Biomedical Engineering I Unit 6 Review: Biomedical Signal Processing

Key Concepts and Fundamentals

• Biomedical signal processing involves the analysis and interpretation of signals generated by biological systems
• Signals can be classified as continuous-time or discrete-time, depending on whether they are defined at every point in time or only at specific time intervals
• Analog signals are continuous-time signals that vary continuously in amplitude, while digital signals are discrete-time signals with a finite set of possible amplitude values
  • Analog signals can be converted to digital signals through a process called analog-to-digital conversion (ADC)
  • Digital signals can be converted back to analog signals using digital-to-analog conversion (DAC)
• Sampling is the process of converting a continuous-time signal into a discrete-time signal by measuring its amplitude at regular time intervals
  • The sampling rate, or sampling frequency, determines how often the signal is measured and directly affects the quality of the digital representation
• Quantization is the process of mapping the continuous amplitude values of a sampled signal to a finite set of discrete values
• Nyquist-Shannon sampling theorem states that a signal can be perfectly reconstructed from its samples if the sampling rate is at least twice the highest frequency component present in the signal

Signal Types and Characteristics

• Biomedical signals can be classified based on their origin, such as electrocardiogram (ECG), electroencephalogram (EEG), and electromyogram (EMG)
• Signal characteristics include amplitude, frequency, phase, and power
  • Amplitude refers to the magnitude of the signal at a given point in time
  • Frequency represents the number of cycles per unit time and is measured in hertz (Hz)
  • Phase describes the relative position of a signal with respect to a reference point
  • Power is the rate at which energy is transferred or consumed by the signal
• Signals can be deterministic or stochastic
  • Deterministic signals have a predictable pattern and can be described by a mathematical function
  • Stochastic signals have random variations and are characterized by statistical properties
• Noise is an unwanted component of a signal that can interfere with the desired information
  • Types of noise include thermal noise, shot noise, and flicker noise
• Signal-to-noise ratio (SNR) is a measure of the strength of the desired signal relative to the level of background noise
• Bandwidth refers to the range of frequencies present in a signal

Data Acquisition and Preprocessing

• Data acquisition involves the collection and digitization of biomedical signals using sensors and analog-to-digital converters
• Sensors convert physical quantities (e.g., voltage, pressure, temperature) into electrical signals
  • Examples of sensors used in biomedical applications include electrodes, strain gauges, and thermistors
• Amplification is often necessary to increase the amplitude of weak signals before digitization
• Filtering is used to remove unwanted frequency components or noise from the signal
  • Low-pass filters remove high-frequency components, while high-pass filters remove low-frequency components
  • Band-pass filters allow a specific range of frequencies to pass through while attenuating others
• Preprocessing techniques such as baseline correction, artifact removal, and signal averaging are applied to improve signal quality
• Resampling involves changing the sampling rate of a signal, either by interpolation (increasing the sampling rate) or decimation (decreasing the sampling rate)
• Normalization is the process of scaling a signal to a specific range (e.g., between 0 and 1) to facilitate comparison or analysis

Time Domain Analysis

• Time domain analysis involves the study of signals as a function of time
• Waveform characteristics such as amplitude, duration, and morphology can be examined in the time domain
• Statistical measures like mean, variance, and standard deviation can be calculated to describe the signal's properties
  • Mean represents the average value of the signal over a given time interval
  • Variance and standard deviation quantify the spread or dispersion of the signal values around the mean
• Autocorrelation is a technique used to find repeating patterns or periodicities within a signal by comparing the signal with a delayed version of itself
• Cross-correlation measures the similarity between two different signals as a function of the time lag between them
• Peak detection algorithms are used to identify and locate local maxima or minima in a signal
  • Thresholding and slope-based methods are common approaches for peak detection
• Segmentation is the process of dividing a signal into smaller, meaningful segments for further analysis
  • Examples of segmentation techniques include fixed-length windowing and adaptive segmentation based on signal characteristics

Frequency Domain Analysis

• Frequency domain analysis involves the study of signals in terms of their frequency components
• Fourier transform is a mathematical tool used to convert a signal from the time domain to the frequency domain
  • Discrete Fourier Transform (DFT) is used for discrete-time signals, while Continuous Fourier Transform (CFT) is used for continuous-time signals
  • Fast Fourier Transform (FFT) is an efficient algorithm for computing the DFT
• Power spectral density (PSD) represents the distribution of power across different frequencies in a signal
  • PSD can be estimated using methods like periodogram, Welch's method, or autoregressive modeling
• Spectral analysis techniques are used to identify and quantify the frequency components present in a signal
  • Examples include band power analysis, which measures the power within specific frequency bands, and peak frequency estimation
• Short-time Fourier transform (STFT) is used to analyze non-stationary signals by dividing them into short segments and applying the Fourier transform to each segment
• Wavelet transform is another tool for analyzing non-stationary signals, providing both time and frequency localization
  • Continuous Wavelet Transform (CWT) and Discrete Wavelet Transform (DWT) are two common types of wavelet transforms

Digital Filters and Filtering Techniques

• Digital filters are used to process discrete-time signals by selectively attenuating or amplifying specific frequency components
• Finite Impulse Response (FIR) filters have a finite duration impulse response and are stable and linear-phase
  • FIR filter design methods include windowing, frequency sampling, and optimization techniques
• Infinite Impulse Response (IIR) filters have an infinite duration impulse response and can achieve sharper frequency responses with fewer coefficients compared to FIR filters
  • IIR filter design methods include bilinear transformation, impulse invariance, and pole-zero placement
• Butterworth, Chebyshev, and elliptic filters are common types of IIR filters with different trade-offs between passband ripple, stopband attenuation, and transition bandwidth
• Adaptive filters can adjust their coefficients in real-time based on the characteristics of the input signal or the desired output
  • Examples of adaptive filtering algorithms include Least Mean Squares (LMS) and Recursive Least Squares (RLS)
• Notch filters are used to remove a specific frequency or a narrow band of frequencies from a signal
• Median filtering is a non-linear technique used to remove impulse noise from a signal by replacing each sample with the median value of its neighboring samples

Advanced Signal Processing Methods

• Time-frequency analysis techniques, such as spectrogram and Wigner-Ville distribution, provide a joint representation of a signal in both time and frequency domains
• Empirical Mode Decomposition (EMD) is a data-driven method for decomposing a signal into a set of intrinsic mode functions (IMFs) that represent different oscillatory modes
• Independent Component Analysis (ICA) is a technique for separating a multivariate signal into its underlying independent components, assuming they are statistically independent
• Principal Component Analysis (PCA) is a dimensionality reduction technique that transforms a set of correlated variables into a smaller set of uncorrelated variables called principal components
• Artificial Neural Networks (ANNs) can be used for signal classification, pattern recognition, and feature extraction
  • Convolutional Neural Networks (CNNs) are particularly well-suited for processing signals with spatial or temporal structure
• Support Vector Machines (SVMs) are another popular machine learning technique for signal classification and regression
• Hidden Markov Models (HMMs) are probabilistic models used for modeling and analyzing time series data, particularly in speech recognition and biosignal analysis

Applications in Biomedical Engineering

• ECG signal processing techniques are used for the analysis and interpretation of electrical activity of the heart
  • Applications include heart rate variability analysis, arrhythmia detection, and ST-segment analysis for ischemia monitoring
• EEG signal processing is used to study the electrical activity of the brain, with applications in neuroscience, psychology, and clinical diagnosis
  • Examples include epileptic seizure detection, brain-computer interfaces, and sleep stage classification
• EMG signal processing involves the analysis of electrical activity generated by muscles, with applications in biomechanics, rehabilitation, and prosthetic control
• Biosignal-based emotion recognition uses various physiological signals (e.g., ECG, EEG, GSR) to detect and classify emotional states
• Gait analysis involves the study of human walking patterns using signals from accelerometers, gyroscopes, and force sensors, with applications in sports biomechanics and clinical assessment
• Wearable devices and remote monitoring systems rely on biomedical signal processing for real-time analysis and transmission of physiological data
  • Examples include smartwatches for fitness tracking, continuous glucose monitoring systems for diabetes management, and wireless ECG patches for cardiac monitoring
• Brain-computer interfaces (BCIs) use EEG or other neural signals to establish direct communication between the brain and external devices, enabling control and communication for individuals with severe motor disabilities