Abstract
Introduction: In studies of pulse wave analysis, single-channel sensors only adopt single temporal pulse signals without spatial information to show pulse-feeling patterns. Multi-channel arterial pulse signals, also named as three-dimensional pulse images (3DPIs), provide the spatial and temporal characteristics of radial pulse signals. When involving single or few-channel sensors, pressing offsets have substantial impacts on obtaining inaccurate physiological parameters like tidal peak ($P_2$). Methods: This study discovers the pressing offsets in multi-channel pulse signals and analyzes the relationship between the pressing offsets and time of $P_2$ ($T_2$) by qualifying the pressing offsets. First, we employ a data acquisition system to capture 3DPIs. Subsequently, the $errorT_2$ is developed to qualify the pressing offsets. Results: The outcomes display a central low and peripheral high pattern. Additionally, the $errorT_2$ increase as the distances from the artery increase, particularly at the radial ends of the blood flow direction. For every 1 mm increase in distances between sensing elements and center sensing elements, the $errorT_2$ in the radial direction escalates by 4.87%. When the distance is greater than 3.42 mm, the $errorT_2$ experiences a sudden increase. Discussion: The results show that increasing the sensor channels can overcome the pressing offsets in radial pulse signal acquisition.
Introduction
Pulse Wave Analysis (PWA) serves as a vital non-invasive tool in cardiovascular diagnostics, enabling the detection of conditions such as arterial stiffness and endothelial dysfunction. A key parameter in this analysis is the tidal peak ($P_2$), which represents the second wave peak within a cardiac cycle. However, the accuracy of PWA is often compromised by "pressing offsets"—measurement errors that occur when a sensor is not perfectly aligned with the radial artery.
Traditional single-channel sensors capture only temporal data, lacking the spatial context necessary to identify if a signal is distorted by misalignment. This study explores how multi-channel tactile sensors can provide three-dimensional pulse images (3DPIs) to characterize and mitigate these pressing offsets.
Methodology
To address the limitations of single-point sensing, the researchers utilized a multi-channel data acquisition system.
Hardware and Data Collection
The system featured a Pressure Profile Systems (PPS) tactile sensor with a $5 \times 5$ grid of sensing elements, providing 24 active channels. This array allows for the simultaneous capture of pulse signals across a spatial area, creating a 3DPI that reflects both temporal and spatial characteristics of the radial pulse.
Signal Processing and Quantification
Raw data was processed using a novel de-noising algorithm: Cross Channel Dynamic Weighting Principal Component Analysis (cc-DWRPCA). This method leverages correlations between adjacent channels to effectively remove motion artifacts and noise.
To quantify the impact of pressing offsets, the researchers developed the $errorT_2$ metric. This measures the time discrepancy of the $P_2$ peak between a peripheral sensing element and the central element positioned directly over the artery. By aggregating these errors across 52 subjects (623 samples), an error Stacked Surface (error$T_2$SS) was generated to map spatial inaccuracy.
Key Results and Insights
The study revealed several critical findings regarding how spatial misalignment affects pulse signal quality:
- Spatial Error Gradient: The $errorT_2$ follows a "central low, peripheral high" pattern. The timing of the tidal peak is most stable directly over the artery and degrades predictably as the sensor moves toward the periphery.
- Directional Sensitivity: Measurement error is significantly more sensitive to offsets in the radial direction (along the blood flow) compared to the axial direction. Specifically, for every 1 mm increase in distance from the center in the radial direction, the $errorT_2$ increases by approximately 4.87%.
- The 3.42 mm Threshold: A critical discovery was the identification of a quality "cliff." When the pressing offset exceeds 3.42 mm, the $errorT_2$ undergoes a sudden, sharp increase, indicating that data beyond this point is highly unreliable for physiological parameter extraction.
Discussion and Conclusion
The findings demonstrate that increasing the number of sensor channels is a robust solution for the pressing offset problem. By providing spatial redundancy, multi-channel arrays ensure that even if the sensor is not perfectly placed, at least one channel will likely fall within the low-error zone (under the 3.42 mm threshold).
This spatial information allows for "self-correcting" diagnostics, where the system can automatically identify the optimal sensing channel. Future work will focus on expanding the demographic diversity of the study and integrating these insights into real-time clinical feedback systems to improve the reliability of cardiovascular monitoring.
Summary updated