Continuous cuffless arterial blood pressure measurement based on PPG quality assessment

Lin Xu, Kunpeng GaoInt. J. Comput. Biol. Drug des 2015 · DOI

TL;DR

This paper presents a novel method for continuous, cuffless blood pressure monitoring by analyzing the quality of photoplethysmography (PPG) signals. By assessing PPG signal integrity, the system can more accurately estimate blood pressure using pulse transit time (PTT), outperforming traditional methods that rely on fixed feature points, especially in the presence of motion artifacts.

What Problem Does This Paper Address?

Traditional blood pressure measurement relies on inflatable cuffs, which are uncomfortable and impractical for continuous, long-term monitoring. Cuffless methods offer a more convenient alternative, often leveraging the relationship between blood pressure and pulse transit time (PTT). PTT is the time it takes for a pulse wave to travel between two points in the arterial system. While the R-wave of an ECG is a reliable marker for the start of a pulse, identifying the corresponding arrival point in the PPG signal (e.g., peak or valley) can be challenging due to signal distortions caused by factors like body motion. This makes accurate PTT calculation and, consequently, reliable continuous blood pressure estimation difficult.

Methods

The core of this paper's contribution lies in its PPG quality assessment mechanism and its integration into a data fusion approach.

  • PPG Quality Assessment: The authors propose a method to quantify the quality of PPG signals. This involves two main hypotheses:
  • Signal Consistency: The PPG signal should closely match its polynomial curve fit. The root mean square error (rMSE) of this fit is used to assess consistency.
  • ABP Consistency: The blood pressure calculated using different PPG feature points (peak vs. valley) should be consistent. Significant disagreement suggests signal distortion. Based on these hypotheses, they define quality indices ($\eta_1$ and $\eta_2$) that reflect the reliability of specific PPG segments.

  • Data Fusion with Kalman Filter: To improve accuracy, the paper employs a Kalman filter for data fusion. This filter combines information from multiple PPG feature points (specifically, PTT derived from the ECG R-wave to PPG peak, and PTT derived from the ECG R-wave to PPG valley). The Kalman filter's gain is dynamically adjusted based on the calculated PPG quality indices. When PPG quality is poor ($\eta < 0.5$), the data is deemed unreliable and discarded, preventing inaccurate blood pressure estimations.

  • ABP Calculation: The relationship between ABP and PTT is modeled as $ABP = a \ln PTT + b$, where $a$ and $b$ are patient-specific parameters determined during an initial calibration phase using a conventional sphygmomanometer and ECG/PPG sensors. Once $a$ and $b$ are known, ABP can be continuously estimated from PTT.

Datasets

The authors validate their method using the Multiparameter Intelligent Monitoring in Intensive Care (MIMIC) II database. This database contains comprehensive physiological data from patients. For their study, they selected 72 sets of data and focused on 10 sets for detailed processing. The blood pressure values from the MIMIC database are treated as the "true" reference values for comparison.

Key Insights and Ideas

  • PPG Quality is Paramount: The paper strongly emphasizes that the accuracy of cuffless blood pressure estimation is critically dependent on the quality of the PPG signal. Poor quality signals, often due to motion artifacts, can lead to significant errors.
  • Dynamic Feature Point Selection: Instead of relying on a single, fixed feature point (like the peak or valley) in the PPG signal, the proposed method intelligently assesses the quality of different segments and uses the most reliable information.
  • Kalman Filter for Robustness: The Kalman filter is crucial for integrating information from multiple PTT measurements and adapting to varying PPG signal quality. By weighting measurements based on quality, it effectively filters out noise and unreliable data.
  • Motion Artifact Mitigation: The quality assessment mechanism is particularly effective at identifying and mitigating the impact of motion artifacts, a common challenge in wearable sensing.

Key Results

The paper demonstrates that their proposed method, which fuses PTT data with PPG quality assessment using a Kalman filter, achieves more accurate blood pressure estimations compared to methods using single, fixed PPG feature points.

  • Reduced rMSE: Table 1 shows that the average root mean square error (rMSE) for the data fusion method with PPG quality assessment is 2.54, which is lower than estimation with PTT1 (3.00) and PTT2 (2.96).
  • Visual Consistency: Figure 2 and Figure 3 visually confirm the strong agreement between the proposed method's ABP estimates and the reference sphygmomanometer values, even during periods of potential signal disturbance. Figure 3, in particular, highlights how the data fusion method significantly reduces errors observed in methods relying on single PTT definitions.

Limitations and Future Work

  • Limitations:
    • The paper acknowledges that the method is verified on the MIMIC database, which consists of intensive care unit patients. The generalizability to a broader, ambulatory population with different physiological conditions might require further investigation.
  • The initial calibration step still requires a conventional sphygmomanometer to determine patient-specific parameters $a$ and $b$.
  • The paper mentions that when PPG quality is poor ($\eta < 0.5$), the data is abandoned. While this prevents errors, it means that continuous monitoring might have intermittent gaps in data.
  • Future Work:
    • The authors suggest that further research could explore more advanced methods for PPG quality assessment and data fusion to potentially achieve even higher accuracy and reduce data gaps.
  • Investigating methods for reducing or eliminating the need for initial calibration would be a significant advancement.

Summary updated

BibTeX
@article{xu2015continuous,
  author = {Lin Xu and Kunpeng Gao},
  title = {{Continuous cuffless arterial blood pressure measurement based on PPG quality assessment}},
  journal = {Int. J. Comput. Biol. Drug des},
  year = {2015},
  doi = {10.1504/ijcbdd.2015.071170},
}
Figure