Biomedical Data Acquisition and Processing Study Guide

Overview of Biomedical Engineering and Measurement Methods

  • Role of the Biomedical Engineer:

    • Serves as a technical specialist in high-tech biomedical devices and equipment across hospital sectors, healthcare establishments, and professional health settings.

    • Operates within clinical environments, medical research institutions, and the broader healthcare technology field.

  • Biomedical Equipment Functional Pipeline:

    • Sensors: Interface with biological tissue to capture physical or chemical phenomena.

    • Signal Acquisition and Processing: Conditions, amplifies, filters, and digitizes signals.

    • Analysis and Display: Interprets signals using algorithms and visualizes data for clinicians.

    • Network Connection and Storage: Facilitates telemetry, cloud storage, electronic health records, and system integration.

  • Taxonomy of Non-Invasive Measurements:

    • Non-invasive measurements do not breach skin or internal mucosal barriers. They are categorized into direct, propagated, and indirect measures.

    • Breathing & Respiratory Dynamics:

    • Respiration rate, breathing amplitude, airflow velocity.

    • Tidal volume, airway pressure, oxygen consumption (VO2VO_2), carbon dioxide production (VCO2VCO_2).

    • Cardiovascular & Hemodynamic Parameters:

    • Electrocardiography (ECG), Gyrocardiography (GCM), Vectorcardiography.

    • Vessel blood flow rate via Ultrasound (US) Doppler.

    • Microvascular capillary flow via Capillaroscopy.

    • Peripheral volume changes via Plethysmography.

    • Acoustic phenomena: Heart sounds, lung sounds, muscle sounds.

    • Hemodynamic pressures: Arterial blood pressure, transcutaneous partial pressure of oxygen (TCpO2TCpO_2), transcutaneous partial pressure of carbon dioxide (TCpCO2TCpCO_2).

    • Neurological & Central Nervous System Activity:

    • Electroencephalography (EEG), Magnetoencephalography (MEG), Nuclear Magnetic Resonance (NMR) imaging/spectroscopy.

    • Nystagmography (eye movement recording), Audiogram (auditory response).

    • Infrared (IR) spectroscopy, voice/acoustic analysis, indirect intracranial pressure estimation.

    • Cognitive and autonomic state indicators: Emotional response, vigilance levels, mental workload.

    • Musculoskeletal & Movement Parameters:

    • Electromyography (EMG) and muscular activity tracking (kinematics, velocity, force output).

    • Body shape and structural dimension measurement.

    • Plantar pressure mapping and force plate measurement.

    • Body acceleration, movement tracking, and ballistocardiography.

    • Dermatological & Thermal Parameters:

    • Skin surface temperature, galvanic skin resistance, skin bioimpedance, microvascular skin blood flow, skin hydration levels, colorimetry, and tactile sensitivity.

    • Body heat loss measurement.

    • Fetal & Obstetric Monitoring:

    • Non-invasive fetal heart rate tracking and fetal movement monitoring.

    • Excretory & Metabolic Parameters:

    • Urinary flow rate measurement.

Non-invasive measurement parameters mapped to body

Minimally Invasive and Fully Invasive Medical Devices

  • Minimally Invasive Measurement Techniques:

    • Involve minimal penetration into body orifices or minor tissue incisions, offering a balance between clinical accessibility and physiological measurement accuracy.

    • Intracranial pressure via specialized catheters.

    • Core body temperature: Tympanic temperature, oral temperature, and rectal temperature.

    • Esophageal blood flow monitoring using specialized esophageal probes.

    • Internal fetal monitoring: Fetal heart rate, intrauterine pressure monitoring, and scalp fetal pH sampling during labor.

    • Laparoscopic optical and physical sensors.

    • Intravascular/endovascular probes deployed within cardiac structures or brain vessels.

    • Continuous glucose monitoring (CGM) via subcutaneous sensors.

    • Cardiac output determination via thermodilution techniques.

    • Gastrointestinal parameters: Internal pressure, internal pH, and internal EMG.

    • Ingestible or implanted telemetry capsules: Internal temperature and pH.

    • Direct arterial pressure via intra-arterial catheters.

    • Accessibility vs. Accuracy Trade-off: Increasing invasiveness yields higher measurement accuracy, whereas decreasing invasiveness improves patient comfort and clinical accessibility.

Minimally invasive measurement techniques mapped to body
  • Invasive Measures & Active Implants:

    • Deep Brain Stimulation (DBS): Brain implants for tremor control in Parkinson's disease.

    • Cochlear Implants: Direct electrical stimulation of auditory nerve fibers.

    • Cardiac Active Implants: Implantable Holter monitors, cardiac pacemakers, implantable cardioverter-defibrillators (ICD), and cardiac resynchronization therapy (CRT) devices.

    • Automated Drug Delivery: Implantable insulin pumps with integrated dosing management.

St. Jude Medical implantable pacemakersCochlear implant system diagram
  • Artificial Organs & Advanced Prosthetics:

    • Total Artificial Heart (TAH): Mechanical replacement for end-stage heart failure.

    • Artificial Pancreas: Closed-loop continuous glucose sensing and automated insulin infusion.

    • Robotic Prosthetic Arms: Myoelectric or neural-controlled robotic limbs for functional restoration.

Artificial heart deviceProsthetic robotic arm in use

Sensor Principles and Influencing Factors

  • Physiological Parameters and Transducer Measurement Principles:

    • Pressure: Piezoresistive sensing, piezoelectric effect, capacitive sensing on silicon substrates.

    • Strength / Force: Piezoresistive sensing, piezoelectric effect, capacitive sensing on silicon substrates.

    • Fluid Flow: Thermodilution, Ultrasound (US) Doppler, Laser Doppler, differential pressure sensing, electromagnetic induction, force transducers.

    • Thermal Parameters: Thermoresistive sensors (RTD, thermistors), thermoelectric effect (thermocouples), infrared (IR) radiometry, microwave radiometry.

    • Position, Speed, & Acceleration: Resistive potentiometers, magnetic encoding/Hall effect, optical encoders, piezoelectric accelerometers.

    • Ionic Concentration in Biological Fluids: Ion-Sensitive Field-Effect Transistors (ISFET), Chemical Field-Effect Transistors (CHEMFET) utilizing thin/thick films or monolayer membranes.

    • Metabolic Enzyme Activity: Substrate immobilization techniques, specific chemical reaction detection, acoustic resonance frequency tracking, fluid viscosity measurements.

    • Respiratory Gas: Bioimpedance spectroscopy, substance-specific gas conductivity, acoustic resonance frequency, viscosity measurement.

    • Cerebral Activity, Emotional Reactivity, Vigilance, & Stress: Electrical skin conductance, bioimpedance analysis, acoustic resonance, viscosity changes.

    • Electrical Biopotentials (ECG, EEG, EMG): Surface/needle electrodes, potential difference measurement, contact resistance sensing.

    • Tissue Characterization: Nuclear Magnetic Resonance (NMR), tissue impedance spectroscopy, ultrasound scattering, thermal or optical spectroscopy.

    • Magnetic Fields: Superconducting Quantum Interference Devices (SQUID).

    • Biochemical & Physical Parameters: Optical mechanisms including light diffusion, luminescence, and optical absorption spectrum analysis.

  • Factors Influencing Sensor Selection and Design:

    • Transduction Principle: Optical, mechanical, electrical, chemical, or thermal conversion.

    • Power & Autonomy: Battery capacity, power consumption, energy harvesting, wireless power transfer.

    • Patient Comfort: Level of invasiveness, device geometry, materials biocompatibility.

    • Deployment Location: Intensive Care Unit (ICU), ambulatory monitoring, home healthcare environments.

    • Measurement Objective: Passive diagnostic measurement, active continuous monitoring, rehabilitation feedback.

    • Commercial & Manufacturing Constraints: Component cost, manufacturing complexity, reusability vs. single-use disposable design.

Biomedical Signal Characteristics and Processing Pipeline

  • Nature of Biological Signals:

    • Composed of useful physiological information combined with unwanted noise.

    • Requires a hybrid sequence of analog signal preprocessing and digital signal processing.

  • Biological vs. Technical Signal Acquisition Pipeline:

    • Biological Vision Pipeline: Physical Image →\rightarrow Eye (biological sensor) →\rightarrow Optical Nerve (transmission path) →\rightarrow Brain (processing unit).

    • Technical Acquisition Pipeline: Physiological Parameter →\rightarrow Sensor / Electrodes →\rightarrow Analog Preprocessing (amplification, hardware filtering) →\rightarrow Analog-to-Digital Converter (sampling, quantization) →\rightarrow Digital Processing (Fourier transform, AI models, digital filtering) →\rightarrow Computer Output.

Biological signal acquisition and processing flow
  • Key Specificities of Biomedical Signals:

    • Accessibility vs. Invasiveness: Direct signal access requires higher invasiveness.

    • Low Signal-to-Noise Ratio (SNR): Demands specialized low-noise hardware and advanced denoising algorithms.

    • Non-Stationarity: Statistical properties change over time, requiring time-frequency or adaptive processing.

    • High Structural Complexity: Demands sophisticated digital signal processing models.

    • Physiological Link: Requires robust mathematical modeling to link measured signals to underlying physiological mechanisms.

  • Taxonomy of Signal Characteristics:

    • Classification by Source: Bioelectric, Biomechanical, Bioimpedance, Biochemical, Biomagnetic.

    • Classification by Application: Cardiology, Neurology, Myology, Pulmonology.

    • Classification by Deterministic vs. Stochastic Nature:

    • Deterministic Signals (Mathematical expression exists):

      • Periodic: Waveshape repeats strictly periodically over time (e.g., pure sine wave).

      • Quasi-periodic: Waveshape repeats almost periodically with subtle beat-to-beat variations (e.g., Electrocardiogram - ECG).

      • Transient: Waveshape occurs as a single isolated event (e.g., single-cell action potential response).

    • Stochastic Signals (Governed by probabilistic distributions):

      • Stationary: Statistical properties (mean, variance, spectral content) do not change over time (e.g., steady-state alpha brain waves).

      • Non-stationary: Statistical properties evolve continuously over time (e.g., long-term raw Electroencephalogram - EEG).

Classification tree of deterministic and stochastic signals

Complete Signal Acquisition Chain Architecture

  • 7-Stage Data Acquisition Chain Structure:

    • Stage 1 - Sensor: Converts physiological parameter into electrical output; typically exhibits low amplitude, high output impedance, and high noise content.

    • Stage 2 - Input Amplifier: Provides electrical isolation, impedance matching, and initial differential amplification.

    • Stage 3 - Analog Filters: Removes out-of-band noise, suppresses powerline hum, prevents spectral aliasing.

    • Stage 4 - Gain Amplifier: Scales signal amplitude to match the standardized input range of the A/D converter.

    • Stage 5 - Storage: Retains raw or processed signals temporarily or permanently.

    • Stage 6 - Analog-to-Digital Conversion (ADC): Discretizes continuous time (sampling) and continuous amplitude (quantization).

    • Stage 7 - Digital Processing: Executes software filtering, spectral analysis, feature extraction, or machine learning algorithms.

  • Signal Transformation State Across the Acquisition Chain:

    • Input State: Low Amplitude, High Impedance, Noisy Analog Signal.

    • Conditioned State: Standard Amplitude, Adapted Impedance, Clean Analog Signal.

    • Output State: Discretized Digital Representation ready for microprocessor computation.

Front-End Signal Conditioning: Input Amplifiers and Isolation

  • Sensor Stage Considerations:

    • Sensor output impedance, signal voltage level, full dynamic range, linearity, long-term stability, calibration requirement, and physical packaging.

  • Optical Isolation Amplifiers:

    • Patient Safety Requirement: Prevents dangerous leakage currents from flowing through the patient by maintaining complete galvanic isolation.

    • Uses optically coupled isolation amplifiers powered by an isolated independent power supply on the patient side.

Optically coupled isolation amplifier diagram
  • Impedance Matching Principles:

    • Loaded measured voltage equation:     EMeasured=ESensor×ZInputZInput+ZSensorE_{\text{Measured}} = E_{\text{Sensor}} \times \frac{Z_{\text{Input}}}{Z_{\text{Input}} + Z_{\text{Sensor}}}

    • If input impedance equals sensor impedance (ZInput=ZSensorZ_{\text{Input}} = Z_{\text{Sensor}}), signal attenuation is 50%50\%:     EMeasured=12ESensorE_{\text{Measured}} = \frac{1}{2} E_{\text{Sensor}}

    • To prevent signal loading, input impedance must be vastly larger than sensor impedance (ZInput≫ZSensorZ_{\text{Input}} \gg Z_{\text{Sensor}}), giving:     EMeasured≈ESensorE_{\text{Measured}} \approx E_{\text{Sensor}}

    • Implemented using unity-gain (G=1G = 1) buffer amplifiers featuring extremely high input impedance (ZInputZ_{\text{Input}}) and very low output impedance (ZOutputZ_{\text{Output}}).

  • Instrumentation Amplifier Architecture:

    • Two-Stage Circuit Topology:

    • First Stage: Dual operational amplifiers in non-inverting buffer configuration providing ultra-high input impedance with differential gain:       Gain1=1+R2R1\text{Gain}_1 = 1 + \frac{R_2}{R_1}

    • Second Stage: Standard differential amplifier stage converting differential signal to single-ended output:       Gain2=R2′R1′=1\text{Gain}_2 = \frac{R'_2}{R'_1} = 1

    • AD620 Integrated Instrumentation Amplifier:

    • Input Impedance: Ze=10 GΩZ_e = 10\,\text{G}\Omega (1010 Ω10^{10}\,\Omega).

    • Common Mode Rejection Ratio: CMRR≈100 dB\text{CMRR} \approx 100\,\text{dB}.

    • High-performance bioelectric front-end IC used extensively in ECG monitoring.

AD620 instrumentation amplifier pinoutAD620 Instrumentation Amplifier setupECG acquisition circuit with AD620 and driven right leg AD705J
  • Common Mode Rejection Ratio (CMRR):

    • Differential Amplifier Real Output Voltage:     X=Gd(x1−x2)+12GCM(x1+x2)X = G_d (x_1 - x_2) + \frac{1}{2} G_{\text{CM}} (x_1 + x_2)

    • CMRR Definition:     CMRR=GdGCM\text{CMRR} = \frac{G_d}{G_{\text{CM}}}

    • Where GdG_d represents differential gain and GCMG_{\text{CM}} represents common-mode gain.

    • Operational amplifier LM741 exhibits a nominal CMRR=90 dB\text{CMRR} = 90\,\text{dB}.

    • Real-world CMRR degrades with increasing signal frequency.

    • Impact of Resistor Tolerances (TT) on CMRR:     CMRR≈1+R2R14T\text{CMRR} \approx \frac{1 + \frac{R_2}{R_1}}{4T}

    • For 1%1\% resistor tolerance (T=0.01T = 0.01) and gain ratio R2R1=10\frac{R_2}{R_1} = 10:       CMRR=275≈48 dB\text{CMRR} = 275 \approx 48\,\text{dB}

    • For 0.1%0.1\% precision resistor tolerance (T=0.001T = 0.001) and gain ratio R2R1=10\frac{R_2}{R_1} = 10:       CMRR=2750≈68 dB\text{CMRR} = 2750 \approx 68\,\text{dB}

CMRR vs Frequency curve
  • Bioelectric Signal Magnitudes and Frequency Bands:

    • Local Field Potential (LFP): Frequency Band = 1 Hz−500 Hz1\,\text{Hz} - 500\,\text{Hz}, Amplitude Range = 10 μV−5 mV10\,\mu\text{V} - 5\,\text{mV}.

    • Action Potential (AP): Frequency Band = 300 Hz−7 kHz300\,\text{Hz} - 7\,\text{kHz}, Amplitude Range = 10 μV−500 μV10\,\mu\text{V} - 500\,\mu\text{V}.

    • Electroencephalogram (EEG): Frequency Band = <1 Hz−100 Hz<1\,\text{Hz} - 100\,\text{Hz}, Amplitude Range = 1 μV−10 μV1\,\mu\text{V} - 10\,\mu\text{V}.

    • Electrocardiogram (ECG): Frequency Band = 5 Hz−500 Hz5\,\text{Hz} - 500\,\text{Hz}, Amplitude Range = 1 mV−10 mV1\,\text{mV} - 10\,\text{mV}.

    • Electromyogram (EMG): Frequency Band = 20 Hz−1 kHz20\,\text{Hz} - 1\,\text{kHz}, Amplitude Range = 100 μV−10 mV100\,\mu\text{V} - 10\,\text{mV}.

Analog Filtering Techniques and Applications

  • Signal-to-Noise Ratio (SNR):   SNR=20log⁡10(PSPN)\text{SNR} = 20 \log_{10}\left(\frac{P_S}{P_N}\right)

    • Where PSP_S is useful signal power and PNP_N is noise power. Biomedical signals intrinsically suffer from low initial SNR values.

  • Origins of Noise in Biomedical Measurements:

    • Thermal noise, electronic component noise, physical/electromagnetic environmental interference (50 Hz/60 Hz50\,\text{Hz} / 60\,\text{Hz} powerline interference), motion artifacts, interference from secondary biological signals (e.g., EMG interference during ECG recording).

  • Analog Noise Mitigation Techniques:

    • Physical hardware shielding using shielded coaxial cables and grounded Faraday cages/antennas.

    • Impedance matching buffer amplifiers.

    • High-CMRR differential instrumentation amplifiers.

    • Frequency-selective analog active/passive filtering.

  • Analog Filter Transfer Characteristics:

    • Linear time-invariant filtering model: x(t)=s(t)+n(t)→H(f)y(t)=s(t)x(t) = s(t) + n(t) \xrightarrow{H(f)} y(t) = s(t).

    • Effective when noise is additive and noise spectrum does not overlap signal spectrum.

    • Filter Topologies: Low-Pass, High-Pass, Band-Pass, Band-Reject (Notch).

    • Filter Parameters: Cutoff frequency (−3 dB-3\,\text{dB} attenuation frequency point, ωc\omega_c), Filter order (determines roll-off steepness in transition band between passband ωf\omega_f and stopband ωs\omega_s).

Real low-pass filter frequency response
  • Specific Clinical Filtering Applications:

    • Low-Pass Filtering: Rejection of high-frequency muscle activation artifacts (EMG noise) from cardiac ECG signals.

    • Notch / Band-Reject Filtering: Suppression of continuous powerline hum (50 Hz50\,\text{Hz} in Europe or 60 Hz60\,\text{Hz} in North America).

    • High-Pass Filtering: Elimination of baseline wander, motion-induced DC offset, and respiration-induced baseline drift without distorting signal morphology.

ECG signal with high frequency noise before and after low pass filteringECG signal notch filter response

Programmable Gain Amplification and Signal Storage

  • Gain Amplifier Stage (Stage 4):

    • Positioned post-filtering to boost conditioned signal amplitude to match standard ADC input voltage ranges (0−5 V0 - 5\,\text{V}, ±5 V\pm 5\,\text{V}, etc.).

    • Programmable Gain Amplifiers (PGA): Microcontroller-selectable gain configurations.

    • LT1991 Precision Programmable Gain Amplifier:

    • Offers pin-configurable gain options: 11, 22, 3.253.25, 44, 55, 5.55.5, 77, 1010, 1111, 1313, and 1414.

    • Configured via input selection logic across pin sets P1, P3, P9, M1, M3, and M9.

    • Contains internal matched precision resistor networks (50 kΩ50\,\text{k}\Omega, 150 kΩ150\,\text{k}\Omega, 450 kΩ450\,\text{k}\Omega) and feedback capacitors (4 pF4\,\text{pF}).

LT1991 programmable gain choices diagram
  • Storage Technologies (Stage 5):

    • Permanent Storage Methods:

    • Analog: FM-modulated magnetic tape reels, thermal millimeter strip chart paper.

    • Digital: Magnetic Hard Disk Drives (HDD), solid-state Flash memory drives (USB), optical discs (CD/DVD), centralized cloud storage servers.

    • Temporary Storage Methods:

    • Analog or digital storage oscilloscopes for real-time waveform inspection.

Analog-to-Digital Conversion (ADC) Principles

  • Core Operations of A/D Conversion (Stage 6):

    • Sampling: Discretizes continuous time into discrete time intervals at sample rate FeF_e (or SRSR).

    • Quantization: Discretizes continuous voltage amplitude into discrete numeric representation using NN binary bits.

  • Shannon-Nyquist Sampling Theorem and Aliasing:

    • Nyquist Criterion:     Fe≥2×FMaxF_e \ge 2 \times F_{\text{Max}}

    • Case 1 (Fe≥2FMaxF_e \ge 2 F_{\text{Max}}): Non-overlapping spectrum repeats; continuous signal can be reconstructed perfectly without distortion.

    • Case 2 (Fe<2FMaxF_e < 2 F_{\text{Max}}): Overlapping spectral copies cause irreversibly corrupted signal profiles termed spectral aliasing (spectrum folding / repliement de spectre).

    • Anti-Aliasing Filter: Low-pass filter installed prior to sampling switch with cutoff frequency fc≤Fe2f_c \le \frac{F_e}{2}. Example: LTC1068-25 8th-order switched-capacitor low-pass filter IC.

Spectrum aliasing due to undersamplingAnti-aliasing low pass filter prior to sampling switch
  • Quantization and Resolution Calculations:

    • Quantization Step Size (qq) for input range PEPE:     q=PE2Nq = \frac{PE}{2^N}

    • Bit Depth Resolution Breakdown:

    • 8 bits→256 discrete levels→0.39%8\,\text{bits} \rightarrow 256\,\text{discrete levels} \rightarrow 0.39\% of Full Scale.

    • 10 bits→1,024 discrete levels→0.098%10\,\text{bits} \rightarrow 1,024\,\text{discrete levels} \rightarrow 0.098\% of Full Scale.

    • 12 bits→4,096 discrete levels→0.024%12\,\text{bits} \rightarrow 4,096\,\text{discrete levels} \rightarrow 0.024\% of Full Scale.

    • 16 bits→65,536 discrete levels→0.0015%16\,\text{bits} \rightarrow 65,536\,\text{discrete levels} \rightarrow 0.0015\% of Full Scale.

    • 20 bits→1,048,576 discrete levels→0.000095%20\,\text{bits} \rightarrow 1,048,576\,\text{discrete levels} \rightarrow 0.000095\% of Full Scale.

    • 24 bits→16,777,216 discrete levels→0.000006%24\,\text{bits} \rightarrow 16,777,216\,\text{discrete levels} \rightarrow 0.000006\% of Full Scale.

    • Signal Degradation Rule: Reducing bit depth by 1 bit decreases achievable Signal-to-Noise Ratio by −6 dB-6\,\text{dB}. Utilizing the total dynamic range (PEPE) of the ADC is essential.

  • ADC Parameters and Error Classification:

    • Dynamic input voltage span (PEPE).

    • Bit resolution (NN) and voltage step (qq).

    • Conversion duration time (tct_c), defining maximum allowable sample rate.

    • Clock source configuration (internal crystal oscillator vs. external pacer clock).

    • Standard ADC Error Types:

    • Offset Error (Erreur de décalage): Constant voltage shift across all converted values.

    • Scale/Gain Error (Erreur d'échelle): Slope deviation in converter transfer curve.

    • Linearity Error (Erreur de linéarité): Non-linear response across converter input range.

ADC offset scale and linearity errors

ADC Architectures and Multi-Channel Data Acquisition Systems

  • Detailed Converter Architectures:

    • Successive Approximation Register (SAR) ADC:

    • Employs binary search (dichotomy) logic with a DAC, voltage comparator, and successive approximation register.

    • Requires exactly NN clock cycles per conversion.

    • Balanced architecture offering medium speed and high resolution.

SAR ADC block diagram and binary search tree
  • Flash (Direct Comparison) ADC:

    • Uses a precision resistor voltage ladder feeding 2N−12^N - 1 parallel voltage comparators into a priority encoder.

    • Provides sub-microsecond conversion time.

    • Exponential hardware complexity (2N2^N comparators).

Flash ADC comparator array architecture
  • Sigma-Delta (ΣΔ\Sigma\Delta) ADC:

    • Uses oversampling, 1-bit DAC feedback, analog integration, high-speed comparator, and digital decimation filtering.

    • Low cost, extremely high resolution (up to 24 bits), but limited sample rate (suitable for audio, motor control, and low-frequency biological signals up to ~20 kHz).

Sigma-Delta ADC architecture with decimation filter
  • ADC Performance Domain Mapping:

    • High Speed / Low Resolution (4−8 bits4-8\,\text{bits}, 100 MHz−10 GHz100\,\text{MHz} - 10\,\text{GHz}): Flash ADCs.

    • Medium Speed / High Resolution (10−16 bits10-16\,\text{bits}, 100 kHz−100 MHz100\,\text{kHz} - 100\,\text{MHz}): SAR ADCs.

    • Low Speed / Ultra-High Resolution (16−24 bits16-24\,\text{bits}, 10 Hz−100 kHz10\,\text{Hz} - 100\,\text{kHz}): Sigma-Delta ADCs.

  • Multi-Channel Data Acquisition Topologies:

    • Conventional Multiplexed System: Single ADC shared across nn channels via analog multiplexer (MUX). Results in non-synchronous channel sampling.

    • Simultaneous Sample and Hold (S/H): Dedicated S/H block per channel prior to MUX and ADC. Maintains exact phase synchronization across channels.

    • Simultaneous Sigma-Delta: Independent ΣΔ\Sigma\Delta ADC per channel.

    • Simultaneous Flash: Independent Flash ADC per channel for ultra-high-speed multi-channel systems.

    • Microprocessor Interface Integration: Utilizes programmable pacer clock timers, interrupt requests (IRQ), data/address buses, and parallel peripheral interface ports (e.g., PIA Ports A & B with control lines CA1, CA2, CB1, CB2).

Digital Signal Processing and Academic Curriculum

  • Digital Processing Stage (Stage 7):

    • Executed in software on embedded microprocessors, Digital Signal Processors (DSP), or Field Programmable Gate Arrays (FPGA).

    • Includes fast Fourier transforms (FFT), digital FIR/IIR filtering, feature extraction, and machine learning/AI classification.

  • Curricular Structure (BM04 - Biomedical Data Acquisition and Processing):

    • Introduction & Acquisition Chain Structure.

    • Electrical Safety in Healthcare (B. Dauzat).

    • Biomedical Sensors and Physics.

    • Advanced Signal Processing Techniques (S. Boudaoud).

    • Clinical Applications of Signal Processing (S. Boudaoud).

    • Connected Architectures & Data Exchange/Storage (J. Laforêt).

    • Embedded Signal Processing Systems (J. Terrien).

    • Connected Objects and E-Health Systems.

    • Industrial Directed Studies (TDs): Implantable Devices (Biotronik) and Green Biomedical Engineering (KickMaker).

  • Academic References and Literature:

    • New Perspectives in Biomedical Signal Processing, J. L. Coatrieux, IEEE EMBS, 1995.

    • A Concise Guide to Intraoperative Monitoring, G. Zouridakis and A. C. Papanicolaou Eds., CRC Press, 2001.

    • The Biomedical Engineering Handbook (2nd Edition), Vol. I, J. D. Bronzino Ed., CRC Press & IEEE Press, 2000.

    • Data Acquisition, G. Ash et al. Eds., Dunod, Paris, 2003.