Quantitative real-time pulse oximetry with ultrafast frequency-domain diffuse optics and deep neural network processing

Ninety years of pulse oximetry: history, current status, and outlook

Significance

This year, 2024, marks the 50th anniversary of the invention of pulse oximetry (PO), which was first presented by Takuo Aoyagi, an engineer from the Nihon Kohden Company, at the 13th Conference of the Japanese Society of Medical Electronics and Biological Engineering in Osaka in 1974. His discovery and the development of PO for the non-invasive measurement of peripheral arterial oxygenation represents one of the most significant chapters in the history of medical technology. It resulted from research and development efforts conducted by biochemists, engineers, physicists, physiologists, and physicians since the 1930s.

Aim

The objective of this work was to provide a narrative review of the history, current status, and future prospects of PO.

Approach

A comprehensive review of the literature on oximetry and PO was conducted.

Results and Conclusions

Our historical review examines the development of oximetry in general and PO in particular, tracing the key stages of a long and fascinating story that has unfolded from the first half of the twentieth century to the present day-an exciting journey in which serendipity has intersected with the hard work of key pioneers. This work has been made possible by the contributions of numerous key pioneers, including Kurt Kramer, Karl Matthes, Glenn Millikan, Evgenii M. Kreps, Earl H. Wood, Robert F. Show, Scott A. Wilber, William New, and, above all, Takuo Aoyagi. PO has become an integral part of modern medical care and has proven to be an important tool for physiological monitoring. The COVID-19 pandemic not only highlighted the clinical utility of PO but also revealed some of the problems with the technology. Current research in biomedical optics should address these issues to make the technology even more reliable and accurate. We discuss the necessary innovations in PO and present our thoughts on what the next generation of PO might look like.

Tutorial on methods for estimation of optical absorption and scattering properties of tissue

Significance

The estimation of tissue optical properties using diffuse optics has found a range of applications in disease detection, therapy monitoring, and general health care. Biomarkers derived from the estimated optical absorption and scattering coefficients can reflect the underlying progression of many biological processes in tissues.

Aim

Complex light-tissue interactions make it challenging to disentangle the absorption and scattering coefficients, so dedicated measurement systems are required. We aim to help readers understand the measurement principles and practical considerations needed when choosing between different estimation methods based on diffuse optics.

Approach

The estimation methods can be categorized as: steady state, time domain, time frequency domain (FD), spatial domain, and spatial FD. The experimental measurements are coupled with models of light-tissue interactions, which enable inverse solutions for the absorption and scattering coefficients from the measured tissue reflectance and/or transmittance.

Results

The estimation of tissue optical properties has been applied to characterize a variety of ex vivo and in vivo tissues, as well as tissue-mimicking phantoms. Choosing a specific estimation method for a certain application has to trade-off its advantages and limitations.

Conclusion

Optical absorption and scattering property estimation is an increasingly important and accessible approach for medical diagnosis and health monitoring.

Compressed single-shot 3D photoacoustic imaging with a single-element transducer

Three-dimensional (3D) photoacoustic imaging (PAI) can provide rich information content and has gained increasingly more attention in various biomedical applications. However, current 3D PAI methods either involves pointwise scanning of the 3D volume using a single-element transducer, which can be time-consuming, or requires an array of transducers, which is known to be complex and expensive. By utilizing a 3D encoder and compressed sensing techniques, we develop a new imaging modality that is capable of single-shot 3D PAI using a single-element transducer. The proposed method is validated with phantom study, which demonstrates single-shot 3D imaging of different objects and 3D tracking of a moving object. After one-time calibration, while the system could perform single-shot 3D imaging for different objects, the calibration could remain effective over 7 days, which is highly beneficial for practical translation. Overall, the experimental results showcase the potential of this technique for both scientific research and clinical applications.

Review of recent advances in frequency-domain near-infrared spectroscopy technologies [Invited]

Over the past several decades, near-infrared spectroscopy (NIRS) has become a popular research and clinical tool for non-invasively measuring the oxygenation of biological tissues, with particular emphasis on applications to the human brain. In most cases, NIRS studies are performed using continuous-wave NIRS (CW-NIRS), which can only provide information on relative changes in chromophore concentrations, such as oxygenated and deoxygenated hemoglobin, as well as estimates of tissue oxygen saturation. Another type of NIRS known as frequency-domain NIRS (FD-NIRS) has significant advantages: it can directly measure optical pathlength and thus quantify the scattering and absorption coefficients of sampled tissues and provide direct measurements of absolute chromophore concentrations. This review describes the current status of FD-NIRS technologies, their performance, their advantages, and their limitations as compared to other NIRS methods. Significant landmarks of technological progress include the development of both benchtop and portable/wearable FD-NIRS technologies, sensitive front-end photonic components, and high-frequency phase measurements. Clinical applications of FD-NIRS technologies are discussed to provide context on current applications and needed areas of improvement. The review concludes by providing a roadmap toward the next generation of fully wearable, low-cost FD-NIRS systems.

Single-shot photoacoustic imaging with single-element transducer through a spatiotemporal encoder

Significance

Current photoacoustic (PA) imaging modalities typically require either serial detection with a single-element transducer or parallel detections with an ultrasonic array, indicating a dilemma between system cost and imaging throughput. PA topography through ergodic relay (PATER) was recently developed to address this bottleneck. However, PATER requires object-specific calibration due to varied boundary condition and must be recalibrated through pointwise scanning for each object before measurements, which is time-consuming and severely limits practical application.

Aim

We aim to develop a new single-shot PA imaging technique that only requires a one-time calibration for imaging different objects using a single-element transducer.

Approach

We develop an imaging method, PA imaging through a spatiotemporal encoder (PAISE), to address the above issue. The spatial information is effectively coded into unique temporal features by the spatiotemporal encoder, which allows for compressive image reconstruction. An ultrasonic waveguide is proposed as a critical element to guide the PA waves from the object into the prism, which effectively accounts for the varied boundary condition of different objects. We further add irregular-shaped edges on the prism to introduce randomized internal reflections and further facilitate the scrambling of acoustic waves.

Results

The proposed technique is validated through comprehensive numerical simulations and experiments, and it is demonstrated that PAISE can successfully overcome the changed boundary condition and can image different samples given a single calibration.

Conclusions

The proposed PAISE technique is capable of single-shot widefield PA imaging with a single-element transducer and does not require sample-specific calibration, which successfully overcomes the major limitation of previous PATER technology.

A scalable, multi-wavelength, broad bandwidth frequency-domain near-infrared spectroscopy platform for real-time quantitative tissue optical imaging

Frequency-domain near-infrared spectroscopy (FD-NIRS) provides quantitative noninvasive measurements of tissue optical absorption and scattering, as well as a safe and accurate method for characterizing tissue composition and metabolism. However, the poor scalability and high complexity of most FD-NIRS systems assembled to date have contributed to its limited clinical impact. To address these shortcomings, we present a scalable, digital-based FD-NIRS platform capable of measuring optical properties and tissue chromophore concentrations in real-time. The system provides single-channel FD-NIRS amplitude/phase, optical property, and chromophore data at a maximum display rate of 36.6 kHz, 17.9 kHz, and 10.2 kHz, respectively, and can be scaled to multiple channels as well as integrated into a handheld format. The entire system is enabled by several innovations including an ultra-high-speed k-nearest neighbor lookup table method (maximum of 250,000 inversions/s for a large 2500x700 table of absorption and reduced scattering coefficients), embedded FPGA and CPU high-speed co-processing, and high-speed data transfer (due to on-board processing). We show that our 6-wavelength, broad modulation bandwidth (1-400 MHz) system can be used to perform 2D high-density spatial mapping of optical properties and high speed quantification of hemodynamics.

Imaging sub-diffuse optical properties of cancerous and normal skin tissue using machine learning-aided spatial frequency domain imaging

SIGNIFICANCE

Sub-diffuse optical properties may serve as useful cancer biomarkers, and wide-field heatmaps of these properties could aid physicians in identifying cancerous tissue. Sub-diffuse spatial frequency domain imaging (sd-SFDI) can reveal such wide-field maps, but the current time cost of experimentally validated methods for rendering these heatmaps precludes this technology from potential real-time applications.

AIM

Our study renders heatmaps of sub-diffuse optical properties from experimental sd-SFDI images in real time and reports these properties for cancerous and normal skin tissue subtypes.

APPROACH

A phase function sampling method was used to simulate sd-SFDI spectra over a wide range of optical properties. A machine learning model trained on these simulations and tested on tissue phantoms was used to render sub-diffuse optical property heatmaps from sd-SFDI images of cancerous and normal skin tissue.

RESULTS

The model accurately rendered heatmaps from experimental sd-SFDI images in real time. In addition, heatmaps of a small number of tissue samples are presented to inform hypotheses on sub-diffuse optical property differences across skin tissue subtypes.

CONCLUSION

These results bring the overall process of sd-SFDI a fundamental step closer to real-time speeds and set a foundation for future real-time medical applications of sd-SFDI such as image guided surgery.

The Challenges and Pitfalls of Detecting Sleep Hypopnea Using a Wearable Optical Sensor: Comparative Study

Background

Obstructive sleep apnea (OSA) is the most prevalent respiratory sleep disorder occurring in 9% to 38% of the general population. About 90% of patients with suspected OSA remain undiagnosed due to the lack of sleep laboratories or specialists and the high cost of gold-standard in-lab polysomnography diagnosis, leading to a decreased quality of life and increased health care burden in cardio- and cerebrovascular diseases. Wearable sleep trackers like smartwatches and armbands are booming, creating a hope for cost-efficient at-home OSA diagnosis and assessment of treatment (eg, continuous positive airway pressure [CPAP] therapy) effectiveness. However, such wearables are currently still not available and cannot be used to detect sleep hypopnea. Sleep hypopnea is defined by ≥30% drop in breathing and an at least 3% drop in peripheral capillary oxygen saturation (Spo₂) measured at the fingertip. Whether the conventional measures of oxygen desaturation (OD) at the fingertip and at the arm or wrist are identical is essentially unknown.

Objective

We aimed to compare event-by-event arm OD (arm_OD) with fingertip OD (finger_OD) in sleep hypopneas during both naïve sleep and CPAP therapy.

Methods

Thirty patients with OSA underwent an incremental, stepwise CPAP titration protocol during all-night in-lab video-polysomnography monitoring (ie, 1-h baseline sleep without CPAP followed by stepwise increments of 1 cmH₂O pressure per hour starting from 5 to 8 cmH₂O depending on the individual). Arm_OD of the left biceps muscle and finger_OD of the left index fingertip in sleep hypopneas were simultaneously measured by frequency-domain near-infrared spectroscopy and video-polysomnography photoplethysmography, respectively. Bland-Altman plots were used to illustrate the agreements between arm_OD and finger_OD during baseline sleep and under CPAP. We used t tests to determine whether these measurements significantly differed.

Results

In total, 534 obstructive apneas and 2185 hypopneas were recorded. Of the 2185 hypopneas, 668 (30.57%) were collected during baseline sleep and 1517 (69.43%), during CPAP sleep. The mean difference between finger_OD and arm_OD was 2.86% (95% CI 2.67%-3.06%, t₆₆₇=28.28; P<.001; 95% limits of agreement [LoA] –2.27%, 8.00%) during baseline sleep and 1.83% (95% CI 1.72%-1.94%, t₁₅₁₆=31.99; P<.001; 95% LoA –2.54%, 6.19%) during CPAP. Using the standard criterion of 3% saturation drop, arm_OD only recognized 16.32% (109/668) and 14.90% (226/1517) of hypopneas at baseline and during CPAP, respectively.

Conclusions

arm_OD is 2% to 3% lower than standard finger_OD in sleep hypopnea, probably because the measured arm_OD originates physiologically from arterioles, venules, and capillaries; thus, the venous blood adversely affects its value. Our findings demonstrate that the standard criterion of ≥3% OD drop at the arm or wrist is not suitable to define hypopnea because it could provide large false-negative results in diagnosing OSA and assessing CPAP treatment effectiveness.

Hemodynamics of the sternocleidomastoid measured with frequency domain near-infrared spectroscopy towards non-invasive monitoring during mechanical ventilation

Mechanical ventilation (MV) is used to assist spontaneous breathing in critically ill patients in the intensive care unit (ICU). MV is a cornerstone of critical care medicine but it is now known that inspiratory muscle dysfunction due to injury, disuse, and/or atrophy during MV plays a major role in outcomes for these patients. For example, prolonged MV is strongly correlated with dysfunction of the sternocleidomastoid (SCM), an accessory inspiratory muscle that has been linked to weaning failure from MV. Hemodynamic monitoring of the SCM may provide an important non-invasive and real-time means to monitor MV. In this work, we first conducted multi-layer Monte Carlo simulations to confirm the ability of near infrared light to detect changes in the oxygenation of the SCM over wide ranges of skin tones and adipose layer thicknesses. We then optimized a custom digital frequency domain near-infrared spectroscopy (FD-NIRS) system for continuous 10 Hz measurements of the SCM at 730 nm and 850 nm. A healthy volunteer study was conducted (N=10); subjects performed sets of isometric neck flexions of the SCM. Substantial changes in oxyhemoglobin + oxymyoglobin (oxy[Hb + Mb]), deoxyhemoglobin + deoxymyoglobin (deoxy[Hb + Mb]), and total hemoglobin + myoglobin (total[Hb + Mb]) were observed during sustained and intermittent isometric flexions. There were notable sex differences observed in the magnitude of hemodynamic changes (∼2x larger changes in males for oxy[Hb + Mb] and deoxy[Hb + Mb]). The magnitude of hemodynamic changes when taking into account µ_s' changes during flexions was ∼ 2-2.5x larger as compared to assuming constant scattering (CS), which is a common assumption used for continuous wave (CW) NIRS methods. This study suggests that FD-NIRS provides improved accuracy for hemodynamic monitoring of the SCM compared to CW-NIRS, and that FD-NIRS may provide value for SCM monitoring during MV.

Image reconstruction with the chaotic fiber laser in scattering media

The reconstruction of the size, position, optical properties, and structure of the object in scattering media was realized with a chaotic fiber laser. The light from the chaotic fiber laser was split into two parts. One part was used as the detection signal to detect the object, and the other was used as the reference signal; then, the two signals were cross correlated. The attenuation of light in scattering media was attributed to scattering and absorption. The theoretical model of the peak value of cross correlation of the chaotic signals as projection data were established by the attenuation law, and the filtered back-projection algorithms were used to realize the image reconstruction. The mean squared error, the normalized mean squared error, the peak signal-to-noise ratio, and the structural similarity index of the reconstructed image were analyzed. The results show that the high resolution of the reconstructed image benefits from the high signal-to-noise ratio with the chaotic fiber laser based on a delta-like cross-correlation function.

Real-Time Handheld Probe Tracking and Image Formation Using Digital Frequency-Domain Diffuse Optical Spectroscopy

OBJECTIVE

Frequency-domain diffuse optical spectroscopic imaging (FD-DOS) is a non-invasive method for measuring absolute concentrations of tissue chromophores such as oxy- and deoxy-hemoglobin in vivo. The utility of FD-DOS for clinical applications such as monitoring chemotherapy response in breast cancer has previously been demonstrated, but challenges for further clinical translation, such as slow acquisition speed and lack of user feedback, remain. Here, we propose a new high speed FD-DOS instrument that allows users to freely acquire measurements over the tissue surface, and is capable of rapidly imaging large volumes of tissue.

METHODS

We utilize 3D monocular probe tracking combined with custom digital FD-DOS hardware and a high-speed data processing pipeline for the instrument. Results are displayed during scanning over the surface of the sample using a probabilistic Monte Carlo light propagation model.

RESULTS

We show this instrument can measure absorption and scattering coefficients with an error of 7% and 1% respectively, with 0.7 mm positional accuracy. We demonstrate the equivalence of our visualization methodology with a standard interpolation approach, and demonstrate two proof-of-concept in vivo results showing superficial vasculature in the human forearm and surface contrast in a healthy human breast.

CONCLUSION

Our new FD-DOS system is able to compute chromophore concentrations in real-time (1.5 Hz) in vivo.

SIGNIFICANCE

This method has the potential to improve the quality of FD-DOS image scans while reducing measurement times for a variety of clinical applications.

Gradient-assisted focusing light through scattering media

Focusing light through a scattering medium is a longstanding challenge in biomedical optics, to which wavefront shaping is a powerful solution. The state-of-the-art feedback-based approach is the widely used genetic algorithm method. However, it can only achieve relatively low enhancement of the focus, and the genetic algorithm is known to be time-consuming. To tackle those issues, we propose a gradient-assisted strategy for wavefront shaping. The proposed method conducts optimization in the function distribution space. Specifically, when optimizing the parameters along each iteration, the consequent function distribution changes within a distance as measured by the Kullback-Leibler divergence. Taking advantage of the gradient information, the proposed method is over 60× faster to obtain the same peak-to-background ratio (PBR) level. Compared with the genetic algorithm that is able to optimize a number of 64×64 phase segments, the proposed gradient strategy is able to optimize 256×256 phase segments, and gives 20× higher focus enhancement as quantified by the PBR.

Modulation frequency selection and efficient look-up table inversion for frequency domain diffuse optical spectroscopy

SIGNIFICANCE

Frequency domain diffuse optical spectroscopy (FD-DOS) uses intensity modulated light to measure the absorption and reduced scattering coefficients of turbid media such as biological tissue. Some FD-DOS instruments utilize a single modulation frequency, whereas others use hundreds of frequencies. The effect of modulation frequency choice and measurement bandwidth on optical property (OP) extraction accuracy has not yet been fully characterized.

AIM

We aim to assess the effect of modulation frequency selection on OP extraction error and develop a high-speed look-up table (LUT) approach for OP estimation.

APPROACH

We first used noise-free simulations of light transport in homogeneous media to determine optimized iterative inversion model parameters and developed a new multi-frequency LUT method to increase the speed of inversion. We then used experimentally derived noise models for two FD-DOS instruments to generate realistic simulated data for a broad range of OPs and modulation frequencies to test OP extraction accuracy.

RESULTS

We found that repeated measurements at a single low-frequency (110 MHz) yielded essentially identical OP errors as a broadband frequency sweep (35 evenly spaced frequencies between 50 and 253 MHz) for these noise models. The inclusion of modulation frequencies >300  MHz diminished overall performance for one of the instruments. Additionally, we developed a LUT inversion algorithm capable of increasing inversion speeds by up to 6  ×  , with 1000  inversions  /  s and ∼1  %   error when a single modulation frequency was used.

CONCLUSION

These results suggest that simpler single-frequency systems are likely sufficient for many applications and pave the way for a new generation of simpler digital FD-DOS systems capable of rapid, large-volume measurements with real-time feedback.

Direct mapping from diffuse reflectance to chromophore concentrations in multi-fx spatial frequency domain imaging (SFDI) with a deep residual network (DRN)

Spatial frequency domain imaging (SFDI) is an emerging technology that enables label-free, non-contact, and wide-field mapping of tissue chromophore contents, such as oxy- and deoxy-hemoglobin concentrations. It has been shown that the use of more than two spatial frequencies (multi-fx ) can vastly improve measurement accuracy and reduce chromophore estimation uncertainties, but real-time multi-fx SFDI for chromophore monitoring has been limited in practice due to the slow speed of available chromophore inversion algorithms. Existing inversion algorithms have to first convert the multi-fx diffuse reflectance to optical absorptions, and then solve a set of linear equations to estimate chromophore concentrations. In this work, we present a deep learning framework, noted as a deep residual network (DRN), that is able to directly map from diffuse reflectance to chromophore concentrations. The proposed DRN is over 10x faster than the state-of-the-art method for chromophore inversion and enables 25x improvement on the frame rate for in vivo real-time oxygenation mapping. The proposed deep learning model will help enable real-time and highly accurate chromophore monitoring with multi-fx SFDI.

Recent Developments in Instrumentation of Functional Near-Infrared Spectroscopy Systems

In the last three decades, the development and steady improvement of various optical technologies at the near-infrared region of the electromagnetic spectrum has inspired a large number of scientists around the world to design and develop functional near-infrared spectroscopy (fNIRS) systems for various medical applications. This has been driven further by the availability of new sources and detectors that support very compact and wearable system designs. In this article, we review fNIRS systems from the instrumentation point of view, discussing the associated challenges and state-of-the-art approaches. In the beginning, the fundamentals of fNIRS systems as well as light-tissue interaction at NIR are briefly introduced. After that, we present the basics of NIR systems instrumentation. Next, the recent development of continuous-wave, frequency-domain, and time-domain fNIRS systems are discussed. Finally, we provide a summary of these three modalities and an outlook into the future of fNIRS technology.

Hemodynamic responses during standing and sitting activities: a study toward fNIRS-BCI

In this paper, we utilized functional near-infrared spectroscopy (fNIRS) technology to examine the hemodynamic responses in the motor cortex for two conditions, namely standing and sitting tasks. Nine subjects performed five trials of standing and sitting (SAS) tasks with both real movements and imagery thinking of SAS. A group level of statistical parametric mapping (SPM) analysis during these tasks showed bilateral activation of oxy-hemoglobin for both real movements and imagery experiments. Interestingly, the SPM analysis clearly revealed that the sitting tasks induced a higher oxy-hemoglobin level activation compared to the standing task. Remarkably, this finding is persistent across the 22 measured channels at the individual and group levels for both experiments. Furthermore, six features were extracted from pre-processed HbO signals and the performance of four different classifiers was examined in order to test the viability of using SAS tasks in future fNIRS-brain-computer interface (fNIRS-BCI) systems. In particular, two features-combination tests revealed that the signal slope with signal variance represents one of the three best two-combined features for its consistency in providing high accuracy results for both real and imagery experiments. This study shows the potential of implementing such tasks into the fNIRS-BCI system. In the future, the results of this work could pave the way towards the application of fNIRS-BCI in lower limb rehabilitation.

Frequency-Domain Techniques for Cerebral and Functional Near-Infrared Spectroscopy

This article reviews the basic principles of frequency-domain near-infrared spectroscopy (FD-NIRS), which relies on intensity-modulated light sources and phase-sensitive optical detection, and its non-invasive applications to the brain. The simpler instrumentation and more straightforward data analysis of continuous-wave NIRS (CW-NIRS) accounts for the fact that almost all the current commercial instruments for cerebral NIRS have embraced the CW technique. However, FD-NIRS provides data with richer information content, which complements or exceeds the capabilities of CW-NIRS. One example is the ability of FD-NIRS to measure the absolute optical properties (absorption and reduced scattering coefficients) of tissue, and thus the absolute concentrations of oxyhemoglobin and deoxyhemoglobin in brain tissue. This article reviews the measured values of such optical properties and hemoglobin concentrations reported in the literature for animal models and for the human brain in newborns, infants, children, and adults. We also review the application of FD-NIRS to functional brain studies that focused on slower hemodynamic responses to brain activity (time scale of seconds) and faster optical signals that have been linked to neuronal activation (time scale of 100 ms). Another example of the power of FD-NIRS data is related to the different regions of sensitivity featured by intensity and phase data. We report recent developments that take advantage of this feature to maximize the sensitivity of non-invasive optical signals to brain tissue relative to more superficial extracerebral tissue (scalp, skull, etc.). We contend that this latter capability is a highly appealing quality of FD-NIRS, which complements absolute optical measurements and may result in significant advances in the field of non-invasive optical sensing of the brain.