A wearable micro-in vivo flow cytometer

Abstract

The present invention provides a wearable miniature in vivo flow cytometer, characterized by comprising a multi-channel miniature fluorescence in vivo imaging microscope, the multi-channel miniature fluorescence in vivo imaging microscope comprising a light source, an incident light modulation optical path, a fluorescence receiving optical path, and a CMOS imaging unit. The excitation light emitted by the light source is filtered by the incident light modulation optical path, and then directed toward the sample by a micro-filter. The fluorescence emitted by the sample after irradiation is collected by a gradient refractive index micro-optical lens, filtered by the fluorescence receiving optical path, and then directed to the CMOS imaging unit for imaging. The fluorescence is then transmitted to a host computer for processing via an interface circuit, and an intelligent image processing algorithm is stored in the host computer. The multi-channel miniature fluorescence in vivo imaging microscope designed and constructed by the present invention is miniaturized and wearable, providing a new technical means for the monitoring of circulating tumor cells in freely moving small animals.

Description

Technical Field

[0001] The present invention belongs to the technical field of infrared analyte detection, and in particular relates to a wearable micro-in vivo flow image cytometer. Background Art

[0002] Cancer is a major health threat to public health. According to data released by the World Health Organization's International Agency for Research on Cancer, 19.29 million new cancer cases were reported worldwide in 2020, of which 4.57 million were in China, accounting for 23.7% of the global total. In 2020, 9.66 million cancer deaths were reported worldwide, of which 3 million were in China, accounting for 30% of the total. As China has the world's largest population, the number of new cancer cases and deaths far exceeds that of any other country. According to data released by the China Cancer Center, in 2015, my country had approximately 3.929 million new cases of malignant tumors and 2.338 million deaths. On average, approximately 11,000 people are diagnosed with cancer each day, and approximately 6,000 die from cancer. Cancer morbidity and mortality rates increase with age. With the aging population, my country's healthcare system will continue to face pressure from cancer. Cancer severely impacts patients' quality of life and, as a costly disease, places a heavy burden on patients' families and the healthcare system.

[0003] As cancer research deepens, scientists are discovering that the majority of cancer deaths stem not from the primary tumor but from metastasis. Tumor metastasis is a crucial step in tumor proliferation and spread, particularly through hematogenous metastasis via the circulatory system. Tumor cells break away from the primary tumor site, break through the blood vessel walls, and then invade the blood vessels. These metastatic tumor cells are called circulating tumor cells (CTCs). As an emerging marker of tumor metastasis, the study of CTCs is gaining increasing attention.

[0004] Traditional circulating tumor cell detection technologies are usually based on in vitro liquid testing of patient blood samples.

[0005] Technology. The above-mentioned circulating tumor cell enrichment, separation and analysis technology is an in vitro detection technology. The limitation of in vitro detection technology is that it requires in vitro blood collection for detection. The in vitro blood collection process separates circulating tumor cells from the blood flow environment of the living animal, and changes in blood parameters and cell physiological environment may cause changes in circulating tumor cell parameters, resulting in inaccurate test results. In addition, the limited amount of blood collected leads to limited detection sensitivity, and it is difficult to draw blood frequently for dynamic monitoring.

[0006] Charles Lin's research team pioneered in vivo flow cytometry. Combining the strengths of traditional flow cytometry and intravital confocal microscopy, in vivo flow cytometry can be used for real-time, quantitative in vivo monitoring of circulating cells (including CTCs). Traditional in vivo microscopic imaging techniques enable in vivo observation of tumor-immune activities within the tumor microenvironment. Two-photon microscopy can be used to observe perivascular macrophages. Research results indicate that these perivascular macrophages increase vascular permeability and enhance tumor cell invasion. In T cell immunotherapy research, confocal microscopy is used to observe the destruction of solid tumors during T cell therapy. Spinning disk confocal microscopy is used to observe the adhesion of circulating tumor cells to hepatic sinusoids, a phenomenon that includes the promotion of neutrophils. Most of these reported methods focus on solid tumors or reticuloendothelial cells. Solid tumors are typically the primary site of tumor origin, while reticuloendothelial cells are often the site of metastasis. However, the interactions between tumor cells and immune cells in the bloodstream during tumor dissemination are less well understood. This is related to the lack of suitable in vivo high-speed and high-sensitivity imaging methods and data processing methods.

[0007] Previously reported in vivo image flow cytometry and computer vision-based in vivo flow cytometry systems have high temporal resolution, allowing them to image and count rapidly moving cells in the bloodstream. However, these methods are limited by the number of detection channels, meaning they can only monitor one type of cell at a time.

[0008] To address the above technical challenges, the project team applied multi-channel high-speed fluorescence imaging combined with intelligent image processing algorithms to achieve simultaneous tracking and spatial co-localization analysis of circulating tumor cells and dendritic cells in the blood circulation. In this study, an artificial neural network algorithm was introduced to identify blood vessels and cells in the image data. Through this method, the researchers observed the interaction between circulating tumor cells and dendritic cells, and further analyzed their movement speed in the blood vessels. When circulating tumor cells and dendritic cells come into contact with each other, they form cell clumps. These cell clumps move at a low speed in the blood vessels and appear to move along the blood vessel walls. This provides some new perspectives and evidence for the study of tumor metastasis. In addition, in vivo flow cytometry has the ability to quantitatively and dynamically monitor fluorescent substances in living organisms.

[0009] All of the above-mentioned in vivo imaging methods are limited by the anesthesia state of small animals and cannot be applied to in vivo detection of freely moving small animals. Therefore, the development of miniaturized and wearable in vivo imaging devices has become an important research direction. Summary of the Invention

[0010] To address the above problems, the present invention provides a wearable miniature in vivo flow cytometer to achieve real-time monitoring of circulating tumor cells in a freely moving small animal tumor model.

[0011] The technical solution adopted in the present invention is:

[0012] A wearable miniature in vivo flow cytometer includes a multi-channel miniature fluorescence in vivo imaging microscope, which includes a light source, an incident light modulation optical path, a fluorescence receiving optical path and a CMOS imaging unit. The excitation light emitted by the light source is filtered by the incident light modulation optical path, and then the filtered excitation light is directed toward the sample by a micro filter. The fluorescence emitted by the sample after being irradiated is collected by a gradient refractive index micro-optical lens, filtered by the fluorescence receiving optical path, and then directed to the CMOS imaging unit for imaging. The fluorescence is then transmitted to a host computer through an interface circuit for processing.

[0013] Preferably, the incident light modulation optical path includes an incident light micro-glued filter waveplate micro-lenticular mirror and an incident light micro-glued filter waveplate arranged in the same direction as the light source. The excitation light emitted by the light source passes through the incident light micro-glued filter waveplate micro-lenticular mirror and is filtered by the incident light micro-glued filter waveplate to allow light in the blue band and the red band to pass through. The microfilter guides the light in the blue band and the red band in the direction of the sample, and the sample emits fluorescence of different wavelengths after being irradiated.

[0014] Preferably, the wavelength range of the light in the blue band is 458.9nm-489.7nm; the wavelength range of the light in the red band is 623.1nm-646.9nm.

[0015] Preferably, the fluorescence receiving optical path includes an emission light micro-glued filter and a plano-convex mirror arranged in the same direction as the CMOS imaging unit. The light in the blue band and the red band is collected by a gradient refractive index micro-optical lens, filtered by the emission light micro-glued filter, and then directed to the CMOS imaging unit.

[0016] Preferably, the CMOS imaging unit includes

[0017] Green channel, used to detect fluorescence signals in the 497.0nm-532.6nm band;

[0018] The red channel is used to detect fluorescence signals in the 656.9nm-804.1nm band.

[0019] Preferably, the interface circuit is embedded with an image processor, which processes and analyzes the image data in real time and uploads the extracted characteristic information of the target detection object to a host computer.

[0020] Preferably, the image processor stores an intelligent image processing algorithm, and the process of processing image data using the intelligent image processing algorithm is as follows:

[0021] Step 1: Effectively identify and segment the blood vessel area in the image, and identify and remove the tissue area outside the blood vessel;

[0022] Step 2: Prepare training data and test data for circulating tumor cells, dendritic cells, and circulating tumor cell-dendritic cell clusters in the vascular area after manual labeling, and pass the training data and test data to the convolutional neural network for training respectively.

[0023] Preferably, in step 1, the process of effectively identifying and segmenting the blood vessel region in the image is as follows:

[0024] (1) Mark the region of interest (ROI) of the image by drawing a rectangular window in the vascular region, and extract all the image pixel values ​​in this ROI as the training data of the vascular region. Use the same method to extract another ROI in the vascular region, and extract all the image pixel values ​​in this ROI as the test data of the vascular region.

[0025] (2) Similarly, a region of interest is drawn in the surrounding tissue area, and all image pixel values ​​within this region of interest are extracted as the training data set of the surrounding tissue area. Another region of interest is extracted in the surrounding tissue area using the same method, and all image pixel values ​​within this region of interest are extracted as the test data of the surrounding tissue area.

[0026] (3) The training data and test data of the blood vessel area and the training data and test data of the surrounding tissue area are sequentially input into the blood vessel recognition neural network for training, and the tissue area outside the blood vessel is identified and removed.

[0027] Beneficial effects of the present invention:

[0028] The multi-channel, wearable, miniaturized fluorescence in vivo imaging microscope designed in this paper provides a novel technique for monitoring circulating tumor cells in freely moving small animals. This technique can be used to dynamically monitor the dynamic changes in circulating tumor cells under different autonomous motion states, studying the interactions between tumor hematogenous metastasis, autonomous behavior, and biological rhythms, providing a means for evaluating the efficacy of tumor immunotherapy and drug treatments. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 This is a system diagram of the multi-channel miniature fluorescence in vivo imaging microscope of the present invention;

[0031] Figure 2 Schematic diagram of coupling a miniature fluorescence intravital microscope with an animal model;

[0032] Figure 3 This is a flowchart of image processing and target recognition based on convolutional neural networks.

[0033] The reference numerals are as follows:

[0034] 1. Light source; 2. Micro-cemented filter for incident light and micro-lenticular mirror; 3. Micro-cemented filter for incident light; 4. Micro-filter; 5. Gradient-index micro-optical lens; 6. Micro-cemented filter for emitted light; 7. Plano-convex mirror; 8. CMOS imaging unit. DETAILED DESCRIPTION

[0035] To make the objectives, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0036] This embodiment provides a wearable micro-in vivo flow cytometer, such as Figure 1 As shown, it includes a multi-channel miniature fluorescence in vivo imaging microscope, which includes a light source 1, an incident light modulation optical path, a fluorescence receiving optical path and a CMOS imaging unit 8. The excitation light emitted by the light source 1 is filtered by the incident light modulation optical path, and the filtered excitation light is guided toward the sample by the micro filter 4. The fluorescence emitted by the sample after being irradiated is collected by the gradient refractive index micro-optical lens 5, filtered by the fluorescence receiving optical path, and then guided to the CMOS imaging unit 8 for imaging, and transmitted to the host computer through the interface circuit for processing.

[0037] In this embodiment, the light source 1 is a white LED lamp that provides excitation light within the visible light wavelength range.

[0038] Specifically, the incident light modulation optical path includes a micro-lenticular mirror 2 and a micro-lenticular filter 3, both coaxially arranged with the light source 1. Excitation light emitted by the light source 1 passes through the micro-lenticular mirror 2 and is filtered by the micro-lenticular filter 3, allowing light in the blue and red bands to pass through. The micro-filter 4 directs the blue and red light toward the sample, which emits fluorescence of different wavelengths after exposure. The wavelength range of the blue light is 458.9 nm to 489.7 nm, and the wavelength range of the red light is 623.1 nm to 646.9 nm.

[0039] The fluorescence receiving optical path includes an emission light micro-cemented filter 6 and a plano-convex mirror 7, coaxially arranged with a CMOS imaging unit 8. Light in the blue and red bands is collected by a gradient refractive index micro-optical lens 5, filtered by the emission light micro-cemented filter 6, and then directed to the CMOS imaging unit 8. The green channel of the CMOS imaging unit 8 is used to detect fluorescence signals in the 497.0nm-532.6nm band, while the red channel is used to detect fluorescence signals in the 656.9nm-804.1nm band. Dual-channel images are acquired synchronously at a rate of at least 30 frames per second to meet the frame rate requirements of real-time imaging.

[0040] The collected image data can be processed online or offline. One approach involves directly transmitting it to a host computer through an interface circuit and saving it in a video file format for subsequent offline processing and data analysis. Another approach involves integrating image processing functionality into the interface circuit, such as through a field-programmable gate array (FPGA) or graphics processing unit (GPU), to enable real-time processing and analysis of image data, transmitting the extracted target object feature information to the host computer.

[0041] The image processor stores an intelligent image processing algorithm. The process of processing image data using the intelligent image processing algorithm is as follows:

[0042] Step 1: Effectively identify and segment the blood vessel area in the image, and identify and remove the tissue area outside the blood vessel;

[0043] Step 2: Prepare training data and test data for circulating tumor cells, dendritic cells, and circulating tumor cell-dendritic cell clusters in the vascular area after manual labeling, and pass the training data and test data to the convolutional neural network for training, such as Figure 3 shown.

[0044] In step 1, the process of effectively identifying and segmenting the blood vessel region in the image is as follows:

[0045] (1) Mark the region of interest (ROI) of the image by drawing a rectangular window in the vascular region, and extract all the image pixel values ​​in this ROI as the training data of the vascular region. Use the same method to extract another ROI in the vascular region, and extract all the image pixel values ​​in this ROI as the test data of the vascular region.

[0046] (2) Similarly, a region of interest is drawn in the surrounding tissue area, and all image pixel values ​​within this region of interest are extracted as the training data set of the surrounding tissue area. Another region of interest is extracted in the surrounding tissue area using the same method, and all image pixel values ​​within this region of interest are extracted as the test data of the surrounding tissue area.

[0047] (3) The training data and test data of the blood vessel area and the training data and test data of the surrounding tissue area are sequentially input into the blood vessel recognition neural network for training, and the tissue area outside the blood vessel is identified and removed.

[0048] While focusing on the accuracy of image segmentation and object recognition, the algorithm design also emphasizes improving image processing speed to meet the needs of real-time, multi-channel, and massive data processing. The trained neural network algorithm can also be deployed on a host computer for offline data processing and analysis.

[0049] Animal experiments

[0050] Study on the dynamic changes of circulating tumor cells in small animals under different autonomous activity states

[0051] Experimental animals will be SD rats or BALB / c mice. These are commonly used experimental animal species in cancer research. Animals of appropriate size will be selected based on the final size of the micro-fluorescence in vivo imaging microscope. Four models are available for coupling the micro-fluorescence in vivo imaging microscope with experimental animals: the skull bone marrow model, the cranial window model, the visual window model, and the ear microvascular model. The skull bone marrow model is a minimally invasive animal imaging model that requires only a scalp incision. Direct imaging of the skull bone marrow region using the micro-fluorescence in vivo imaging device allows for observation of tumor cell migration within the superficial bone marrow region. The cranial window model is a commonly used imaging model in brain science research. The experiment involves partially destroying the skull tissue, allowing the micro-in vivo microscope to directly image blood vessels in the cerebral cortex and observe circulating tumor cells and metastatic lesions in the brain. The visual window model is a commonly used model for in vivo tumor imaging. By creating a visual window in the epidermis of the animal, the micro-in vivo microscope can observe blood vessels within the skin, enabling the study of tumor cell shedding and the formation of circulating tumor cells at the primary tumor site. The ear microvascular model is one of the commonly used models for in vivo flow cytometry. It is a non-invasive detection model. Since the skin of the mouse ear is thin and the blood vessels are relatively superficial, non-invasive circulating tumor cell monitoring can be achieved after hair removal. Figure 2 shown.

[0052] The 4T1 breast cancer model was established as the primary tumor model for this study. 4T1 breast cancer cells were labeled with green fluorescent protein (GFP). The absorption and emission center wavelengths of GFP are 488 nm and 510 nm, respectively, which match the green channel of the miniature fluorescence intravital microscope we constructed. GFP-labeled 4T1 breast cancer cells (4T1-GFP cells) were orthotopically implanted in SD rats or BALB / c mice. Live imaging flow cytometry was used to monitor circulating tumor cells in breast cancer in real time, recording the dynamics of circulating tumor cells under different autonomous activity states. For example, the number of circulating tumor cells present during feeding, resting, exercising, and sleeping states can be compared, as can differences in cell morphology (single cells versus cell clusters).

[0053] In animal experiments, a micro-in vivo flow cytometer was used to achieve high-speed imaging and feature extraction of circulating tumor cells and dendritic cells in the bloodstream. The researchers also investigated their spatial co-localization and its impact on overall cellular motility. Furthermore, in vivo flow cytometry enabled quantitative analysis of the clearance and aggregation of nanoparticles in the bloodstream.

[0054] The above description is only used to illustrate the technical solution of the present invention and is not intended to limit it. Other modifications or equivalent substitutions made to the technical solution of the present invention by ordinary technicians in this field should be included in the scope of the claims of the present invention as long as they do not depart from the spirit and scope of the technical solution of the present invention.

Claims

1. A wearable micro-in vivo flow cytometer, characterized in that: The invention comprises a multi-channel miniature fluorescence in vivo imaging microscope, which comprises a light source, an incident light modulation optical path, a fluorescence receiving optical path and a CMOS imaging unit. The excitation light emitted by the light source is filtered by the incident light modulation optical path, and then the filtered excitation light is directed toward the sample by a micro filter. The fluorescence emitted by the sample after being irradiated is collected by a gradient refractive index micro-optical lens, filtered by the fluorescence receiving optical path, and then directed to the CMOS imaging unit for imaging, and then transmitted to a host computer through an interface circuit for processing. The incident light modulation optical path includes a micro-lenticular mirror and an incident light micro-cemented filter waveplate arranged in a coaxial direction with the light source. The excitation light emitted by the light source passes through the micro-lenticular mirror and is filtered by the incident light micro-cemented filter waveplate to allow light in the blue and red bands to pass through. The micro-filter guides the blue and red bands toward the sample, and the sample emits fluorescence of different wavelengths after being irradiated. The wavelength range of the blue band light is 458.9nm-489.7nm; the wavelength range of the red band light is 623.1nm-646.9nm; The fluorescence receiving optical path includes an emission light micro-cemented filter and a plano-convex mirror arranged coaxially with the CMOS imaging unit. The light in the blue band and the red band is collected by a gradient refractive index micro-optical lens, filtered by the emission light micro-cemented filter, and then directed to the CMOS imaging unit. The CMOS imaging unit includes Green channel, used to detect fluorescence signals in the 497.0nm-532.6nm band; The red channel is used to detect fluorescence signals in the 656.9nm-804.1nm band.

2. A wearable micro in vivo flow cytometer according to claim 1, characterized in that: The interface circuit is embedded with an image processor, which processes and analyzes the image data in real time and uploads the extracted characteristic information of the target detection object to the host computer.

3. A wearable micro-in vivo flow cytometer according to claim 2, characterized in that: The image processor stores an intelligent image processing algorithm. The process of processing image data using the intelligent image processing algorithm is as follows: Step 1: Effectively identify and segment the blood vessel area in the image, and identify and remove the tissue area outside the blood vessel; Step 2: Prepare training data and test data for circulating tumor cells, dendritic cells, and circulating tumor cell-dendritic cell clusters in the vascular area after manual labeling, and pass the training data and test data to the convolutional neural network for training respectively.

4. A wearable micro-in vivo flow cytometer according to claim 3, characterized in that: In step 1, the process of effectively identifying and segmenting the blood vessel regions in the image is as follows: (1) Mark the region of interest (ROI) of the image by drawing a rectangular window in the vascular region, and extract all the image pixel values ​​in this ROI as the training data of the vascular region. Use the same method to extract another ROI in the vascular region, and extract all the image pixel values ​​in this ROI as the test data of the vascular region. (2) Similarly, a region of interest is drawn in the surrounding tissue area, and all image pixel values ​​within this region of interest are extracted as the training data set of the surrounding tissue area. Another region of interest is extracted in the surrounding tissue area using the same method, and all image pixel values ​​within this region of interest are extracted as the test data of the surrounding tissue area. (3) The training data and test data of the blood vessel area and the training data and test data of the surrounding tissue area are sequentially input into the blood vessel recognition neural network for training, and the tissue area outside the blood vessel is identified and removed.

Images