Introduction to Live Compression Artifacts

Live compression techniques are essential in fields such as medical imaging (e.g., ultrasound elastography, MRI compression sequences, and optical coherence elastography) and industrial nondestructive testing (e.g., compression testing of polymers, composites, biological tissues, and food products). During a live compression procedure, the system applies a controlled force to the subject while simultaneously acquiring data—whether it be an ultrasound image, an MRI signal, a force-displacement curve, or a strain map. The goal is to measure the material’s or tissue’s response under load in real time, often with high temporal resolution to capture transient behavior.

However, a persistent challenge in these measurements is the appearance of pumping and breathing artifacts. Pumping artifacts manifest as rhythmic fluctuations in the recorded data, often resembling a pulsating pattern that can obscure true mechanical properties. These may appear as alternating bright/dark bands in strain images or sinusoidal oscillations in force readings. Breathing artifacts arise from involuntary respiratory movements or other small body motions, leading to shifts, blurring, or distortion of the signal. Both artifact types reduce the signal-to-noise ratio, compromise repeatability, and can lead to incorrect clinical diagnoses or flawed material characterization. This article provides a comprehensive, authoritative guide to understanding the root causes of these artifacts and implementing effective prevention strategies, drawing on decades of research in elastography, motion correction, and mechanical testing.

What Are Pumping and Breathing Artifacts?

Pumping Artifacts

Pumping artifacts are periodic, usually sinusoidal variations in the measured compression data that occur without a corresponding change in the applied force. They are most often caused by a mismatch between the compression rate and the data acquisition rate, or by resonance in the mechanical system. In ultrasonic elastography, for example, if the transducer oscillates or the compression platen vibrates at a frequency close to the frame rate, the resulting motion creates alternating bright and dark bands in the strain image. These bands “pump” in and out, hiding real tissue stiffness abnormalities. In MRI elastography, pumping artifacts can result from vibrations of the passive driver or from coupling between the motion-encoding gradients and the compression waveform.

In material testing, pumping artifacts can result from hydraulic or pneumatic actuator instabilities, backlashes in mechanical linkages, or even electrical noise in the load cell signal. The artifact appears as a low-frequency oscillation superimposed on the true force-displacement curve, making it difficult to determine the elastic modulus, yield point, or viscoelastic parameters. In high-speed compression tests, the inertial effects of the specimen and fixture can also introduce mechanical ringing that is misinterpreted as a pumping artifact.

Common Frequencies and Identification

Pumping artifacts typically occur at frequencies between 0.5 Hz and 10 Hz, depending on the system configuration. In medical ultrasound, the frame rate is often 10–30 fps, so a pumping artifact at 2–5 Hz will be clearly visible as a slow oscillation across frames. A simple spectral analysis of the time-series data (e.g., using a fast Fourier transform) can reveal the dominant frequency. If that frequency coincides with the mechanical resonance of the compression device (e.g., a spring-mass system), the artifact is almost certainly mechanical. Conversely, if the frequency matches the line frequency (50/60 Hz) or a subharmonic, electrical interference is likely.

Breathing Artifacts

Breathing artifacts stem from respiratory motion—movement of the chest, abdomen, or other body parts during the scanning or testing procedure. In medical imaging, the patient’s breathing shifts the organ or tissue of interest relative to the compression device. For example, during a breast compression ultrasound for cancer screening, the patient’s breathing can cause the lesion to move out of the imaging plane, producing a false stiffness reading. In MRI-guided compression, respiratory motion leads to ghosting artifacts in k-space, corrupting the entire image. Even in industrial contexts, if the specimen is placed on a compliant table that moves with building vibrations (similar to breathing), the resulting drift can mimic respiratory-induced shifts.

Breathing artifacts are typically non-periodic in a strict sense: the respiratory cycle varies in amplitude and period, and breath holds create plateaus. However, during free breathing, the motion approximates a low-frequency drift with superimposed periodic components (the inhalation and exhalation phases). In ex vivo or phantom testing, breathing artifacts are analogous to thermal expansion or creep movements in a specimen under continuous load. If the testing environment is not temperature-stabilized or if the specimen relaxes over time, the resulting drift mimics respiratory-induced movements. Controlling these factors is essential for obtaining reliable mechanical properties.

Why These Artifacts Matter in Live Compression

The presence of pumping and breathing artifacts is not merely a cosmetic issue—it directly impacts the accuracy and clinical utility of live compression data. In diagnostic imaging, these artifacts can cause false positives (e.g., misidentifying a stiff region as a tumor) or false negatives (e.g., missing a real lesion because it is masked by fluctuations). In quantitative elastography, artifacts bias the measured shear wave speed or strain ratio, leading to incorrect staging of liver fibrosis or characterization of breast masses. In the industrial sector, artifacts lead to erroneous material specifications, increased scrap rates, and potential field failures of components.

Furthermore, artifacts degrade reproducibility across operators, machines, and sessions. Without mitigation, it becomes impossible to establish consistent thresholds for disease classification or material acceptance. Regulatory bodies such as the U.S. Food and Drug Administration (FDA) and organizations like ISO require validation that compression-based measurements are free from systematic errors. For instance, ISO 18437 specifies methods for calibrating mechanical testing machines, including artifact quantification. Therefore, investing in artifact prevention is a prerequisite for clinical translation and industrial certification.

Proven Prevention Strategies

1. Subject Preparation and Stabilization

The first line of defense is ensuring that the subject remains as still as possible. For human patients, this involves careful positioning using foam pads, vacuum cushions, or straps to immobilize the body part being compressed. The imaging table or test fixture should be rigid and free of vibration. For ex vivo tissue specimens or phantom tests, embedding the sample in a low-acoustic gel or using a custom holder prevents unwanted translation during compression. In breast imaging, using a dedicated compression paddle with a concave shape can help anchor the tissue.

In industrial testing, clamping the specimen securely and using anti-creep grips minimizes drift. Pre-conditioning cycles (loading/unloading several times before data collection) stabilize the material response by reducing the viscoelastic creep. For soft materials like elastomers or gels, it is advisable to apply a small preload (e.g., 1–5% strain) and allow the force to settle before starting the live compression sequence. Additionally, isolating the test environment from foot traffic, HVAC systems, and nearby machinery with vibration-damping mats or air tables can reduce extraneous motion.

2. Breath-Hold Techniques and Coaching

For medical applications, instructing patients to hold their breath during the critical acquisition window is a simple yet highly effective method. The breath hold should be performed at normal expiration rather than deep inspiration, as the latter can increase intrathoracic pressure and inadvertently shift organs. Provide clear, calm instructions: “Take a normal breath, then breathe out, and hold it for about 10 seconds.” Use a metronome or visual countdown to time the hold with the compression sequence. In MRI, the technologist can signal the patient via the intercom and monitor respiratory motion using a bellows or navigator echoes.

When multiple compressions are required, allow recovery breaths between holds (typically 3–5 seconds of normal breathing). For patients who cannot follow commands (e.g., pediatric or sedated), consider using a respiratory belt to detect breathing cycles and automatically trigger acquisition during the expiratory pause. Some modern ultrasound systems include real-time respiratory motion detection that freezes acquisition when motion exceeds a threshold. For cooperative adult patients, training them to perform a gentle breath hold while relaxing the diaphragm can reduce chest wall motion.

3. Optimizing Compression Parameters

Adjusting the compression rate, amplitude, and waveform can dramatically reduce pumping artifacts. Use a ramp-up rather than an abrupt step function, and apply compression at a moderate, steady speed to avoid exciting mechanical resonances. In elastography, setting the frame rate to an integer multiple of the compression frequency helps synchronize image acquisition and eliminate beat frequencies. If using a motorized compression device, enable closed-loop force control with a low-pass filter (e.g., 5–10 Hz cutoff) to dampen oscillations. For example, a proportional-integral-derivative (PID) controller with a bandwidth of 5 Hz ensures that the actuator does not overshoot or ring.

For manual compression (e.g., freehand elastography), train operators to apply a gentle, constant, and slightly tilted angle to maintain contact without bouncing. Preload the tissue by applying a small static force (e.g., 0.5–1 N) before dynamic acquisition. Some systems allow the operator to see a real-time force indicator; maintaining this within a narrow range (±0.1 N) dramatically reduces artifacts. In material testing, using a trapezoidal compression profile (ramp up, hold, ramp down) gives the specimen time to equilibrate and reduces inertial pumping.

4. Equipment Calibration and Maintenance

Regular calibration of force transducers, position encoders, and imaging systems is critical. A drift in the load cell zero-point can mimic a breathing artifact. Perform daily checks with a known calibration weight or phantom. Lubricate moving parts, check for mechanical play, and verify that mounting bolts are tight. In MRI environments, ensure that gradient coil currents are stable and that the compression actuator is non-ferromagnetic and properly shielded. Many commercial elastography phantoms (e.g., the CIRS Model 049) provide a known stiffness and can be used weekly to verify system performance.

Keep a log of calibration results and schedule preventive maintenance quarterly. An annual performance verification using a standardized phantom (such as those specified in AAPM reports) can reveal subtle changes that precede artifact onset. For motorized test stands, check for wear in the leadscrew or timing belt; backlash as small as 0.1 mm can introduce pulsations in the force reading. Replacing worn components and performing a resonant frequency analysis of the entire mechanical assembly can identify potential issues before they affect data quality.

5. Signal Processing and Post-Processing

Even with perfect hardware and subject cooperation, residual noise remains. Digital filtering is a powerful tool to separate artifacts from true signals. Use a bandstop filter at the known pumping frequency (e.g., 0.5–2 Hz for respiratory motion) or apply adaptive filtering that tracks the artifact rhythm in real time. Motion compensation algorithms, such as cross-correlation-based registration of successive frames, can align images and remove breathing shifts. In material testing, applying a moving average or low-pass Butterworth filter to the force-displacement curve removes high-frequency noise; a cutoff of 10–20 Hz is typical for static compression, while dynamic tests may require higher bandwidth.

Be cautious not to cut off legitimate high-frequency features like stick-slip transitions or tissue discontinuities. Advanced denoising techniques like wavelet decomposition or principal component analysis (PCA) are available in commercial software packages. PCA can separate artifact components from true signal by decomposing the data into orthogonal modes; the first few principal components often contain the artifact pattern while later components capture the true material behavior. In MRI, navigator echoes or diaphragmatic tracking improve reconstruction of breathing-free images.

6. Operator Training and Protocol Standardization

Human factors play a major role in artifact generation. Develop a standardized operating procedure (SOP) that covers patient positioning, compression speed, breath-hold timing, and acquisition triggering. Use checklists and visual aids to ensure every operator follows the same steps. Periodically audit operators and compare their artifact scores using a metric like the signal-to-artifact ratio (SAR). SAR can be computed as the ratio of the mean signal power in a region of interest to the mean power in a region known to be artifact (e.g., a homogeneous phantom). In clinical settings, board-certified sonographers and radiologists should undergo competency assessments every six months, with refresher training if SAR falls below 10 dB.

For industrial labs, ensure technicians follow ASTM or ISO standards (e.g., ASTM D695 for compression testing of plastics). Document any deviations and their effect on artifact levels. A training module that includes side-by-side examples of artifact-free and artifact-laden data helps operators recognize and avoid problematic techniques. Implementing a “two-person verification” for critical tests (operator plus supervisor) further minimizes human error.

Advanced Techniques for Artifact Mitigation

Respiratory and Cardiac Gating

Gating is a technique where data acquisition is synchronized with the subject’s physiological cycles. Respiratory gating uses a belt or camera to detect the expiratory plateau and only collects data during that stable phase. Similarly, cardiac gating can reduce pulsation artifacts from large vessels adjacent to the compression site. In MRI, gating can be combined with triggering to acquire a single k-space line per heartbeat, yielding motion-free images. For ultrasound, some systems offer ECG-based gating that acquires frames during diastole when cardiac motion is minimal. Recent advances in real-time gating using machine learning can predict optimal acquisition windows from previous cycles, improving efficiency.

Real-Time Motion Tracking and Feedback

Modern systems incorporate optical or ultrasonic trackers to monitor the position of the compression plate relative to the subject. If motion exceeds a threshold, the system automatically pauses acquisition and re-adjusts the compression. Some research setups use a stereo camera pair to track skin markers and feed the displacement vector into a motion-correction algorithm. In industrial testing, laser displacement sensors can detect platen tilt or drift and correct the force reading in real time. These feedback loops are especially useful when testing large specimens or when environmental vibrations are unavoidable.

Adaptive Filtering and Averaging

Adaptive filters (e.g., least mean squares, recursive least squares) can learn the artifact pattern and cancel it without prior knowledge of the frequency. This is especially useful when breathing rate varies. The filter updates its coefficients based on the error between the measured signal and a reference (e.g., from a respiratory belt). Another powerful method is ensemble averaging: repeating the compression multiple times and averaging the results. This improves SNR by the square root of the number of repetitions while canceling random phase artifacts. However, it requires patient compliance and consistent compression conditions. A typical protocol might acquire 10–15 compressions with 5-second breath holds between each, then average the strain maps.

Machine Learning for Artifact Detection and Correction

Deep learning models trained on large datasets can automatically identify and correct pumping and breathing artifacts. For example, convolutional neural networks (CNNs) can segment artifact-affected regions in ultrasound strain images and replace them with interpolated values. Long short-term memory (LSTM) networks can model the temporal dynamics of breathing and predict the artifact component, enabling real-time subtraction. These methods are particularly beneficial in 3D elastography where manual correction is impractically time-consuming. While still a research area, several commercial vendors have started integrating AI-based motion correction in their products.

Practical Checklist for Minimizing Artifacts

  • Before each session: Calibrate load cell and imaging system; verify mechanical integrity; check for vibrations in the environment (e.g., air conditioning, pumps). Perform a quick test on a phantom to establish baseline artifact level.
  • Patient/subject preparation: Explain the procedure; use stabilization aids (foam pads, vacuum cushions, straps); have subject empty bladder and wear loose clothing. For sensitive regions, apply coupling gel and ensure good acoustic contact.
  • Breath-hold: Instruct for normal expiration hold; time acquisition during hold; allow rest periods. Use a respiratory belt to confirm plateau. For uncooperative subjects, use gating.
  • Compression settings: Use moderate speed (e.g., 1–2 mm/s), ramp-up profile; set frame rate to match or exceed compression frequency; enable force feedback. Preload with 0.5–1 N.
  • During acquisition: Monitor real-time display for rhythmic fluctuations or drift; abort and re-acquire if artifacts exceed threshold (e.g., >10% variation in strain). Watch for sudden shifts that indicate patient movement.
  • Post-processing: Apply bandstop or adaptive filtering; register images; reject frames with excessive motion (using a quality index like normalized cross-correlation). If available, use wavelet denoising or PCA.
  • Documentation: Log artifact type, severity, and corrective action taken. Include screenshot of artifact-affected frames. Track SAR over time to identify trends.

Conclusion

Pumping and breathing artifacts are not inevitable in live compression. By understanding their physical origins, implementing robust preparation and stabilization methods, optimizing compression parameters, maintaining equipment, and employing advanced signal processing, practitioners can achieve clean, reliable data. Consistent use of breath-hold techniques, gating, and operator training further reduces variability. The investment in artifact prevention pays dividends in diagnostic confidence, regulatory compliance, and product quality. As technology evolves, real-time adaptive and AI-based corrections will increasingly automate these steps, but the fundamental practices described here remain indispensable for any laboratory or clinic performing live compression measurements.

For further reading, consult the Radiopaedia article on breathing artifacts, the NIST Physical Measurement Laboratory for calibration standards, and the 2013 review on shear wave elastography artifacts for a deeper dive into medical applications. For industrial practitioners, ASTM D695 and ISO 18437 provide relevant standards. Implement these best practices, and live compression will become a robust, artifact-free tool in your workflow.