Photoplethysmography detection system based on vertical cavity surface emitting laser

Abstract

The invention discloses a photoelectric volume pulse wave detection system based on a vertical cavity surface emitting laser, which is characterized in that a multi-wavelength VCSEL (vertical cavity surface emitting laser) is used as a core light emitting unit and is matched with a signal processing module, an embedded control module and an AI (artificial intelligence) sign analysis module to construct an integrated detection framework. According to the system, stable lasing is realized by optimizing a VCSEL epitaxial structure and a preparation process; a multi-wavelength PPG signal is accurately acquired in cooperation with a low-noise amplification-multistage filtering-adjustable gain signal processing link; the embedded control module dynamically adjusts light source parameters, and the AI engine realizes real-time accurate calculation of health parameters through multi-dimensional feature extraction and integrated learning model and fusion of multi-wavelength signals. The system has the advantages of high integration level, low power consumption and strong anti-interference performance, the core unit is packaged compactly, the system adapts to the long endurance requirement of wearable equipment, the detection precision is obviously improved compared with the traditional LED equipment, and a high-cost-performance solution can be provided for the scenes of consumer electronic health management, primary medical screening and the like.

Description

Technical Field

[0001] This invention relates to the field of optical sensing technology. The system uses a vertical cavity surface-emitting laser (VCSEL) as the core optical emitting unit and integrates photoelectric detection, signal processing and vital sign analysis modules to realize real-time monitoring of photoplethysmography (PPG) signals, which is suitable for portable medical testing. Background Technology

[0002] With the rising global incidence of cardiovascular disease and the widespread use of wearable medical devices, non-invasive continuous vital sign monitoring (NVR) technology has become a research hotspot. Current mainstream health monitoring devices (such as smart bracelets and portable pulse oximeters) primarily use light-emitting diodes (LEDs) as their core optoelectronic components. These LEDs suffer from problems such as isotropic spontaneous emission, broad spectral linewidth, low power density, and unadjustable power, resulting in limited skin penetration and susceptibility to stray light interference. Particularly during exercise, the signal-to-noise ratio of PPG signals significantly decreases, affecting the accuracy of heart rate and blood oxygen calculations. Furthermore, most devices rely on simple time-domain algorithms to extract PPG signal features, exhibiting weak suppression of motion artifacts and baseline drift, and lacking real-time adaptive analysis capabilities. This prevents them from dynamically optimizing monitoring accuracy based on the user's physiological state, making it difficult to meet the precise detection needs of primary healthcare settings.

[0003] VCSELs, with their advantages of small size, circular output spot, single longitudinal mode lasing, low threshold current, wide operating temperature range, and adjustable power, have become an ideal light source to replace traditional LEDs. Meanwhile, the lightweight development of embedded AI technology has made local real-time signal analysis possible. However, there is currently no solution for building an integrated health monitoring system based on VCSELs and combined with an embedded AI engine, making it difficult to simultaneously meet the requirements of device miniaturization, low power consumption, and high monitoring accuracy.

[0004] To address this, the present invention proposes a VCSEL-based photoplethysmography (PPG) pulse wave detection system. By optimizing the VCSEL epitaxial structure and fabrication process, the system improves the performance of the light source and integrates photoelectric detection and signal processing modules as well as a vital sign analysis module. This solves the problems of low integration and poor anti-interference in existing technologies, providing a cost-effective solution for the portable medical field. Summary of the Invention

[0005] This invention discloses a VCSEL-based photoplethysmography (PPG) system. This system uses a VCSEL as the core light-emitting unit, combined with photoelectric detection, signal processing, and vital sign analysis modules. Through PPG wavelength analysis technology, it achieves real-time monitoring of health parameters such as heart rate and blood oxygen saturation. It boasts advantages such as high integration, low power consumption, and high monitoring accuracy, and is adaptable to wearable devices and primary healthcare instruments, providing a cost-effective solution for portable medical testing. The purpose of this invention is to overcome the contradiction between integration, power consumption, and monitoring accuracy in existing health monitoring devices by combining the high stability of the VCSEL light source with the low power consumption control of the embedded system, thus achieving a PPG system that is both portable, low-power, and provides real-time, accurate detection. To achieve the objectives of this invention, the technical solution adopted includes:

[0006] This invention first designs the structure and fabrication process of the core optical emitting device, VCSEL, to ensure it meets the performance requirements of the light source in vital sign monitoring scenarios. In terms of structural design, the VCSEL device, as... Figure 1 As shown, the structure includes a top electrode 1, an upper DBR2, an active region 3, a lower DBR4, and a bottom electrode 5 arranged sequentially from top to bottom. Both the upper DBR2 and lower DBR4 achieve high reflectivity through an alternating layer structure. A gradient upper / lower distributed Bragg mirror (DBR) is constructed based on transfer matrix theory, and the reflectivity of the upper and lower DBRs is... satisfy:

[0007] (1);

[0008] In the formula, It is the vacuum wavelength. It is the effective absorption coefficient. It is the difference in refractive index between the two layers of DBR material.

[0009] The interface barrier is reduced by gradually transitioning components, which reduces carrier transport loss and ensures that the laser is preferentially emitted from the top and maximizes the efficiency of intracavity optical feedback; the active region adopts a multi-quantum well structure.

[0010] The peak optical gain of this structure It can be expressed by the following formula:

[0011] (2);

[0012] In the formula, These are momentum matrix elements. It is the density of states. It is the net reversal factor.

[0013] By setting confinement layers on both sides of the active region and optimizing the band structure to efficiently confine charge carriers within the quantum well, radiative recombination efficiency is improved, ultimately achieving stable lasing. Using a VCSEL as a highly stable, low-power light source, the emitted light of a specific wavelength interacts with the pulsating blood volume in human tissues (such as fingertips or earlobes), converting this interaction into changes in light intensity that can be received by a photodetector, ultimately demodulating the pulse wave signal.

[0014] To achieve efficient acquisition and processing of multi-wavelength PPG signals, this invention constructs a modular signal processing architecture based on an embedded core. Through time-division control and dual-channel acquisition design, it adapts to the needs of multi-wavelength optical emission units and signal processing.

[0015] The architecture is centered on an embedded control module, which is responsible for the driving timing control, signal acquisition scheduling, and data interaction functions of the multi-wavelength optical emitting unit. The driving control of the optical emitting unit adopts a time-division strategy, switching the working state of different wavelength light sources according to a preset cycle to ensure interference-free acquisition of multi-wavelength signals. The signal acquisition link adopts a three-level architecture of "peripheral signal processing - high-precision conversion - data temporary storage and transmission" to form a complete signal processing flow.

[0016] The peripheral signal processing module adopts a generalized "low-noise amplification-multi-stage filtering" architecture to adapt to the weak photocurrent signals output by different wavelength optical detection units: the first stage converts the photocurrent into a voltage signal through a low-noise amplification unit, and simultaneously integrates a low-pass filter component to filter out high-frequency noise; the second stage suppresses baseline drift through coupling capacitors and is equipped with an adjustable gain unit to dynamically adjust the signal amplification factor according to the detection scenario to ensure that the signal strength is adapted to the subsequent conversion requirements.

[0017] The signal conversion module uses a high-precision analog-to-digital converter (ADC) unit, featuring low noise and high resolution. The sampling rate can be dynamically configured by the embedded control module, accurately capturing subtle fluctuations in multi-wavelength PPG signals. The converted digital signal is transmitted to the embedded control module via a standard communication protocol and temporarily stored in the on-chip memory. Simultaneously, it establishes bidirectional interaction with the host computer via a serial communication protocol, enabling real-time data upload and control command reception. Furthermore, the embedded control module can adaptively adjust the drive power of the optical emitting unit according to the monitoring scenario (static / dynamic), balancing power consumption and signal strength requirements.

[0018] To improve the accuracy and anti-interference capability of multi-wavelength PPG signal vital signs analysis, lightweight machine learning algorithms can be introduced to build a modular AI vital signs analysis module, realizing fully automated processing from signal preprocessing to vital signs parameter calculation.

[0019] The module first performs standardized preprocessing on the multi-wavelength PPG raw signal: power frequency interference is removed by an adaptive filtering component, motion artifacts are suppressed by wavelet transform technology, and baseline drift is reduced by sliding window smoothing, ensuring that the signal quality meets the requirements for feature extraction.

[0020] The feature extraction process employs a multi-dimensional modular design, extracting core features for each wavelength PPG signal: the time-domain feature module captures key morphological parameters of the pulse wave through energy operators, reflecting the periodicity of heart rate and pulse intensity; the frequency-domain feature module extracts the dominant frequency and energy proportion of different frequency bands through power spectral density analysis, correlating vascular elasticity and microcirculation characteristics; and the nonlinear feature module assesses the cardiovascular system's dynamic characteristics through long-range correlation analysis. The feature extraction process for each wavelength signal operates independently, ultimately forming a multi-dimensional dual-path feature set.

[0021] The core of the vital sign analysis employs a lightweight ensemble learning model, trained on a large-scale, multi-scenario PPG dataset. The model parameters are optimized to fit the storage and computing resources of embedded platforms, eliminating reliance on cloud computing. Through an ensemble inference mechanism using multiple decision trees, the model integrates feature information from multi-wavelength PPG signals to achieve real-time calculation of vital sign parameters such as heart rate and blood oxygen saturation. It also possesses dynamic error correction capabilities to ensure analysis accuracy across different scenarios. The overall system is as follows: Figure 2 As shown.

[0022] Compared with the prior art, the beneficial effects of the present invention are:

[0023] 1. Higher detection accuracy: Replacing traditional LEDs with VCSELs as the core light source, the VCSEL's single-mode lasing and high power density significantly improve the stability of the light signal. Adjustable power allows for adjustable skin penetration depth, enabling the acquisition of multi-dimensional signals while reducing environmental stray light interference. Combined with multi-dimensional signal processing and intelligent analysis algorithms, it effectively suppresses motion artifacts and power frequency interference, significantly improving the accuracy of monitoring vital signs such as heart rate and blood oxygen saturation, meeting the needs for higher-precision health monitoring.

[0024] 2. Superior integration and power consumption: Adopting an integrated architecture, the VCSEL light emission, photoelectric detection, signal processing and analysis functions are highly integrated, significantly reducing the system size and making it more suitable for miniaturized application scenarios such as wearable devices; The low threshold current characteristics of VCSEL combined with the low power consumption design of embedded systems significantly reduce the overall system energy consumption, extend the device's battery life, and solve the pain points of large size and high power consumption of traditional discrete device solutions.

[0025] 3. Supports dynamic parameter adjustment: The VCSEL power can be flexibly adjusted according to the user's physiological state, monitoring site and usage scenario, avoiding the poor signal adaptability problem caused by the "fixed parameters" of traditional equipment; the lightweight analysis algorithm can be deployed locally without relying on the cloud, and at the same time, it reserves function expansion interfaces to facilitate the addition of detection dimensions in the future. Attached Figure Description

[0026] Figure 1 This is a structural diagram of a VCSEL device.

[0027] Figure 2 This is a block diagram of the detection system.

[0028] Figure 3 The measured PPG signal. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Example 1: Design and fabrication of a VCSEL laser;

[0031] This embodiment details the fabrication method of the core light source of the present invention—the red-light vertical cavity surface-emitting laser (VCSEL).

[0032] First, epitaxial structure growth was performed. The following structures were sequentially grown on a semi-insulating GaAs substrate using metal-organic chemical vapor deposition (MOCVD):

[0033] 1. 56.5 pairs / The gradient n-DBR's reflectivity design follows the principle described in formula (1), with a theoretical reflectivity >99.98%.

[0034] 2. 50nm thick n-type Lower constraint layer.

[0035] 3, 3 pairs / The active region of the multi-quantum well has a well thickness of 8 nm and a barrier thickness of 12 nm. Its gain characteristics are characterized by formula (2).

[0036] 4. 50nm thick p-type Upper constraint layer.

[0037] 5. 30nm thick Oxide layer.

[0038] 6, 30 pairs / Gradient-type p-DBR, theoretical reflectivity >98.89%.

[0039] Mesa etching was then performed. Photoresist was spin-coated onto the epitaxial wafer surface, and ultraviolet exposure and development were used to form a circular mesa pattern with a diameter of 40 μm. Inductively coupled plasma (ICP) dry etching was then performed using SiCl4 (30 sccm) and Ar (10 sccm) as etching gases at a power of 200 W for 62 seconds to form a mesa structure with a height of approximately 3.6 μm, exposing the oxide layer.

[0040] Next, wet oxidation was performed. The sample was placed in an oxidation furnace at 410°C, and nitrogen gas carrying water vapor was introduced at a flow rate of 1.5 sccm for 20 minutes. The oxide layer was oxidized laterally from the outside to the inside, eventually forming a current-limiting hole with a diameter of 10 μm, which precisely limited the injected current to the center of the active region.

[0041] Finally, electrode fabrication and annealing were completed. Using photolithography, sputtering, and lift-off processes, a Ti / Au (15nm / 400nm) ring-shaped p-type ohmic contact electrode was fabricated on the top of the p-DBR, and an AuGeNi / Au (50nm / 300nm) n-type ohmic contact electrode was fabricated on the back side of the substrate. After fabrication, annealing was performed in a rapid annealing furnace under the following conditions: under a nitrogen atmosphere, the temperature was increased to 200℃ in 15 seconds and held for 15 seconds, then increased to 430℃ in 20 seconds and held for 35 seconds, and finally decreased to 100℃ in 20 seconds and held for 20 seconds.

[0042] The VCSEL device fabricated in this embodiment fully meets the light source requirements of the health monitoring system, and the measured PPG signal is as follows: Figure 3 As shown.

[0043] Example 2: PPG signal acquisition and embedded system implementation;

[0044] This embodiment illustrates how to construct a PPG signal acquisition and processing system using the VCSEL prepared in Example 1 as the core light source.

[0045] The system hardware architecture is based on the STM32F407 microcontroller and integrates the following modules:

[0046] 1. Dual-wavelength VCSEL driver module: The microcontroller outputs adjustable signals through two independent PWM pins (PA1, PA2) to control two constant current source circuits, providing adjustable drive current for the 660nm red VCSEL and the 940nm near-infrared VCSEL respectively. The system uses a time-division multiplexing method to alternately light up the two VCSELs to avoid spectral crosstalk.

[0047] 2. Photoelectric Detection and Signal Processing Module: This module uses a silicon photodiode (PD) to receive the red light signal reflected back from human tissue. The subsequent processing circuitry employs a "transimpedance amplification-filtering-AC coupling" architecture. The first stage is a transimpedance amplifier composed of an MCP6001 low-noise operational amplifier, which converts the nA-level photocurrent into a voltage signal. Its output voltage... ,in The first stage uses a 470kΩ feedback resistor. The second stage consists of an RC circuit (470Ω, 2.2μF) forming a low-pass filter with a cutoff frequency of approximately 150Hz to filter out high-frequency noise. The third stage uses a 4.7μF coupling capacitor to suppress baseline drift.

[0048] 3. High-precision ADC module: The ADS1115 16-bit ADC chip is selected to digitize and sample the processed analog signal at a sampling rate of 1kHz. The ADS1115 communicates with the STM32F4 via the I²C bus (SDA:PB7, SCL:PB6).

[0049] The system software flow is as follows: After the STM32F4 is powered on, all peripherals are initialized. In the main loop, PPG sampling data is read from the ADS1115 via the I²C protocol and temporarily stored in the internal SRAM. At the same time, the data is uploaded to the host computer in real time for display and recording via the UART interface (PA9, PA10, baud rate 115200), and the host computer can be received to dynamically adjust the VCSEL drive current.

Claims

1. A photoplethysmography (PPG) system based on a vertical-cavity surface-emitting laser (VCSEL), characterized in that, It includes a multi-wavelength optical emission unit, a signal processing unit, an embedded control unit, and an AI vital sign analysis unit, which are connected in sequence.

2. The photoplethysmography (PPG) system based on a vertical-cavity surface-emitting laser according to claim 1, characterized in that, The multi-wavelength optical emitting unit includes VCSELs of various wavelengths, which operate alternately according to a preset period using a time-division driving method to avoid spectral crosstalk between different wavelengths. The epitaxial structure of each wavelength VCSEL includes a top electrode, an upper DBR, an active region, a lower DBR, and a bottom electrode. Both the upper and lower DBRs achieve high reflectivity through alternating layer structures. The active region confines charge carriers through a combination of multiple quantum wells and confinement layers, ensuring stable lasing at the corresponding wavelength. The reflectivity of the upper and lower DBRs... satisfy: （1）； In the formula, It is the vacuum wavelength. It is the effective absorption coefficient. It is the difference in refractive index between the two layers of DBR material; The active region employs a multi-quantum-well structure, which has a peak optical gain. Expressed by the following formula: （2）； In the formula, These are momentum matrix elements. It is the density of states. It is the net reversal factor; By setting confinement layers on both sides of the active region, the charge carriers are efficiently confined in the quantum well through band structure optimization, thereby improving the radiative recombination efficiency and ultimately achieving stable lasing. Using VCSEL as a light source, the specific wavelength light emitted by it interacts with the pulsation of blood volume in human tissue, and this interaction is converted into changes in light intensity received by a photodetector, ultimately demodulating the pulse wave signal.

3. The photoplethysmography (PPG) system based on a vertical-cavity surface-emitting laser according to claim 1, characterized in that, The signal processing unit adopts a low-noise amplification, multi-stage filtering, and adjustable gain architecture to adapt to the processing requirements of PPG signals corresponding to multi-wavelength VCSELs: the low-noise amplification circuit converts the photocurrent signals corresponding to each wavelength into voltage signals and amplifies them initially; the filtering circuit filters out high-frequency noise and baseline drift; and the adjustable gain circuit dynamically adjusts the gain according to the intensity differences of signals of different wavelengths to ensure that the signal strength of each wavelength is adapted to the subsequent processing requirements.

4. The photoplethysmography (PPG) system based on a vertical-cavity surface-emitting laser according to claim 1, characterized in that, The embedded control unit is based on a microcontroller and is connected to a high-precision analog-to-digital converter (ADC) module and a multi-wavelength VCSEL driving circuit. The ADC module performs digital conversion on the analog signals of each wavelength after signal processing and transmits the data to the microcontroller through a communication protocol. The multi-wavelength VCSEL driving circuit receives the control signal output by the microcontroller and provides adjustable driving current for each wavelength VCSEL.

5. The photoplethysmography (PPG) system based on a vertical-cavity surface-emitting laser according to claim 1, characterized in that, The AI ​​vital sign analysis unit includes a signal preprocessing module, a multi-wavelength feature extraction module, and a lightweight fusion learning model. The signal preprocessing module removes power frequency interference and motion artifacts through filtering to ensure the quality of PPG signals at each wavelength. The feature extraction module extracts time-domain, frequency-domain, and nonlinear features for each wavelength PPG signal. The lightweight fusion learning model is built based on decision tree integration, which integrates feature data from various wavelengths for inference calculation and outputs health parameters such as heart rate and blood oxygen saturation in real time.

6. The photoplethysmography (PPG) system based on a vertical-cavity surface-emitting laser according to claim 1, characterized in that, Core features are extracted for PPG signals at different wavelengths: the time-domain feature module captures key morphological parameters of the pulse wave through energy operators, reflecting the periodicity of heart rate and pulse intensity; the frequency-domain feature module extracts the main frequency of the signal and the energy ratio of different frequency bands through power spectral density analysis, and correlates them with vascular elasticity and microcirculation characteristics. The nonlinear feature module assesses the cardiovascular system dynamics through long-range correlation analysis. The feature extraction process for each wavelength signal runs independently, ultimately forming a multi-dimensional dual-path feature set.

Images