A circulating tumor multi-target in vivo synchronous identification device and method based on pulse waveform features

Abstract

The present application belongs to the technical field of tumor cell detection, and aims at the problem that circulating tumor cells, circulating tumor cell clusters and extracellular vesicles cannot be recognized synchronously at present. The present application provides a circulating tumor multi-target living body synchronous recognition device and method based on pulse waveform characteristics. The device comprises a light source and excitation assembly, a blood vessel positioning system, a living body monitoring unit, a light folding system and a signal processing and analysis assembly. The shaped laser light source provided by the light source and excitation assembly is guided to the living body monitoring unit by the light folding system for excitation. The excited fluorescent signal is conducted to the signal processing and analysis assembly by the light folding system to synchronously recognize and count the circulating tumor cells, circulating tumor cell clusters and extracellular vesicles. The present application realizes the living body synchronous recognition and dynamic monitoring of the three entities of circulating tumor cells, circulating tumor cell clusters and extracellular vesicles for the first time, and ensures that the detection result can truly reflect the dynamic behavior of each target under the physiological state.

Description

Technical Field

[0001] This invention belongs to the field of tumor cell detection technology, specifically relating to a device and method for simultaneous identification of multiple targets in living organisms of circulating tumors based on pulse waveform characteristics. Background Technology

[0002] Circulating tumor cells (CTCs), as important carriers of cancer metastasis, have become a key research direction in oncology. In recent years, CTC aggregates—complexes formed by multiple cell clusters—have been shown to possess metastatic potential far exceeding that of single cells, making them valuable for prognostic assessment. Meanwhile, extracellular vesicles, as nanoscale particles secreted by tumor cells, play crucial roles in tumor-microenvironment communication, pre-metastatic microenvironment formation, and immune regulation, and have emerged as new biomarkers in liquid biopsy. These three entities exhibit complex interactions during tumor metastasis, making it difficult to comprehensively assess tumor progression and treatment response using a single indicator.

[0003] Currently, the detection of circulating tumor cells (CTCs), CTC clusters, and extracellular vesicles primarily relies on in vitro analysis. While immunomagnetic bead enrichment combined with fluorescence imaging can identify CTCs and their clusters, its dependence on ex vivo blood samples means it cannot capture their real-time dynamic behavior within the circulatory system. Centrifugation and washing during sample processing easily disintegrate fragile cell clusters, causing systematic biases in the test results. For extracellular vesicles, existing detection methods mainly include nanoparticle tracking analysis, dynamic light scattering, transmission electron microscopy, and traditional flow cytometry. These methods all require in vitro blood collection and sample processing, making in vivo real-time monitoring impossible and difficult to distinguish between tumor-derived and normal cell-derived extracellular vesicles.

[0004] In the field of in vivo detection technology, while traditional small animal optical imaging systems can provide a holistic observation perspective, their technological limitations severely restrict their application value in circulating tumor cell research. These systems typically rely on bioluminescence or fluorescence imaging principles, enabling a direct visualization of the distribution of labeled cells within the organism. However, their spatial resolution is generally limited to the millimeter level, which is clearly insufficient for observing micrometer-sized cell clusters. When researchers attempt to improve resolution by increasing detector sensitivity or optimizing algorithms, they face the inherent contradiction between tissue penetration depth and signal attenuation: visible light has limited penetration in biological tissues, while near-infrared light, although having better penetration, offers relatively fewer available fluorescent labeling options. Furthermore, the complex optical environment within the organism, including tissue autofluorescence, hemoglobin absorption, and scattering effects at different tissue depths, severely interferes with the detection signal. These factors collectively make it difficult for traditional in vivo imaging techniques to accurately distinguish between individual circulating tumor cells and cell clusters, let alone detect extracellular vesicles, greatly limiting their application value in tumor metastasis mechanism research.

[0005] In the field of in vivo flow cytometry, studies have achieved non-invasive real-time monitoring of circulating tumor cells (CTCs) and preliminary identification of CTC clusters. In recent years, in vivo flow cytometry has also been shown to detect extracellular vesicle signals in circulating blood. Studies have shown that by analyzing the peak width characteristics of fluorescence pulses, it is possible to distinguish between larger cells and smaller vesicles to some extent. However, current technologies mainly focus on the detection and counting of single-type targets. Effective simultaneous identification methods are lacking when CTCs, CTC clusters, and extracellular vesicles coexist. Current methods often rely on a single fluorescence signal amplitude threshold for judgment. When multiple targets pass through the detection area sequentially or simultaneously, their superimposed signals are easily misjudged as a single event or rejected as noise due to complex waveform morphology. Although manual observation of peak shape can assist in judgment to some extent, this method is highly subjective, lacks standardized criteria, and has limited accuracy, resulting in a high rate of missed detection for cell clusters and making accurate quantification of extracellular vesicles even more difficult.

[0006] In fact, these three biological entities exhibit complex interactions and synergistic mechanisms during tumor metastasis, making it difficult to comprehensively assess tumor progression and treatment response using a single indicator. While some studies have attempted to differentiate between different types of targets using multicolor fluorescent labeling strategies, this method requires multiple fluorescent probes and multichannel detection systems, increasing system complexity and cost, and presenting technical challenges such as signal registration and spectral crosstalk between different channels. These limitations of existing technologies severely restrict in-depth exploration of the mechanisms of tumor hematogenous metastasis.

[0007] Therefore, developing a novel detection method that can simultaneously identify circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles has become an urgent technical problem to be solved in this field. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide a device and method for simultaneous identification of multiple targets in living organisms of circulating tumors based on pulse waveform characteristics.

[0009] The first objective of this invention is to provide a multi-target live synchronous identification device for circulating tumors based on pulse waveform characteristics, comprising a light source and excitation assembly, a blood vessel positioning system, a live monitoring unit, a refractive system, and a signal processing and analysis assembly; The shaped laser light source provided by the light source and excitation component is guided to the live monitoring unit by the refractive system for excitation. The excited fluorescence signal is then transmitted to the signal processing and analysis component by the refractive system to simultaneously identify and count circulating tumor cells, circulating tumor cell clusters and extracellular vesicles. The vascular positioning system automatically locates the target blood vessel's position information and adjusts the position of the live monitoring unit according to the blood vessel position information, so that the monitoring position of the live monitoring unit is aligned with the center of the target blood vessel.

[0010] Furthermore, the light source and excitation assembly include: Laser source; A laser reshaping system is used to reshape a laser source to obtain a reshaped laser source.

[0011] Furthermore, the laser shaping system includes a beam expander group and a cylindrical lens group arranged sequentially along the optical path.

[0012] Furthermore, the blood vessel localization system includes: A high-resolution imaging module is used to generate a real-time image stream of blood vessels and to identify and select the target blood vessel's real-time image stream through the real-time image stream. The autofocus and tracking module is used to automatically identify and locate the target blood vessel in the real-time image stream to obtain the blood vessel location information; The central control unit is used to generate control commands in real time based on the blood vessel location information to drive the live monitoring unit to move and adjust its position, so that the monitoring position of the live monitoring unit is aligned with the center of the target blood vessel.

[0013] Furthermore, the autofocus and tracking module includes: The blood vessel recognition submodule is used to extract the blood vessel region from the background tissue of the real-time image stream of the target blood vessel; The centerline extraction submodule is used to extract the subpixel coordinates of the centerline of blood vessels in the vascular region. The displacement compensation submodule is used to calculate the periodic displacement of the blood vessel centerline in real time based on the subpixel coordinates of the blood vessel centerline and predict the amount of displacement of the blood vessel centerline. The tracking control submodule is used to generate a feedforward compensation signal based on the predicted displacement to obtain the vessel position information.

[0014] Furthermore, the liveness monitoring unit includes: Animal experimental samples; Animal placement platform is used to fix animal experimental samples and monitor physiological signs in real time; The laser detection area is located on the blood vessels in the auricle or tail of the animal experimental sample.

[0015] Furthermore, the signal processing and analysis components include: Fluorescence detector; The signal analysis equipment includes a high-speed data acquisition card for continuously acquiring fluorescence signals detected by a fluorescence detector, and an analysis host for multi-target classification and identification of acquired fluorescence pulse signals and counting circulating tumor cells, circulating tumor cell clusters and extracellular vesicles passing through the laser detection area.

[0016] Furthermore, the circulating tumor multi-target live synchronous identification device based on pulse waveform features provided by the present invention also includes: The data processing and generation module is used to process the results of multi-target classification and recognition, as well as the counting results of circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles, and generate a local preliminary report. A cloud server is used to integrate and analyze the generated local preliminary reports to produce a professional analysis report; The user terminal interaction and results display unit is used to receive and display professional analysis reports in real time.

[0017] A second objective of this invention is to provide a method for simultaneous in vivo identification of multiple targets in circulating tumors based on pulse waveform features, implemented using the apparatus described in any of the preceding claims, comprising: Animal experimental samples labeled with fluorescent proteins and anesthetized were fixed on the animal placement platform, and the high-resolution imaging module was turned on to generate a real-time image stream of blood vessels. The real-time image stream of the target blood vessel was identified and selected to complete the initial localization. The laser source is activated and the excitation wavelength is selected. The laser shaping system then shapes and focuses the source to obtain the shaped laser source. The autofocus and tracking module is activated to automatically identify and locate the target blood vessel in the real-time image stream, and obtain the blood vessel position information. The central control unit generates control commands in real time based on the blood vessel position information to move the live monitoring unit so that the laser detection area is aligned with the center of the target blood vessel, thus completing the final positioning. The live monitoring unit and signal processing and analysis components are activated. The shaped laser source is guided to the laser detection area by the refractive system for excitation. The excited fluorescence signal is then transmitted to the signal processing and analysis components by the refractive system. The original fluorescence pulse signal flowing through the laser detection area is detected and continuously acquired by the fluorescence detector and high-speed data acquisition card. The acquired raw fluorescence pulse signals are evaluated in real time, and valid fluorescence pulse signals that meet the quality requirements are screened. The analysis host performs multi-target classification and identification on the selected effective fluorescence pulse signals, and counts the circulating tumor cells, circulating tumor cell clusters and extracellular vesicles that pass through the laser detection area.

[0018] Furthermore, the analysis host performs multi-target classification and identification on the screened valid fluorescence pulse signals, including: The host analyzer preprocesses the selected valid fluorescence pulse signals and performs multi-scale adaptive threshold detection to obtain fluorescence pulse events; The characteristic parameters of the fluorescence pulse event are extracted, including pulse peak intensity, pulse half width at half maximum (FWHM), rise time, fall time, pulse area, kurtosis coefficient, skewness coefficient, peak-to-valley ratio, and number of sub-peaks. Particle size inversion calculations are performed based on a quantitative relationship model between target particle size and pulse peak width to obtain particle size inversion results; Multi-target classification and recognition are performed based on the extracted feature parameters and particle size inversion results.

[0019] Furthermore, the analysis host preprocesses the selected valid fluorescence pulse signals and performs multi-scale adaptive threshold detection to obtain fluorescence pulse events, including: The analysis host preprocesses the selected valid fluorescence pulse signals; The preprocessed signal is divided into a high-threshold layer and a low-threshold layer based on the peak intensity. For the high-threshold layer, an adaptive thresholding method combined with sliding window energy is used to extract high-amplitude pulses corresponding to circulating tumor cells and circulating tumor cell clusters. For the low-threshold layer, wavelet packet reconstruction combined with peak detection algorithm is used to extract weak pulses corresponding to extracellular vesicles, thus obtaining fluorescence pulse events.

[0020] Furthermore, the multi-target classification and recognition based on the extracted feature parameters and particle size inversion results includes: Amplitude screening is performed based on peak intensity. If the peak intensity is less than one-tenth of the high threshold layer, it is included in the extracellular vesicle candidate set; otherwise, it is included in the circulating tumor cell or mass candidate set. For the candidate set of extracellular vesicles, the determination is made based on the particle size inversion results: if the inverted particle size is less than 1 μm and the pulse half width at half maximum is in the range of 0.5~1.0 ms, it is determined to be an extracellular vesicle. For candidate sets of circulating tumor cells or clumps, a comprehensive judgment is made based on the particle size inversion results and waveform morphology: if the inverted particle size is in the range of 10~30 μm, and the pulse half width at half maximum is in the range of 2~8 ms, and the waveform is a single peak, smooth and symmetrical, then it is determined to be a single circulating tumor cell. If the inverted particle size is greater than 30 μm, the pulse half width at half maximum is greater than 8 ms, the waveform exhibits multi-peak characteristics with the number of sub-peaks greater than or equal to 2 and the peak-to-valley ratio less than 0.8, it is determined to be a circulating tumor cell mass. If the inverted particle size is within the cellular range but the waveform is flat-topped with multiple peaks and the peak-to-valley ratio is greater than 0.8, then it represents multiple circulating tumor cells.

[0021] Furthermore, the method for simultaneous live in vivo identification of circulating tumors based on pulse waveform features provided by the present invention also includes: The data processing module processes the multi-target classification and identification results and the counting results of circulating tumor cells, circulating tumor cell clusters and extracellular vesicles, and generates a local preliminary report. The cloud server integrates and analyzes the generated local preliminary report to produce a professional analysis report; Professional analysis reports are received and displayed in real time through the user terminal interaction and results display unit.

[0022] Compared with the prior art, the advantages of this invention are as follows: (1) By combining in vivo flow cytometry with fluorescent protein labeling, this invention has for the first time achieved simultaneous in vivo identification and dynamic monitoring of three entities: circulating tumor cells, circulating tumor cell clumps, and extracellular vesicles. This effectively avoids the problems of clumping disintegration and extracellular vesicle loss caused by sample processing in traditional in vitro detection, and ensures that the detection results can truly reflect the dynamic behavior of each target under physiological conditions.

[0023] (2) By designing a vascular positioning system, this invention achieves automatic identification of the vascular centerline and dynamic compensation for physiological displacement. This improvement enables the laser detection area to be accurately aligned with the target vascular for a long time, significantly improving the collection efficiency and signal-to-noise ratio of fluorescence signals, and providing a stable and reliable positioning basis for subsequent accurate multi-target detection.

[0024] (3) In view of the differences in signal characteristics of circulating tumor cells, circulating tumor cell clusters with multiple peaks of high amplitude and extracellular vesicles, this invention designs a two-layer adaptive threshold system with high and low amplitude, and combines wavelet packet transform to extract weak signals. This solves the problem that the traditional single threshold method cannot simultaneously take into account strong and weak signals, making it possible to simultaneously detect multiple targets with size spans from nanometers to hundreds of micrometers, and significantly expanding the detection dynamic range of in vivo flow cytometer.

[0025] (4) This invention establishes a quantitative relationship between target particle size and pulse peak width based on a particle kinematic model, thereby determining the characteristic peak width range of various targets. By combining peak height screening, peak width-to-particle size inversion, and waveform morphology analysis, specifically for clumping identification, two features, the number of sub-peaks and the peak-to-valley ratio, are introduced. By setting the judgment rules for both, clumps and single cells can be effectively distinguished. For extracellular vesicle detection, weak signals are extracted through wavelet packet transform, and cells and vesicles are distinguished through peak width analysis, thus achieving accurate classification of the three types of targets.

[0026] (5) This invention constructs a high-speed data acquisition and local-cloud collaborative analysis platform. By calculating multiple parameters such as EV / CTC ratio and clumping ratio, it can generate professional analysis reports including migration risk assessment. The user terminal interaction and result display unit intuitively displays the classification results and counting curves, reducing the difficulty of using the equipment. Attached Figure Description

[0027] Figure 1 The structural block diagram of the circulating tumor multi-target live synchronous identification device based on pulse waveform features provided by the present invention; Figure 2 A schematic diagram of the principle of the circulating tumor multi-target live synchronous identification method based on pulse waveform features provided by the present invention; In the diagram: 1. Light source and excitation assembly; 2. Blood vessel positioning system; 3. In vivo monitoring unit; 4. Refractive system; 5. Signal processing and analysis assembly; 6. High-brightness LED illumination lamp; 7. CMOS camera; 8. Laser light source; 9. Beam expander assembly; 10. Cylindrical lens assembly; 11. Animal experimental sample; 12. Animal placement platform; 13. Laser detection area; 14. Spectrometer; 15. Fluorescence detector; 16. Signal analysis equipment. Detailed Implementation

[0028] Exemplary embodiments of this application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of this application and to fully convey the scope of this application to those skilled in the art.

[0029] Example 1 like Figure 1 As shown, this embodiment of the invention provides a multi-target live synchronous identification device for circulating tumors based on pulse waveform characteristics, including a light source and excitation component 1, a blood vessel positioning system 2, a live monitoring unit 3, a refractive system 4, and a signal processing and analysis component 5; The shaped laser light source provided by the light source and excitation component 1 is guided to the live monitoring unit 3 by the refractive system 4 for excitation. The excitation fluorescence signal is then transmitted to the signal processing and analysis component 5 by the refractive system 4 to simultaneously identify and count circulating tumor cells, circulating tumor cell clusters and extracellular vesicles. The vascular positioning system 2 automatically locates the target vascular location information and generates a control signal based on the vascular location information to drive the live monitoring unit 3 or its location to adjust so that the monitoring position of the laser detection area is aligned with the center of the target vascular.

[0030] In this embodiment, the light source and excitation assembly 1 includes a laser light source 8 and a laser shaping system arranged sequentially along the optical path. The laser shaping system is used to shape the laser light source 8 to obtain a shaped laser light source. Specifically, the laser shaping system includes a beam expander group 9 and a cylindrical lens group 10 arranged sequentially along the optical path. The laser light source 8 uses a multi-wavelength semiconductor laser with a wavelength range of 488~640 nm and an adjustable output power range of 10~100 mW. The multi-wavelength design can adapt to animal experimental samples 11 labeled with different fluorescent proteins, improving the versatility of the device provided by this invention. The animal experimental sample 11 can be constructed by inoculating experimental animals with tumor cells labeled with fluorescent proteins, and the specific method can adopt conventional tumor model construction methods in the art. When using a single fluorescent labeling strategy, a specific wavelength matching the fluorescent protein can be selected for excitation.

[0031] In this embodiment, the beam expander group 9 and the cylindrical lens group 10 can shape the laser beam into an elliptical spot with a size of 5×50~70 μm. This spot size matches the diameter of the target blood vessel, ensuring that the laser beam can completely cover the blood vessel lumen. The beam expander group 9 and the cylindrical lens group 10 are fixed to the optical platform by a precision adjustment frame equipped with a locking mechanism, ensuring the long-term stability of the optical path. Specifically, the precision adjustment frame can be implemented using conventional optical adjustment components or equivalent structures in the art. This embodiment does not involve any improvement to the structure of the precision adjustment frame, so its structure will not be described in detail here. Specifically, the precision adjustment frame can be an optical mounting bracket with multi-degree-of-freedom adjustment functions, such as a two-dimensional or three-dimensional optical adjustment frame, which achieves position adjustment through a threaded fine-tuning mechanism and is fixed by a locking structure. In practical applications, commercially available optical adjustment frames, such as similar products provided by manufacturers like Thorlabs and Daheng Optoelectronics, can be used.

[0032] In this embodiment, the blood vessel positioning system includes a high-resolution imaging module, an autofocus and tracking module, and a central control unit. The high-resolution imaging module generates a real-time image stream of the blood vessel and identifies and selects the real-time image stream of the target blood vessel through the real-time image stream. The autofocus and tracking module automatically identifies and locates the real-time image stream of the target blood vessel to obtain the blood vessel position information. The central control unit generates control commands in real time according to the blood vessel position information to drive the electric three-dimensional displacement platform and the animal experimental sample 11 to move, so that the monitoring position of the live monitoring unit (i.e., the laser detection area 13) is always aligned with the center of the target blood vessel, thereby realizing dynamic compensation and tracking of physiological displacement.

[0033] Specifically, the high-resolution imaging module includes a high-brightness LED illumination lamp 6 and a CMOS camera 7. The high-brightness illumination lamp 6 uses a high-brightness LED light source with a center wavelength of 530 nm, and its luminous intensity can be precisely adjusted via current to adapt to the light transmission characteristics of different tissues. The CMOS camera 7 is equipped with a long working distance objective lens with a magnification of 10-40x, a resolution of 1920×1080 pixels, and a frame rate of up to 60 fps, capable of clearly resolving vascular structures with diameters of 20-100 μm. The entire high-resolution imaging module is fixed by an electrically driven three-dimensional displacement platform with a displacement accuracy of 0.1 μm; the electrically driven three-dimensional displacement platform can be implemented using conventional precision displacement devices or equivalent structures in the art. For example, the electrically driven three-dimensional displacement platform can consist of three orthogonally driven linear motion modules, specifically using commercially available products such as the electrically driven three-dimensional displacement stages from Thorlabs or Daheng Optoelectronics, or implemented using equivalent mechanical structures known in the art.

[0034] In this embodiment, the autofocus and tracking module receives the real-time image stream from the CMOS camera 7 and automatically identifies and locks the central axis position of the target blood vessel through the built-in image processing algorithm; specifically, it includes a blood vessel recognition submodule, a centerline extraction submodule, a displacement compensation submodule, and a tracking control submodule. The blood vessel recognition submodule uses a tubular structure enhancement algorithm based on the Hessian matrix combined with an adaptive threshold segmentation algorithm to extract the blood vessel region from the background tissue of the real-time image stream of the target blood vessel; the centerline extraction submodule extracts the subpixel coordinates of the blood vessel centerline of the blood vessel region through a morphological thinning algorithm. The displacement compensation submodule calculates the lateral and longitudinal displacements of the vessel centerline relative to a reference position at adjacent time points based on the subpixel coordinate sequence of the vessel centerline in continuously acquired image frames, constructing a time displacement curve of the vessel centerline. Then, it performs periodic analysis on the time displacement curve, separating the low-frequency and high-frequency displacement components caused by respiration and heartbeat, and predicts the vessel centerline displacement at the next or subsequent time points based on the amplitude, frequency, and phase information of the aforementioned displacement components. The tracking control submodule generates a feedforward compensation signal based on the predicted displacement to obtain the vessel position information.

[0035] In this embodiment, the live animal monitoring unit 3 includes an animal experimental sample 11, an animal placement platform 12, and a laser detection area 13. The animal placement platform 12 includes a motorized three-dimensional displacement platform and a precision temperature control system, a multi-degree-of-freedom body positioning device, and a physiological sign monitoring sensor mounted on the motorized three-dimensional displacement platform. The precision temperature control system uses a feedback heating element, which can maintain the animal's body temperature within the range of 35~37.5℃, with fluctuations of less than ±0.2℃. The multi-degree-of-freedom body positioning device uses an adjustable flexible clamp, suitable for experimental animals of different sizes. The physiological sign monitoring sensor monitors parameters such as respiration and heart rate of the animal experimental sample in real time. The laser detection area 13 is located on the auricle or tail blood vessels of the animal experimental sample 11.

[0036] In this embodiment, the refractive system 4 includes a condenser lens and at least one beam splitter 14. The beam splitter 14 is preferably a dichroic mirror or a beam splitter, employing multilayer dielectric film technology. By setting the reflectivity of the excitation light and the transmittance of the emitted light, the optical path allocation between the light source, the excitation component 1, and the signal processing and analysis component 5 is realized. Simultaneously, the imaging optical path of the blood vessel positioning system 2 is coupled to the IVFC detection optical path through the refractive system 4, so that the imaging field of view and the laser detection area 13 are kept in the same spatial position, thereby realizing automatic positioning and alignment based on image feedback. The blood vessel positioning system 2 automatically locates the blood vessel position information of the target blood vessel and generates a control signal according to the blood vessel position information to drive the live monitoring unit 3 or its location to adjust, so that the monitoring position of the laser detection area 13 is aligned with the center of the target blood vessel.

[0037] Specifically, when using a single fluorescent label, a dichroic mirror and filter matching the fluorescence wavelength can be selected for configuration. The invention also includes an environmental control system comprising a vibration isolation platform and a precision temperature and humidity controller; the vibration isolation platform and precision temperature and humidity controller can be implemented using conventional environmental control equipment or equivalent structures in the art. The vibration isolation platform employs air spring vibration isolation technology, with an intrinsic frequency less than 2 Hz, effectively isolating building vibrations. The precision temperature and humidity controller maintains an ambient temperature preferably 23 ± 1 ℃ and a relative humidity preferably 50% ± 5%, ensuring stable operation of the optical system and electronic equipment. All optical components are dynamically stabilized and fixed to the vibration-isolated optical platform, ensuring long-term reproducibility and stability of the optical path.

[0038] In this embodiment, the signal processing and analysis component 5 includes a fluorescence detector 15 and a signal analysis device 16. The fluorescence detector 15 is used to detect the fluorescence signal generated by the target blood vessel under the action of excitation light. Specifically, the fluorescence detector 15 adopts a photon-counting photomultiplier tube or a low-noise avalanche photodiode, and its detection sensitivity can reach the single-photon level. The signal analysis device includes a high-speed data acquisition card for continuously acquiring the fluorescence signal detected by the fluorescence detector 15, and an analysis host for multi-target classification and identification of the acquired fluorescence pulse signals and counting circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles passing through the laser detection area. The data acquisition card has a sampling rate of up to 100 MS / s, a quantization accuracy of 16 bits, and is equipped with a large-capacity onboard buffer to ensure no data loss during long-term monitoring. The analysis host is equipped with multi-target synchronous recognition algorithm software based on pulse waveform characteristics, which can process the fluorescence signal acquired by a single channel in real time and synchronously identify and count circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles by analyzing the peak width, amplitude, waveform morphology, and other characteristics of the pulse.

[0039] In this embodiment, the circulating tumor multi-target live synchronous identification device based on pulse waveform features provided by the present invention further includes a data processing and generation module, a cloud server, and a user terminal interaction and result display unit. Specifically, the data processing and generation module processes the multi-target classification and identification results and the counting results of circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles, and generates a local preliminary report; the cloud server integrates and analyzes the generated local preliminary report to generate a professional analysis report; and the user terminal interaction and result display unit receives and displays the professional analysis report in real time.

[0040] Example 2 like Figure 2 As shown, this embodiment of the invention provides a method for simultaneous live identification of multiple targets in circulating tumors based on pulse waveform features, implemented using the device described in Embodiment 1, including: Step 1: Fix the anesthetized animal experimental sample 11 labeled with fluorescent protein onto the animal placement platform 12, and turn on the high-resolution imaging module to generate a real-time image stream of blood vessels. Then, identify and select the real-time image stream of the target blood vessel through the real-time image stream to complete the initial localization. In this embodiment, a single fluorescent labeling strategy is employed. All test targets, including circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles, are labeled with the same fluorescent protein, eliminating the need for additional fluorescent probes. Extracellular vesicles derived from tumor cells encapsulate part of the cytoplasm during formation and therefore also carry fluorescent protein signals, allowing detection at the same excitation wavelength. The high-brightness LED illumination lamp 6 and CMOS camera 7 of the high-resolution imaging module are activated to generate a real-time image stream of blood vessels. Suitable target blood vessels are identified in the real-time image stream manually or semi-automatically, typically selecting vessels in the auricle or tail with a diameter of 30-50 μm, stable blood flow, and easy observation, to complete the initial localization. Simultaneously, the physiological sign monitoring sensor is activated to ensure that the animal experimental sample 11 is in a stable physiological state.

[0041] Step 2: Start the laser source 8 and select the excitation wavelength. Shape and focus the light source through the beam expander group 9 and the cylindrical lens group 10 to obtain the shaped laser source. In this embodiment, the laser source is activated, and an excitation wavelength matching the fluorescent protein is selected. The beam is shaped and focused using a beam expander group 9 and a cylindrical lens group 10. A beam analyzer is used to measure the spot size and energy distribution on the detection plane. The beam expander group 9 and the cylindrical lens group 10 are adjusted to match the spot size with the target blood vessel diameter and to achieve a uniform elliptical flat-top energy distribution. This completes the calibration of the laser source 8, providing a stable excitation source for subsequent detection.

[0042] Step 3: The autofocus and tracking module is activated to automatically identify and locate the target blood vessel in the real-time image stream, obtaining the vessel position information (i.e., the coordinates of the vessel centerline). The central control unit generates control commands in real time based on the vessel position information to drive the electric three-dimensional displacement platform and the animal experimental sample 11 to move, ensuring that the laser detection area 13 is always aligned with the center of the target blood vessel, that is, the laser beam waist plane is precisely coincident with the central axis of the blood vessel, completing the final positioning. The autofocus and tracking module monitors changes in the vessel position at regular intervals and continuously performs dynamic tracking compensation. When the vessel position drift is detected to exceed the allowable range, repositioning is automatically performed.

[0043] Step 4: Activate the liveness monitoring unit 3 and the signal processing and analysis component 5. The shaped laser source 8 is guided by the refractive system 4 to the laser detection area 13 for excitation. The excited fluorescence signal is then transmitted to the signal processing and analysis component 5 via the refractive system 4. The original fluorescence pulse signal flowing through the laser detection area 13 is detected and continuously acquired by the fluorescence detector 15 and the high-speed data acquisition card. All targets to be tested, including circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles, are excited by a laser wavelength matching the fluorescent protein. The generated fluorescence signals are received by the same detection channel, establishing a single-channel signal sequence with a timestamp. Data acquisition adopts a trigger mode. When the signal exceeds the preset baseline threshold, the complete pulse waveform is recorded, and the pulse arrival time is also recorded.

[0044] The blood vessel location information is output to the central control unit in the form of time-stamped data for determining the alignment status of the laser detection area 13 and generating control signals to adjust the position of the live monitoring unit 3 or its location. At the same time, the blood vessel location information can be synchronously transmitted to the signal processing and analysis component 5 for time synchronization and spatial correlation with the fluorescence pulse signal for data fusion analysis and signal quality assessment.

[0045] Step 5: Perform real-time quality assessment on the acquired raw fluorescence pulse signals and screen for valid fluorescence pulse signals that meet the quality requirements. Specifically, calculate parameters such as the signal-to-noise ratio (SNR) and baseline stability of the raw fluorescence pulse signals to determine whether they meet the analysis requirements. The specific analysis requirements are: comprehensively judge the SNR, baseline stability, and waveform characteristics of the raw fluorescence pulse signals. The SNR must be higher than a preset threshold, the baseline drift and fluctuation amplitude must be within a preset range, and the pulse signal should have a clear peak structure and complete rising and falling edges. When the above indicators simultaneously meet the preset conditions, the raw fluorescence pulse signal is determined to be a valid signal. When the signal quality deteriorates or abnormal noise occurs continuously, the system issues a prompt and automatically adjusts detection parameters such as laser power and detector gain according to preset rules, or prompts to check the sample to be tested, until valid fluorescence pulse signals that meet the quality requirements are obtained.

[0046] Step 6: The host analyzer performs multi-target classification and identification on the selected valid fluorescence pulse signals, and counts circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles passing through the laser detection area. Specifically, the host analyzer is equipped with multi-target synchronous identification algorithm software based on pulse waveform characteristics. The core principle of this algorithm is to infer the target particle size from the pulse peak width and combine it with peak height information to comprehensively determine the target type.

[0047] In this embodiment, the analysis host performs multi-target classification and identification on the screened valid fluorescence pulse signals, specifically including: Step 61: The host analyzes the selected valid fluorescence pulse signals, preprocesses them, and performs multi-scale adaptive threshold detection to obtain fluorescence pulse events; including: Step 611: The host analyzer preprocesses the selected effective fluorescence pulse signals; specifically, the host analyzer performs digital filtering on the selected effective fluorescence pulse signals, using wavelet transform or Butterworth low-pass filter to remove high-frequency noise and baseline drift, and extracts clean pulse waveform data. Step 612: Divide the preprocessed signal into high-threshold layers and low-threshold layers according to the peak intensity; specifically targeting the differences in signal characteristics of large-size high fluorescence intensity of circulating tumor cells, large-size multi-peak high fluorescence intensity of circulating tumor cell masses, and nanoscale low fluorescence intensity of extracellular vesicles, design a multi-scale adaptive threshold strategy to divide the fluorescence pulse events detected in the preprocessed signal into high-threshold layers and low-threshold layers according to their peak intensity.

[0048] Step 613: For the high-threshold layer, an adaptive thresholding method combined with sliding window energy is used to extract high-amplitude pulses corresponding to circulating tumor cells and circulating tumor cell clusters. For the low-threshold layer, wavelet packet reconstruction combined with peak detection algorithm is used to extract weak pulses corresponding to extracellular vesicles, thus obtaining fluorescence pulse events. For closely connected or overlapping pulses, a curve fitting method is used for separation to ensure that each fluorescence pulse event corresponds to a target passage event.

[0049] Step 62: Extract the characteristic parameters of the fluorescence pulse event; among which, the characteristic parameters include pulse peak intensity, pulse half width at half maximum, rise time, fall time, pulse area, kurtosis coefficient, skewness coefficient, peak-to-valley ratio, and number of sub-peaks; Specifically, the peak pulse intensity reflects the fluorescence intensity of the target and is related to the expression level and size of the target's fluorescent protein; the pulse half-width at half-maximum (FWHM), i.e., the duration of the target passing through the laser spot, is related to the target's size, blood flow velocity, and laser spot width; the rise time and fall time reflect the rate at which the target enters and leaves the spot; the pulse area is the integral intensity; the kurtosis coefficient reflects the sharpness of the waveform; the skewness coefficient reflects the symmetry of the waveform; the peak-to-valley ratio, i.e., the ratio of peak values ​​to valley values ​​in a multi-peak signal, reflects the aggregation morphology of the target; and the number of sub-peaks, i.e., the number of local maxima in the pulse waveform, reflects the number of cells in the cluster.

[0050] Step 63: Perform particle size inversion calculation based on the quantitative relationship model between target particle size and pulse peak width to obtain the particle size inversion results; Specifically, this invention establishes a quantitative relationship model between target particle size and pulse peak width. When a target blood vessel passes through a laser spot of width L at a velocity v, the generated fluorescence pulse width T and the target blood vessel diameter d satisfy the following relationship: T = (L + d) / v. Based on this model, and using the typical physiological blood flow velocity range (5~10 mm / s) and the known spot width L = 5 μm, the particle size d of the target blood vessel can be calculated by back-calculating the measured fluorescence pulse width T. Through in vitro calibration experiments, using fluorescent microspheres of known size to establish an accurate peak width-size calibration curve can further improve the accuracy of particle size inversion.

[0051] Step 64: Perform multi-target classification and recognition based on the extracted feature parameters and particle size inversion results, specifically including: Step 641: Amplitude screening is performed based on peak intensity. If the peak intensity is less than one-tenth of the high threshold layer, it is included in the extracellular vesicle candidate set; otherwise, it is included in the circulating tumor cell or mass candidate set. Step 642: For the candidate set of extracellular vesicles, make a judgment based on the particle size inversion results: if the inverted particle size is less than 1 μm and the pulse half width at half maximum is in the range of 0.5~1.0 ms, it is determined to be an extracellular vesicle; Step 643: For the candidate set of circulating tumor cells or clumps, a comprehensive judgment is made based on the particle size inversion results and waveform morphology: if the inverted particle size is in the range of 10~30 μm, and the pulse half width at half maximum is in the range of 2~8 ms, and the waveform is a single peak, smooth and symmetrical, then it is determined to be a single circulating tumor cell. Step 644: If the inverted particle size is greater than 30 μm, the pulse half width at half maximum is greater than 8 ms, the waveform shows multi-peak characteristics and the number of sub-peaks is greater than or equal to 2 and the peak-to-valley ratio is less than 0.8, then it is determined to be a circulating tumor cell mass; Step 645: If the inverted particle size is within the cellular range but the waveform is flat-topped with multiple peaks and the peak-to-valley ratio is greater than 0.8, then it is multiple circulating tumor cells, which may be multiple circulating tumor cells passing through continuously. Further analysis of the time interval between sub-peaks is needed for confirmation.

[0052] This invention achieves high-precision simultaneous identification of three types of targets with size ranges from nanometers to hundreds of micrometers by combining peak height screening and peak width particle size inversion.

[0053] Step 65: Based on the identification results from Step 6, the number of individual circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles passing through the laser detection area 13 is counted in real time. The proportion of circulating tumor cell clusters to total circulating tumor cell events and the ratio of extracellular vesicles to circulating tumor cells are calculated. The size distribution of each target is analyzed, and the time-varying pattern is estimated based on pulse width and area to obtain complete dynamic parameters such as flow velocity and transit time.

[0054] The method for simultaneous live in vivo identification of circulating tumors based on pulse waveform features provided in this embodiment of the invention further includes: Step 7: The data processing generation module processes the multi-target classification and identification results and the counting results of circulating tumor cells, circulating tumor cell clusters and extracellular vesicles, and generates a local preliminary report. Specifically, data acquisition automatically stops after monitoring reaches a set time (e.g., 30-60 minutes) or after collecting a preset number of target events. All collected data (i.e., multi-target classification results and counts of circulating tumor cells, circulating tumor cell clusters, and extracellular vesicles) is standardized to remove outliers and generate a local preliminary report containing the original data, identification results, classification statistics, and size distribution. This local preliminary report is stored in a local database according to a preset format and can optionally be transmitted to a cloud server via network.

[0055] Step 8: The cloud server integrates and analyzes the generated local preliminary report to produce a professional analysis report; Specifically, when the data processing and generation module connects to the cloud server, local data can be automatically uploaded. The cloud server integrates and analyzes the preliminary local reports of multiple batches and multiple animal data received, and uses multi-dimensional data modeling technology to explore the correlation between the dynamic change trends of each target and tumor progression and treatment response, generating professional analysis reports that include mass burden index, extracellular vesicle (EV) / circulating tumor cell (CTC) ratio, metastasis risk assessment, and treatment effect monitoring.

[0056] Step 9: Receive and display professional analysis reports in real time through the user terminal interaction and results display unit. The user terminal interaction and results display unit includes a local host, tablet, or mobile app, capable of receiving and displaying monitoring results and analysis reports. Furthermore, the user terminal interaction and results display unit provides a real-time monitoring visualization interface, allowing playback of raw signal waveforms, displaying the classification and identification results of each target, and presenting curves showing the change in the quantity of each target over time. The interactive analysis platform allows researchers to manually review suspicious events, adjust analysis parameters, and export raw data, facilitating in-depth mechanistic research.

[0057] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

Claims

1. A device for simultaneous in vivo identification of multiple targets in circulating tumors based on pulse waveform characteristics, characterized in that, It includes a light source and excitation component (1), a blood vessel positioning system (2), a live monitoring unit (3), a refractive system (4), and a signal processing and analysis component (5). The laser light source provided by the light source and excitation component (1) after shaping is guided to the live monitoring unit (3) by the refractive system (4) for excitation. The excitation fluorescence signal is then transmitted to the signal processing and analysis component (5) by the refractive system (4) to simultaneously identify and count circulating tumor cells, circulating tumor cell clusters and extracellular vesicles. The vascular positioning system (2) automatically locates the vascular position information of the target vascular vessel in a synchronous manner, and adjusts the position of the live monitoring unit (3) according to the vascular position information so that the monitoring position of the live monitoring unit (3) is aligned with the center of the target vascular vessel.

2. The circulating tumor multi-target live synchronous identification device based on pulse waveform characteristics according to claim 1, characterized in that, The light source and excitation components include: Laser source (8); A laser shaping system is used to shape a laser source (8) to obtain a shaped laser source.

3. The circulating tumor multi-target live synchronous identification device based on pulse waveform characteristics according to claim 1, characterized in that, The vascular positioning system (2) includes: A high-resolution imaging module is used to generate a real-time image stream of blood vessels and to identify and select the target blood vessel's real-time image stream through the real-time image stream. The autofocus and tracking module is used to automatically identify and locate the target blood vessel in the real-time image stream to obtain the blood vessel location information; The central control unit is used to generate control commands in real time based on the blood vessel location information to drive the live monitoring unit (3) to adjust its position so that the monitoring position of the live monitoring unit (3) is aligned with the center of the target blood vessel.

4. The circulating tumor multi-target live synchronous identification device based on pulse waveform features according to claim 3, characterized in that, The autofocus and tracking module includes: The blood vessel recognition submodule is used to extract the blood vessel region from the background tissue of the real-time image stream of the target blood vessel; The centerline extraction submodule is used to extract the subpixel coordinates of the centerline of blood vessels in the vascular region. The displacement compensation submodule is used to calculate the periodic displacement of the blood vessel centerline in real time based on the subpixel coordinates of the blood vessel centerline and predict the amount of displacement of the blood vessel centerline. The tracking control submodule is used to generate a feedforward compensation signal based on the predicted displacement to obtain the vessel position information.

5. The circulating tumor multi-target live synchronous identification device based on pulse waveform characteristics according to claim 1, characterized in that, The liveness monitoring unit (3) includes: Animal experimental samples (11); Animal placement platform (12) is used to fix animal experimental samples (11) and monitor physiological signs in real time; The laser detection area (13) is located on the auricle or tail blood vessels of the animal experimental sample (11).

6. The circulating tumor multi-target live synchronous identification device based on pulse waveform features according to claim 1, characterized in that, The signal processing and analysis component (5) includes: Fluorescence detector (15); The signal analysis device (16) includes a high-speed data acquisition card for continuously acquiring fluorescence signals detected by a fluorescence detector, and an analysis host for multi-target classification and identification of acquired fluorescence pulse signals and counting of circulating tumor cells, circulating tumor cell clusters and extracellular vesicles passing through the laser detection area.

7. A method for simultaneous in vivo identification of multiple targets in circulating tumors based on pulse waveform features, characterized in that, Implemented using the apparatus according to any one of claims 1-6, comprising: The animal experimental sample (11) labeled with fluorescent protein and anesthetized was fixed on the animal placement platform (11), and the high-resolution imaging module was turned on to generate a real-time image stream of blood vessels. The real-time image stream of the target blood vessel was identified and selected through the real-time image stream to complete the initial positioning. Start the laser source (8) and select the excitation wavelength. Shape and focus the source through the beam expander (9) and cylindrical lens group (10) to obtain the shaped laser source. The autofocus and tracking module is activated to automatically identify and locate the real-time image stream of the target blood vessel, and obtain the blood vessel location information. The central control unit generates control commands in real time based on the blood vessel location information to move the live monitoring unit (3) so that the laser detection area (13) is aligned with the center of the target blood vessel, and the final positioning is completed. The live monitoring unit (3) and signal processing and analysis component (5) are activated. The shaped laser light source is guided to the laser detection area (13) by the refractive system (4) for excitation. The excited fluorescence signal is then transmitted to the signal processing and analysis component (5) by the refractive system (4). The original fluorescence pulse signal flowing through the laser detection area is detected and continuously acquired by the fluorescence detector (15) and high-speed data acquisition card. The acquired raw fluorescence pulse signals are evaluated in real time, and valid fluorescence pulse signals that meet the quality requirements are screened. The analysis host performs multi-target classification and identification on the selected effective fluorescence pulse signals, and counts the circulating tumor cells, circulating tumor cell clusters and extracellular vesicles that pass through the laser detection area (13).

8. The method according to claim 7, characterized in that, The analysis host performs multi-target classification and identification on the selected valid fluorescence pulse signals, including: The host analyzer preprocesses the selected valid fluorescence pulse signals and performs multi-scale adaptive threshold detection to obtain fluorescence pulse events; The characteristic parameters of the fluorescence pulse event are extracted, including pulse peak intensity, pulse half width at half maximum (FWHM), rise time, fall time, pulse area, kurtosis coefficient, skewness coefficient, peak-to-valley ratio, and number of sub-peaks. Particle size inversion calculations are performed based on a quantitative relationship model between target particle size and pulse peak width to obtain particle size inversion results; Multi-target classification and recognition are performed based on the extracted feature parameters and particle size inversion results.

9. The method according to claim 8, characterized in that, The analysis host preprocesses the selected valid fluorescence pulse signals and performs multi-scale adaptive threshold detection to obtain fluorescence pulse events, including: The analysis host preprocesses the selected valid fluorescence pulse signals; The preprocessed signal is divided into a high-threshold layer and a low-threshold layer based on the peak intensity. For the high-threshold layer, an adaptive thresholding method combined with sliding window energy is used to extract high-amplitude pulses corresponding to circulating tumor cells and circulating tumor cell clusters. For the low-threshold layer, wavelet packet reconstruction combined with peak detection algorithm is used to extract weak pulses corresponding to extracellular vesicles, thus obtaining fluorescence pulse events.

10. The method according to claim 9, characterized in that, The multi-target classification and recognition based on extracted feature parameters and particle size inversion results includes: Amplitude screening is performed based on peak intensity. If the peak intensity is less than one-tenth of the high threshold layer, it is included in the extracellular vesicle candidate set; otherwise, it is included in the circulating tumor cell or mass candidate set. For the candidate set of extracellular vesicles, the determination is made based on the particle size inversion results: if the inverted particle size is less than 1 μm and the pulse half width at half maximum is in the range of 0.5~1.0 ms, it is determined to be an extracellular vesicle. For candidate sets of circulating tumor cells or clumps, a comprehensive judgment is made based on the particle size inversion results and waveform morphology: if the inverted particle size is in the range of 10~30 μm, and the pulse half width at half maximum is in the range of 2~8 ms, and the waveform is a single peak, smooth and symmetrical, then it is determined to be a single circulating tumor cell. If the inverted particle size is greater than 30 μm, the pulse half-width is greater than 8 ms, the waveform shows multi-peak characteristics and the number of sub-peaks is greater than or equal to 2 and the peak-to-valley ratio is less than 0.8, then it is determined to be a circulating tumor cell mass. If the inverted particle size is within the cellular range but the waveform is flat-topped with multiple peaks and the peak-to-valley ratio is greater than 0.8, then it represents multiple circulating tumor cells.

Images