GE Revolution CT GE Discovery IQ (PET/CT) The University of Sydney Page 45 Helical scanning source Mark Hammer, https://xrayphysics.com The University of Sydney Page 46 X-ray absorption… in a circle! The University of Sydney Page 47 Backprojection The University of Sydney Page 48 Backprojection and Filtering 180 angles Filtered backprojection The University of Sydney Page 49 CT image Reconstruction Mark Hammer, https://xrayphysics.com The University of Sydney Page 50 Anatomical planes (left-right) (front-back) a.k.a. axial plane (top- bottom) … Images usually acquired in axial/transverse plane The University of Sydney Page 51 Slices and slice thickness 1mm 1mm – Slice is a chunk of the body – Has some 3mm thickness – Slice Thickness … is the third spatial resolution The University of Sydney Page 52 Effect of slice thickness 140 slices * 3mm (420mm) 512 pixels * 0.98mm 140 pixels * 3mm (500mm) (420mm) transpose voxel 512 pixels * 0.98mm matrix (500mm) compare 420mm? 500mm 512 pixels * 0.98mm 420mm? 512 pixels * 0.98mm original (500mm) (500mm) 430 pixels * 0.98mm correct transpose & resample (420mm) aspect voxel matrix ratio The University of Sydney Page 53 A handy trick for accurate aspect ratios When rotating and resampling: 1. Ensure rows and columns have the same resolution If not: slices #𝑐𝑜𝑙𝑢𝑚𝑛𝑠 ∗ 𝑐𝑜𝑙𝑢𝑚𝑛 𝑟𝑒𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑛𝑒𝑤 #𝑐𝑜𝑙𝑢𝑚𝑛𝑠 = 𝑟𝑜𝑤 𝑟𝑒𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 rows Resize image to #rows x new #columns 2. Calculate number of slice pixels for a correct aspect ratio columns #𝑠𝑙𝑖𝑐𝑒𝑠 ∗ 𝑠𝑙𝑖𝑐𝑒 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 𝑠𝑙𝑖𝑐𝑒 𝑝𝑖𝑥𝑒𝑙𝑠 = 𝑟𝑜𝑤 𝑟𝑒𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 Resize image to #rows x # columns x slice pixels The University of Sydney Page 54 Positron emission tomography (PET) The University of Sydney Page 55 PET and radiotracers Higher # of decays = – Over time radiotracer higher pixel disperses across the body values – Radiotracer concentrates in particular regions based on underlying biochemistry – Signal generated as radiotracer decays – Count number of decay Radiotracer events at every location injected – Eventually excreted The University of Sydney Page 56 Event detection – Each event picked up by two detections 180 degrees apart – Detector does not rotate – At any given time, multiple events may occur – There is a time-of-flight difference between event detection at the two sensors International Atomic Energy Agency – Human Health Campus The University of Sydney Page 57 https://humanhealth.iaea.org/HHW/MedicalPhysics/NuclearMedicine/ImageAnalysis/3Dimagereconstruction/index.html Concentration, decay, and scan time Scan duration and image quality Regional concentration at different times The University of Sydney Badawi et al., J Nucl Med, vol. 60 no. 3, 299-303, 2019. Page 58 Scatter and attenuation – Not all events will be detected: – Absorption by the body (attenuation) – Scattering outside detector field of view – Several methods for correction – For attenuation transmission scans and CT attenuation maps are two common methods – Techniques are out of the scope of this unit, but we aware that the images acquired may not be perfect (even after correction) The University of Sydney Page 59 What does it mean? – Subjects with same condition in exact same location may look different because of: – Body mass – Dose of radiotracer – Time since dose was administered (radiotracer decays) – Sensitivity of detector – Time-of-flight sensitivity The University of Sydney Page 60 Standard Uptake Value (SUV) – Normalises image based on dose, mass, and time since dose – Avoids the need for blood sampling to determine actual radiotracer concentration – Can be sensitive to: – Noise – Resolution – Definition of regions of interest (e.g., measuring SUV within an area) – Care must be taken in multi-centre imaging The University of Sydney Page 61 Standard Uptake Value (SUV) 𝑃𝐸𝑇 𝑆𝑈𝑉𝐵𝑊 = Can overestimate in obese patients as adipose tissue has low uptake 𝐷𝑜𝑠𝑒Τ𝐵𝑊 BW = body weight 𝑃𝐸𝑇 Can underestimate as adipose tissue is not entirely inert 𝑆𝑈𝑉𝐿𝐵𝑀 = Requires specific equations for LBM 𝐷𝑜𝑠𝑒Τ𝐿𝐵𝑀 LBM = lean body mass, BMI = body mass index 9270 × 𝐵𝑊 9270 × 𝐵𝑊 𝐿𝐵𝑀𝑓𝑒𝑚𝑎𝑙𝑒 = 𝐵𝑊 8780 + 244 × 𝐵𝑀𝐼 𝐿𝐵𝑀𝑚𝑎𝑙𝑒 = 𝐵𝑀𝐼 = 6680 + 216 × 𝐵𝑀𝐼 ℎ𝑒𝑖𝑔ℎ𝑡 2 “FDG PET/CT: EANM procedure guidelines for tumour imaging: version 2.0”, Eur J Nucl Med Mol Imaging (2015) 42:328–354. "Reevaluation of the standardized uptake value for FDG: variations with body weight and methods for correction." Radiology 213.2 (1999): 521-525. The University of Sydney Page 62 Microscopy The University of Sydney Page 63 Basic microscope summary Microscopes: instrument that uses lenses to view small objects Basic principle: Bend light to Some microscopes bend streams of electrons focus on and enhance objects Magnification: Apparent Characterised by a factor or a ratio enlargement of the object Lenses: Can have one or May be imperfect, may have surface aberrations (usually) many in sequence Curvature can cause distortions The University of Sydney Page 64 Brightfield microscopy – Standard light microscope – Contrast can be low for thin translucent samples (like cells!) – Out-of-focus materials appear Adherent HeLA cervical cancer cells Ali, Rehan, et al. "Automatic segmentation of adherent blurry biological cell boundaries and nuclei from brightfield microscopy images." Machine Vision and Applications 23.4: 607-621, 2012. https://medical-dictionary.thefreedictionary.com/bright- field+microscope The University of Sydney Page 65 Brightfield microscopy Micrograph of Whatman lens tissue paper. Bright field illumination. 10x magnification, 1.559 μm/px. By Zephyris - Own work, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=10762658 The University of Sydney Page 66 Darkfield microscopy – Some source light is stopped to create an outer illumination ring – Illumination focused towards the sample – Directly transmitted light is blocked but sample scattered light enters – Contrast enhanced resulting in a dark background Darkfield Microscope By CNX OpenStax - https://cnx.org/contents/[email protected] , CC BY 4.0, https://commons.wikimedia.org/w/index.php?curid=53712288 The University of Sydney Page 67 Darkfield microscopy – Some source light is stopped to create an outer illumination ring – Illumination focused towards sample – Directly transmitted light is blocked but sample scattered light enters – Contrast enhanced resulting in a dark background – Low light levels may require stronger illumination, damaging the sample Micrograph of Whatman lens tissue paper. Dark field illumination. 10x magnification, 1.559 μm/px. By Zephyris - Own work, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=10762664 The University of Sydney Page 68 Fluorescence microscopy – Sample labelled with fluorescent stain or fluorescent protein, antibodies to antigens etc. – Sample illuminated with wavelengths for the spectral characteristics of the stain/protein – Limitations: – Sample is changed by the introduction of stain/protein – Photobleaching (less fluorescence over time) – Phototoxicity can be enhanced under https://biologyreader.com/fluorescence- Homogeneous immunofluorescence staining illumination microscopy.html pattern of double stranded DNA antibodies on HEp-20-10 cells. – Only labelled structures are By Simon Caulton - Own work, CC BY-SA 3.0, observable https://commons.wikimedia.org/w/index.php? curid=20521932 The University of Sydney Page 69  From my research… (a) Nano drug(VP) uptake by the PANC-1 cells (a) (b) Treatment progress (nano drug is PLGA- VP-PFOB, RDT is X-ray+nanodrug) Clement et al., Biomedicines, 2021 Clement et al., International journal of molecular sciences, 2021 The University of Sydney Page 70 Microscope sample staining – Staining enhances contrast using dyes to define biological tissues, cell populations, organelles – Similar to fluorescent staining – H&E staining is frequently Kleczek et al., "A novel method for tissue segmentation in high- used in histology resolution H&E-stained histopathological whole-slide images." Computerized Medical Imaging and Graphics 79 (2020). The University of Sydney Page 71 Metadata: DICOM, NifTI and TIFF The University of Sydney Page 72 What is metadata? – Data that provides information about other types of data – In the context of biomedical imaging: – What type of image (MR, CT, PET, etc.) is stored in the file? – What scanner was the image acquired with? – What are the dimensions and resolution? – How many grey levels (bit-depth, contrast resolution) used? – What radiotracer dose was administered? – Focus on a few types: DICOM, TIFF, NifTI The University of Sydney Page 73 DICOM – Digital Imaging and Communications in Medicine – Standard for communication and management of medical image data and related information – Used for storage, exchange, transmission, visualisation and results reporting – Defines data dictionary, data structures, file formats, etc. – Ensures image cannot be separated from relevant information by mistake (but can still be deliberately separated) The University of Sydney Page 74  Definitions Exam 1 Exam 2 StudyUID A StudyUID B One or more imaging Study Study Study exams taken during a single visit CT X-ray PET Series Series Series SeriesUID A A set of images from Series a single modality of a study Image Image Image ImageUID A A single element of Image data from a series Image Image Unique identifier for a Image UID component The University of Sydney Page 75 DICOM tags http://dicom.nema.org/medical/dicom/current/output/html/part06.html The University of Sydney Page 76 What use is metadata? – Obtaining information for processing – Resolution and slice thickness for resampling/resizing – Body weight and dose for 𝑆𝑈𝑉𝐵𝑊 calculation – Body weight and height and gender for 𝑆𝑈𝑉𝐿𝐵𝑀 – Analysis of processing outcomes – Older scanner may result in different noise characteristics – Different protocol may result in visual differences The University of Sydney Page 77 Key lessons – Contrast and signal contribute to image quality – Noise can degrade contrast or signal – Acquisition is recording of a biophysical property – Acquisition affects contrast, signal, and noise – Acquisition affects resolution – Knowing acquisition properties can assist in understanding the challenges in later image analysis – Acquisition properties are often stored as meta-data The University of Sydney Page 78 Why process and analyse? – Image acquisition is not perfect – Trade-offs – Adjusting contrast resolution may require adjusting radiation – Larger magnetic field strengths may increase spectral resolution but require more energy to power, and greater electromagnetic protection for the hospital – Patients have different cameras at home – Not every facility has the latest and best equipment – Not all image content is relevant to the task The University of Sydney Page 79 Biomedical imaging Applications A Sneak Peak The University of Sydney Page 80 Research concept Image Data Prediction Image ML/AI methods that Classification Data analyse image data Delineation … Image Data I design this The University of Sydney Page 81  Photographs to medical images photos re-train with medical (1 million) images (6700) learn generic things can recognise differences in image data The University of Sydney Page 82 Kumar et al., IEEE JBHI 2017 Fetal ultrasound analysis – Measure anatomical structures in ultrasound images – Use measurements to develop healthy/abnormal growth models identify anatomy identify view measure The University of Sydney Kumar et al. IEEE ISBI 2016; Sridar et al., IEEE JBHI 2017; Sridar et al, UMB 2019. Page 83 Enhancing and processing Fuji Film Digital Radiography PET-CT fusion https://www.fujifilmusa.com/products/medical/digital-x-ray/image- By User:MBq, CC BY-SA 3.0, processing/dynamic-visualization/#images https://commons.wikimedia.org/w/index.php?curid=20682407 The University of Sydney Page 84 Decision making Simulating surgery as part of treatment planning Radiotherapy dose planning Oshiro and Ohkohchi. “Three-dimensional liver surgery simulation: computer-assisted https://medicine.umich.edu/dept/radonc/educ surgical planning with three-dimensional simulation software and three-dimensional ation-training/resources-references/contouring- printing.” Tissue engineering part A, 23(11-12), pp.474-480, 2017. atlas-gi-radiotherapy-planning/imrt- unresectable-pancreatic-cancer-case-2/case-2- supplemental The University of Sydney Page 85 Covid 19 classification – Differentiate various lung infections based on various classification. – Both Binary and multi class classifications are done – Deep learning approach is used for feature extraction and classification Kuzinkovas, D., & Clement, S. (2023). The detection of covid-19 in chest x-rays using ensemble cnn techniques. Information, 14(7), 370. The University of Sydney Page 86 Interested to learn more – BMET5933 Biomedical Image Analysis is an option – No Prerequisites – An understanding of biology (1000-level), experience with programming (ENGG1801, ENGG1810, BMET2922 or BMET9922 – This unit is available in Semester 1 – Python is used in Labs. The University of Sydney Page 87 Sources – Motivation: Does biomedical imaging matter? – CT image (1970s): http://www.impactscan.org/CThistory.htm – Head CT scanner: By Philipcosson - English Wikipedia: en:File:Emi1010.jpg., Public Domain, https://commons.wikimedia.org/w/index.php?curid=1006690 – MR: By AndyGaskell, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=54271471 – Electron Microscopy polio: By Photo Credit:Content Providers(s): CDC/ Dr. Fred Murphy, Sylvia Whitfield - This media comes from the Centers for Disease Control and Prevention's Public Health Image Library (PHIL), with identification number #1875.Note: Not all PHIL images are public domain; be sure to check copyright status and credit authors and content providers., Public Domain, https://commons.wikimedia.org/w/index.php?curid=816997 – Electron Microscopy replica: By J Brew, uploaded on the English-speaking Wikipedia by en:User:Hat'nCoat. - originally posted to Flickr as Electron Microscope Deutsches Museum, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=5309032 – Photo 51: Digitisation of Raymond Gosling’s x-ray diffraction image (WP:NFCC#4), Fair use, https://en.wikipedia.org/w/index.php?curid=38068629 The University of Sydney