Integrating Multispectral Hemodynamic Imaging for Bulk Tissue Oxygenation Analysis

Megan Y Y Tran, Robert Amelard, Alexander WongJournal of Computational Vision and Imaging Systems 2017 · DOI

Abstract Tissue perfusion and oxygenation are critical predictors of patient outcomes, yet current non-invasive monitoring is largely limited to contact-based, single-site devices. This paper presents the integration of a multispectral optical-electronic subsystem into a non-contact coded hemodynamic imaging (CHI) device. By acquiring images under three distinct illuminants (740nm, 850nm, and 940nm), the system enables spatial mapping of tissue oxygenation. Validation experiments demonstrated high optical stability over 10 minutes ($σ_{max} = 0.407$ intensity units). A cuff occlusion experiment on the thenar eminence confirmed the system's ability to track relative changes in oxy- and deoxyhemoglobin concentrations consistent with known physiological responses.

Introduction

Monitoring tissue perfusion and oxygenation is vital in clinical settings such as surgical recovery, trauma care, and neonatal monitoring. Traditional tools, like pulse oximeters, are typically contact-based and provide only single-site data. This physical contact can be problematic for patients with fragile skin or complex wounds. While non-contact Coded Hemodynamic Imaging (CHI) has emerged as a solution for mapping blood flow, it has historically lacked the multispectral capabilities needed to distinguish between oxygenated (oxyhemoglobin) and deoxygenated (deoxyhemoglobin) blood. This research bridges that gap by co-integrating a multispectral subsystem into the CHI framework.

Methods

The researchers developed a "plug-and-play" multispectral device by integrating three core subsystems:

  1. Optical Subsystem: Three high-powered LEDs (740nm, 850nm, and 940nm) were selected to span the isosbestic point of hemoglobin (805nm). To ensure uniform illumination across the target tissue, a "fly’s-eye condenser" lens array was utilized to homogenize the light output.
  2. Electronic Subsystem: A constant-current driver and an Arduino microcontroller were used to manage the LEDs. The system employs time-multiplexing, cycling through wavelengths rapidly so the camera captures sequential frames that are later reconstructed into multispectral datasets.
  3. Computational Processing: The Beer-Lambert law was applied to the reflected light data to calculate the relative concentrations of oxy- and deoxyhemoglobin.
Multispectral system hardware components and assembly

Results and Discussion

The system's performance was evaluated through stability testing and a physiological validation experiment.

  • Hardware Stability: In optical imaging, hardware-induced fluctuations can mask subtle physiological signals. The system achieved remarkable temporal stability, with local fluctuations ($σ$) staying under 0.5 units on an 8-bit scale. Over a 10-minute period, the maximum overall intensity drift was limited to 3.1 units, largely attributed to thermal management via custom heat sinks.
Temporal stability of the three LED illuminants over 10 minutes
  • Physiological Validation: During a 45-second cuff occlusion test on the thenar eminence, the system detected a 5.3% decrease in oxyhemoglobin and a 20.1% increase in deoxyhemoglobin. Upon release of the cuff, the system successfully captured "reactive hyperemia"—a rapid spike in blood flow as the tissue compensates for the temporary oxygen debt. These results align with data from traditional contact-based clinical sensors, validating the non-contact approach.

Conclusion

The integration of multispectral capabilities into CHI allows for non-contact, wide-area mapping of tissue oxygenation. While a minor long-term thermal drift was observed, the system provides a robust platform for clinical monitoring. Future work will focus on "sensor fusion," combining this oxygen mapping with spatial pulsatility data to create a comprehensive, non-contact diagnostic tool for real-time hemodynamic assessment.

Summary updated

Abstract

Tissue perfusion and oxygenation are important factors in predicting patient outcomes, but current non-invasive devices for this type of measurement are limited to contact-based single-site monitoring. We present the co-integration of a multispectral optical-electronic subsystem into an existing non-contact coded hemodynamic imaging (CHI) device to enable image acquisition under different illuminants for spatial tissue oxygenation. Stability of the optical output for three illuminants over 10 mins was validated by the imaging system, with σmax=0.407 intensity units, reflecting stability in local fluctuations, and a maximal overall change of 3.1 units. Bulk tissue oxygenation measurement of the thenar eminence during a cuff occlusion experiment revealed relative changes in absorbance due to oxy- and deoxyhemoglobin consistent with concurrent physiological changes in chromophore concentration as described in a previous study.
BibTeX
@article{tran2017integrating,
  author = {Megan Y Y Tran and Robert Amelard and Alexander Wong},
  title = {{Integrating Multispectral Hemodynamic Imaging for Bulk Tissue Oxygenation Analysis}},
  journal = {Journal of Computational Vision and Imaging Systems},
  year = {2017},
  doi = {10.15353/vsnl.v3i1.163},
}
Figure