A method and apparatus for flow cytometry polarization imaging and classification of suspended particulate matter in water.

Abstract

This invention discloses a flow cytometry polarization imaging and classification method and apparatus for suspended particulate matter in water. The apparatus includes a flow cytometry injection system, an acquisition module, and a processing module. The acquisition module includes a detection chamber, a flash lamp, a polarizer, and an image-polarization detection module. The flow cytometry injection system is connected to the detection chamber and is used to inject the sample into the detection chamber. The flash lamp emits a uniformly intense unpolarized light beam. The polarizer adjusts the unpolarized light beam to a predetermined polarization state and maintains it constant during sampling. The image-polarization detection module acquires raw images of the sample in the detection chamber at a fixed frequency. The processing module processes the raw images to obtain Stokes images and simultaneously obtains the polarization and morphological information of suspended particulate matter in the sample based on the Stokes images. This invention can provide rapid, accurate, high-throughput, and detailed classification and detection of suspended particulate matter in water.

Description

Technical Field

[0001] This invention relates to the detection of suspended particulate matter in water, and in particular to a flow cytometry polarization imaging and classification method and apparatus for suspended particulate matter in water. Background Technology

[0002] Aquatic bodies (such as seawater) contain a large amount of suspended particulate matter, and achieving rapid, high-throughput, and detailed classification and detection of these particles is of great significance for ecological and environmental research. Microalgae, as a major type of suspended matter in aquatic bodies, play a vital ecological role and have intricate impacts on human survival and health. On the one hand, some types of microalgae can be used in aquaculture, chemicals, cosmetics, and nutritional products, bringing significant economic value to humanity; on the other hand, the explosive growth of some harmful algae can cause enormous losses to humankind. As a new type of pollutant, microplastics have become an important component of water environment safety monitoring in recent years; sediment information is a fundamental element of water color remote sensing, and the identification and concentration detection of sediment are crucial parts of water quality testing. Therefore, rapid, accurate, and high-throughput detailed classification of suspended particulate matter, especially microalgae, in aquatic bodies is of great importance.

[0003] Several methods based on different principles have been proposed for detecting suspended particulate matter in water, including the Coulter sensor, flow cytometry, and flow imaging. The Coulter sensor counts particles by detecting electrical pulses caused by differences in resistance between the particles and the electrolyte. However, since it only receives electrical pulse signals, its ability to distinguish particle types is very limited; it generally only provides information such as cell size for classification. Flow cytometry uses a sheath flow to guide particles one by one into a measurement chamber, where laser illumination is used and fluorescence is measured at the other end to classify cells. However, it is cumbersome and slow due to the need for hydrodynamic focusing. Flow imaging involves passing the sample through a detection chamber and photographing it with a microscope for classification. However, the morphological information obtained by this method, characterized by light intensity, is poor for classifying particles with low resolution and similar morphologies. Therefore, a rapid, accurate, high-throughput, and detailed classification method for measuring suspended particulate matter in water is still lacking.

[0004] It should be noted that the information disclosed in the background section above is only for understanding the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The main objective of this invention is to address the shortcomings of the aforementioned background technology by providing a flow polarization imaging and classification method and apparatus for suspended particulate matter in water.

[0006] To achieve the above objectives, this application adopts the following technical solution:

[0007] In a first aspect, a flow cytometry polarization imaging and classification device for suspended particulate matter in water is provided, comprising a flow cytometry inlet system, an acquisition module, and a processing module. The acquisition module includes a detection chamber, a flash lamp, a polarizer, and an image-polarization detection module. The flow cytometry inlet system is connected to the detection chamber and is used to inject the sample to be tested into the detection chamber. The flash lamp is used to emit a uniformly intense unpolarized light beam. The polarizer is used to adjust the unpolarized light beam emitted by the flash lamp to a predetermined polarization state and keep it constant during sampling. The image-polarization detection module is used to acquire raw images of the sample to be tested in the detection chamber at a fixed frequency. The processing module is used to process the raw images to obtain a Stokes image, and simultaneously obtain the polarization information and morphological information of suspended particulate matter in the sample based on the Stokes image.

[0008] Preferably, the image-polarization detection module includes an optical imaging system, a beam splitter, a first polarization camera, a quarter-wave plate, and a second polarization camera; the light emitted from the optical imaging system is split into two beams by the polarization-independent beam splitter, one beam directly enters the second polarization camera, and the other beam passes through the quarter-wave plate to convert circularly polarized light into linearly polarized light before entering the first polarization camera.

[0009] Preferably, the sampling frequencies and exposure times of the first polarization camera and the second polarization camera are equal to ensure that the first polarization camera and the second polarization camera sample synchronously.

[0010] Preferably, the flow-through liquid injection system includes an injection pump and a defoamer. The sample to be tested is injected through the injection pump, and the defoamer is connected to the injection pump to remove air bubbles from the sample pumped out by the injection pump, forming a more stable liquid flow. The defoamer is connected to the detection chamber.

[0011] Preferably, the detection cavity is a microchannel manufactured using soft lithography technology to achieve surface imaging.

[0012] Preferably, it also includes a waste liquid tank, which is connected to the detection chamber and is used to collect the sample discharged from the detection chamber after detection.

[0013] Secondly, a flow polarization imaging and classification method for suspended particulate matter in water is provided, characterized by the following steps: S1, injecting the sample to be tested into the acquisition module at a predetermined flow rate through a flow inlet system; S2, acquiring the original image of the sample to be tested through the acquisition module; S3, processing the original image to obtain a Stokes image through a processing module, and simultaneously obtaining the polarization information and morphological information of suspended particulate matter in the sample to be tested based on the Stokes image.

[0014] Preferably, the predetermined flow rate in step S1 is configured to match the timing of the original image acquisition.

[0015] Preferably, in step S2, a dual polarization camera method is used to acquire 8-channel raw images; in step S3, the 8-channel raw images are processed to obtain 4 images, which correspond to the four parameters of the Stokes vector, respectively, to reflect the polarization information and morphological information of suspended particles in the sample to be tested.

[0016] Preferably, in step S3, the polarization and morphological information of suspended particles in the sample to be tested are obtained simultaneously based on the Stokes image, including a target detection algorithm and a target classification algorithm. The target detection algorithm achieves accurate target detection by applying an energy gradient threshold and a detection area threshold, and removes particles adhering to the detection cavity wall of the acquisition module by comparing the target positions of previous and subsequent frames to avoid duplicate detection. The target classification algorithm uses a CNN-based ResNet 50 network structure and applies a sample mixing method to smooth the boundaries between different categories to accelerate the model's convergence speed. The four-channel polarization image of the target detected by the target detection algorithm is input into the target classification algorithm to achieve automated and precise target detection and classification.

[0017] Preferably, the target detection algorithm includes the following steps: (1) binarizing the obtained light intensity image and detecting the target, and calculating the number of pixels occupied by the target, the position and length and width of the smallest rectangle containing the target; (2) removing targets that do not meet the predetermined threshold according to the number of pixels occupied by the target; (3) extracting the light intensity image containing the target according to the position and length and width of the smallest rectangle containing the target, calculating the energy density of the light intensity image, and removing targets with energy density lower than the threshold; (4) calculating the root mean square error between the previous and next frames by comparing the position of the target, and removing targets with root mean square error lower than the threshold, thereby removing particles adhering to the detection cavity wall of the acquisition module to avoid repeated detection.

[0018] The advantages of the embodiments of the present invention are mainly reflected in the following aspects:

[0019] 1. This invention achieves high-throughput measurement through flow cytometry and surface imaging. By acquiring Stokes images, it simultaneously obtains the polarization and morphological information of particulate matter, enabling detailed classification of particulate matter. Compared to traditional measurement methods, this invention combines accuracy and efficiency, rapidly and with high throughput acquiring the morphological and polarization information of particulate matter, thus enabling rapid and accurate detailed classification and detection of particulate matter.

[0020] 2. As a novel water particulate matter classification and detection technology, this invention can not only quickly and accurately detect particulate matter, but also improve the interpretability of the detection results.

[0021] 3. In the preferred technical solution, under the condition of flow cytometry, it is crucial to simultaneously acquire image information and polarization information by acquiring Stokes images. Therefore, by designing an image-polarization detection module, it is possible to obtain the Stokes data corresponding to each pixel point through a single image capture. By simultaneously analyzing the morphological characteristics of particles and the internal structure and physicochemical properties represented by polarization information, detailed classification of particles can be achieved. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of a flow polarization imaging and classification device for suspended particulate matter in water, according to a preferred embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of the image-polarization detection module in the flow polarization imaging and classification device for suspended particulate matter in water according to a preferred embodiment of the present invention.

[0024] Figure 3a , 3b 3c and 3c are respectively the confusion matrix diagrams of the verification set obtained by specific embodiments of the present invention, Comparison 1, and Comparison 2. Detailed Implementation

[0025] The embodiments of the present invention will be described in detail below. It should be emphasized that the following description is merely exemplary and not intended to limit the scope and application of the present invention. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0026] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of the present invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0027] like Figure 1As shown, this embodiment of the invention provides a flow polarization imaging and classification device for suspended particulate matter in water, including a flow inlet system, a data acquisition module, and a processing module. The data acquisition module includes a detection chamber, a flash lamp, a polarizer, and an image-polarization detection module. The flow inlet system is connected to the detection chamber and is used to inject the sample to be tested into the detection chamber. The flash lamp is used to emit a uniformly intense unpolarized light beam. The polarizer is used to adjust the unpolarized light beam emitted by the flash lamp to a predetermined polarization state and keep it constant during sampling. The image-polarization detection module is used to acquire raw images of the sample to be tested in the detection chamber at a fixed frequency. The processing module is used to process the raw images to obtain a Stokes image, and simultaneously obtain the polarization information and morphological information of suspended particulate matter in the sample based on the Stokes image.

[0028] By using the image-polarization detection module in conjunction with other components, it is possible to simultaneously detect four Stokes images (which correspond to the four parameters of the Stokes vector) within the imaging area (i.e., the detection area), which can significantly increase the amount of information acquired without changing the detection speed.

[0029] In a preferred embodiment, the flash unit can flash according to a trigger or a given timing sequence to provide sufficient light intensity (e.g., using a 3W LED as the light source). Then, a polarizer adjusts the unpolarized beam emitted by the flash unit to a predetermined polarization state (preferably, a polarization state that minimizes system error is selected, for example, S = [1, 0.7186, 0.3449, 0.3819]). T The sampling frequency remains constant during the sampling process. The image-polarization detection module acquires raw images of the sample within the microchannel at a fixed frequency (which can be set by the user; for example, the maximum sampling frequency used in this invention is 7fps) (resulting in 8-channel raw images). Each acquisition is synchronized with the flash (i.e., acquisition is performed when the flash flashes). The exposure time of the image-polarization detection module can be shortened by adjusting the flash power. Preferably, the flash and the image-polarization detection module can be triggered simultaneously by commands from the host computer, or other triggering modes. Stokes images contain morphological information of particles, characterizing their macroscopic morphological features. Simultaneously, the Stokes vector corresponding to each pixel in the Stokes image can characterize the internal structure, physicochemical properties, and other features of the particles. Specifically, the acquired 8-channel images are processed to obtain 4 images, which are the four parameters i, q, u, and v of the Stokes vector. i represents the light intensity; this image is a typical photographic image reflecting morphological information. The values ​​of q, u, and v reflect polarization information.

[0030] See Figure 1In a preferred embodiment, the flow-through system includes a syringe pump and a defoamer. The sample to be tested is injected through the syringe pump, and the defoamer is connected to the syringe pump to remove air bubbles from the sample pumped out by the syringe pump, forming a more stable liquid flow. The defoamer is connected to the detection chamber. By using a syringe pump for injection, the flow rate can be controlled more precisely. After the sample is pumped out by the syringe pump, it enters the defoamer to remove tiny air bubbles in the water, forming a more stable liquid flow and preventing air bubbles from clogging the detection chamber.

[0031] See Figure 2 In a preferred embodiment, the image-polarization detection module includes an optical imaging system, a beam splitter, a first polarization camera 1, a quarter-wave plate, and a second polarization camera 2. Light emitted from the optical imaging system is split into two beams by the polarization-independent beam splitter. One beam directly enters the second polarization camera 2, while the other beam, after being converted from circularly polarized light to linearly polarized light by the quarter-wave plate, enters the first polarization camera 1. The images are then acquired using a dual-polarization camera method to obtain an 8-channel raw image. Preferably, the first polarization camera 1 and the second polarization camera 2 are equipped with pixel-level linear polarization analyzers to obtain the linear polarization state of the image. Preferably, the sampling frequencies and exposure times of the first polarization camera 1 and the second polarization camera 2 are equal to ensure synchronous sampling by the two polarization cameras.

[0032] Compared to traditional image-polarization detection module design methods that require multiple acquisitions to obtain Stokes polarization images, such as the dual-rotating waveplate method, this invention uses the aforementioned dual-polarization camera method to obtain Stokes images with a single sampling, significantly shortening the sampling time. This significant reduction in sampling time through the detection method makes the flow imaging of this invention possible.

[0033] In a preferred embodiment, the optical imaging system can be any optical imaging system required by practical needs, such as a microscopy system. For example, it can be added to a commercial microscope system. Figure 2 The beam splitter, first polarizing camera 1, quarter-wave plate, and second polarizing camera 2 are shown.

[0034] In a preferred embodiment, the depth of the detection cavity can be designed according to the depth of field of the optical imaging system used, and the width can be designed according to the field of view. For example, for a microscope system used as the optical imaging system, the depth and width of the detection cavity are generally on the order of micrometers. Because the size of the detection cavity is relatively small (the size of the detection cavity can be selected according to the magnification of the objective lens, matching the depth of field to ensure clear imaging; generally, a 4X objective lens can use a detection cavity with a depth of 100 μm), a flow-through system stably pumps the sample into the detection cavity. The detection cavity is preferably a microchannel manufactured using soft lithography technology, and the microchannel has a predetermined width (for example, for a microscope system used as the optical imaging system, the width of the microchannel is kept as equal to or slightly smaller than the field of view of the microscope as possible to ensure maximum detection throughput while avoiding missed detections), thus enabling surface imaging and improving detection speed.

[0035] See Figure 1 In a preferred embodiment, a waste liquid pool is further included, which is connected to the detection chamber and is used to collect the sample discharged from the detection chamber after detection.

[0036] In a preferred embodiment, the processing module is located in a host computer, where the original image is processed to obtain a Stokes image.

[0037] This invention also provides a flow cytometry polarization imaging and classification method for suspended particulate matter in water, comprising the following steps:

[0038] S1. The sample to be tested is injected into the acquisition module at a predetermined flow rate through a flow-through liquid injection system;

[0039] S2. Acquire the original image of the sample to be tested through the acquisition module;

[0040] S3. The original image is processed by the processing module to obtain a Stokes image, and the polarization information and morphological information of suspended particles in the sample under test are obtained simultaneously based on the Stokes image.

[0041] In a preferred embodiment, the predetermined flow rate in step S1 is configured to match the timing of the original image acquisition. Flow rate control is crucial for imaging quality and efficiency. If the flow rate is too fast, the area through which particles flow during the exposure time will be too long, resulting in a trailing phenomenon, which seriously affects the sampling quality and detection accuracy; while if the flow rate is too slow, particles appearing in the previous sampled image will not have moved out of the detection area in the next image, resulting in repeated sampling, which will also affect the detection accuracy and speed. This invention can obtain an optimal flow rate through design and calculation. Taking the use of a microscope system as an optical imaging system as an example, the specific evaluation method of the flow rate is shown in Equation (1).

[0042]

[0043] Among them, L F The microscope's field of view is 2.16 mm in this example; v represents the sample feed rate; fps represents the sampling frequency of the first and second polarization cameras (the sampling frequencies of the two polarization cameras are equal); s represents the cross-sectional area of ​​the detection cavity, which is 0.00085 cm² in this example. 2 Δt represents the exposure time of the first and second polarization cameras (the exposure times of the two polarization cameras are equal), which is 80 μs in this example; ΔL represents the distance the particles travel during one exposure, i.e., the tail length. In this example, the camera sampling frequency is set to 6.5 fps. Considering the accuracy of the syringe pump and the sampling frequency, the injection rate is set to 4.25 ml / h. At this rate, ΔL is 1.11 μm, which is acceptable for this example.

[0044] In a preferred embodiment, in step S2, a dual-polarization camera method is used to acquire an 8-channel raw image; in step S3, the 8-channel raw image is processed to obtain 4 images, each corresponding to one of the four parameters of the Stokes vector, to reflect the polarization and morphological information of suspended particulate matter in the sample to be tested.

[0045] In this invention, the detection data differs from conventional light intensity images; instead, it consists of Stokes vectors corresponding to each pixel. In a preferred embodiment, a target detection algorithm is designed to sensitively detect particulate matter while simultaneously analyzing morphological features and the physicochemical properties and internal structural features represented by polarization to achieve detailed classification. Specifically, in step S3, the polarization and morphological information of suspended particulate matter in the sample to be tested are simultaneously obtained based on the Stokes image, including a target detection algorithm and a target classification algorithm. The target detection algorithm achieves accurate target detection by applying energy gradient thresholds and detection area thresholds, and removes particulate matter adhering to the detection cavity wall of the acquisition module by comparing the target positions of previous and subsequent frames to avoid duplicate detection. The target classification algorithm uses a ResNet 50 network structure based on CNN and applies a sample mixing method to smooth the boundaries between different categories to accelerate the convergence speed of the model. The four-channel polarization image of the target detected by the target detection algorithm is input into the target classification algorithm to achieve automated and detailed target detection and classification.

[0046] In this example, preferably, the target detection algorithm includes the following steps: (1) binarizing the obtained light intensity image and detecting the target (i.e., particulate matter) and calculating the number of pixels occupied by the target, the position and length and width of the smallest rectangle containing the target; (2) removing targets that do not meet a predetermined threshold based on the number of pixels occupied by the target (for example, under the 4X objective lens in this example, a pixel count of 10 to 1000 can be selected as the predetermined threshold, and targets with a pixel count less than 10 or greater than 1000 can be removed); (3) determining the position and length and width of the smallest rectangle containing the target. The light intensity image containing the target is extracted and the energy density of the light intensity image is calculated. Targets with energy densities lower than the threshold (e.g., in this case, the threshold is 0.0025, so targets with energy densities lower than 0.0025 are removed) are eliminated (low imaging quality); (4) By comparing the positions of the targets in the previous and next frames, the root mean square error between the previous and next frames is calculated and targets with excessively small root mean square errors are eliminated (preferably, when the current target is in the same position as a specific target in the previous frame, the current target is eliminated) to remove particles adhering to the detection cavity wall of the acquisition module to avoid repeated detection.

[0047] The following describes specific embodiments of the present invention.

[0048] Figure 1 A schematic diagram of a flow polarization imaging and classification device for suspended particulate matter in water is shown. Figure 2 A schematic diagram of the image-polarization detection module in the device is shown. In this example, an optical transmission microscope is used as... Figure 2 The optical imaging system used a 4X objective lens as the detection objective. The sample injection rate was set to 4.25 ml / h, the sampling frequency to 6.5 fps, and the depth and width of the detection cavity to 100 μm and 850 μm, respectively. The CNN network was trained and classified using four-channel Stokes images, single-channel light intensity images (Control 1), and the polarization mean of four-channel images (Control 1), respectively. The resulting validation set confusion matrix is ​​shown in Figure 3. Class 0 represents *Gnaphalium tsao-ko*, Class 1 represents *Chaetoceros*, Class 2 represents *Oocystis*, Class 3 represents *Chlorella*, Class 4 represents PS microspheres (microplastics), and Class 5 represents SiO2 microspheres (sand). Figure 3 shows that both single-channel light intensity images (representing morphological information) and polarization mean classification methods (representing polarization information) have limitations. However, Stokes images, which simultaneously utilize polarization and morphological information, significantly improve classification accuracy. Therefore, this invention can achieve fine classification of particulate matter. Therefore, this invention utilizes flow cytometry to achieve rapid, high-throughput detection, and combines polarization images with polarization and morphological information to achieve accurate and detailed classification, thereby realizing rapid, accurate, and high-throughput detailed classification and detection of suspended particulate matter in water.

[0049] The above description provides a further detailed explanation of the present invention in conjunction with specific / preferred embodiments, and it should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various substitutions or modifications can be made to these described embodiments without departing from the concept of the present invention, and all such substitutions or modifications should be considered within the scope of protection of the present invention. In the description of this specification, the reference to terms such as "an embodiment," "some embodiments," "preferred embodiment," "example," "specific example," or "some examples," etc., indicates that the specific features, structures, materials, or characteristics described in connection with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification and the features of different embodiments or examples. Although the embodiments of the present invention and their advantages have been described in detail, it should be understood that various changes, substitutions, and modifications can be made herein without departing from the scope of protection of the patent application.

Claims

1. A flow cytometry polarization imaging and classification device for suspended particulate matter in water, characterized in that, It includes a flow-through liquid inlet system, a data acquisition module, and a processing module. The data acquisition module includes a detection chamber, a flash lamp, a polarizer, and an image polarization detection module. The flow-through liquid inlet system is connected to the detection chamber and is used to inject the sample to be tested into the detection chamber; The flash unit is used to emit a uniform, unpolarized light beam. The polarizer is used to adjust the unpolarized beam emitted by the flash lamp to a predetermined polarization state and keep it constant during sampling; The image-polarization detection module is used to acquire raw images of the sample to be tested in the detection cavity at a fixed frequency, and is synchronized with the flash lamp each time it acquires an image. The image-polarization detection module includes an optical imaging system, a beam splitter, a first polarization camera, a quarter-wave plate, and a second polarization camera. The light emitted from the optical imaging system is split into two beams by the polarization-independent beam splitter. One beam directly enters the second polarization camera, and the other beam passes through the quarter-wave plate to convert the circularly polarized light into linearly polarized light before entering the first polarization camera. The sampling frequencies and exposure times of the first polarization camera and the second polarization camera are equal to ensure that the first polarization camera and the second polarization camera sample synchronously. The processing module is used to process the original image to obtain a Stokes image, and simultaneously obtain the polarization and morphological information of suspended particles in the sample based on the Stokes image, including a target detection algorithm and a target classification algorithm. The target detection algorithm achieves accurate target detection by applying energy gradient thresholds and detection area thresholds, and removes particles adhering to the detection cavity wall of the acquisition module by comparing the target positions of previous and subsequent frames to avoid duplicate detection. The target classification algorithm uses a CNN-based ResNet 50 network structure and applies a sample mixing method to smooth the boundaries between different categories to accelerate model convergence. The four-channel polarization image of the target detected by the target detection algorithm is input into the target classification algorithm to achieve automated and precise target detection and classification.

2. The flow polarization imaging and classification device for suspended particulate matter in water as described in claim 1, characterized in that: The flow-through liquid injection system includes an injection pump and a defoamer. The sample to be tested is injected through the injection pump. The defoamer is connected to the injection pump and is used to remove air bubbles from the sample pumped out by the injection pump to form a more stable liquid flow. The defoamer is connected to the detection chamber.

3. The flow polarization imaging and classification device for suspended particulate matter in water as described in claim 1, characterized in that: The detection cavity is a microchannel manufactured using soft lithography technology to achieve surface imaging.

4. The flow polarization imaging and classification device for suspended particulate matter in water as described in claim 1, characterized in that: It also includes a waste liquid pool, which is connected to the detection chamber and is used to collect samples discharged from the detection chamber after testing.

5. A flow cytometry polarization imaging and classification method for suspended particulate matter in water, characterized in that, The flow cytometry polarization imaging and classification device for suspended particulate matter in water as described in any one of claims 1-4 includes the following steps: S1. The sample to be tested is injected into the acquisition module at a predetermined flow rate through a flow-through liquid injection system; S2. Acquire the original image of the sample to be tested through the acquisition module; S3. The original image is processed by the processing module to obtain a Stokes image, and the polarization information and morphological information of suspended particles in the sample under test are obtained simultaneously based on the Stokes image.

6. The flow cytometry polarization imaging and classification method for suspended particulate matter in water as described in claim 5, characterized in that, The predetermined flow rate in step S1 is configured to match the timing of the original image acquisition.

7. The flow cytometry polarization imaging and classification method for suspended particulate matter in water as described in claim 5, characterized in that, In step S2, a dual polarization camera method is used to acquire 8 channels of raw images; in step S3, the 8 channels of raw images are processed to obtain 4 images, which correspond to the four parameters of the Stokes vector, respectively, to reflect the polarization information and morphological information of suspended particles in the sample to be tested.

8. The flow cytometry polarization imaging and classification method for suspended particulate matter in water as described in claim 5, characterized in that, The target detection algorithm includes the following steps: (1) Binarize the obtained light intensity image, detect the target, and calculate the number of pixels occupied by the target, the position and length and width of the smallest rectangle containing the target; (2) Based on the number of pixels occupied by the target, remove targets that do not meet the predetermined threshold; (3) Extract the light intensity image containing the target based on the position and length and width of the smallest rectangle containing the target, calculate the energy density of the light intensity image, and remove targets with energy density below the threshold. (4) By comparing the positions of the targets in the previous and next frames, the root mean square error between the previous and next frames is calculated and targets with root mean square errors below the threshold are removed, thereby removing particles adhering to the detection cavity wall of the acquisition module to avoid repeated detection.

Images