Fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting

Abstract

The invention provides a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting. The fluorescence-lensless dual-mode microscopic imaging system comprises a microscopic imaging subsystem, a fluorescence subsystem, a sample to be detected and a processor, the microscopic imaging subsystem is used for carrying out imaging processing on the to-be-detected sample by adopting an optical imaging principle to obtain a microscopic imaging result; the fluorescence subsystem is used for carrying out imaging treatment on the to-be-detected sample based on a fluorescence marker in the to-be-detected sample to obtain a fluorescence imaging result; and the processor is used for carrying out merging processing on the microscopic imaging result and the fluorescence imaging result, and carrying out classified counting processing by adopting a pre-trained classified counting network based on the merging processing result to obtain a classified counting result. According to the system and the method, the microimaging subsystem and the fluorescence subsystem are combined to realize marking imaging of cells with similar forms and different functions, and meanwhile, the pre-trained classification and counting network is combined to improve the accuracy and robustness of cell classification and counting.

Description

Technical Field

[0001] The present invention relates to the technical field of optical imaging, and in particular to a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting. Background Art

[0002] In the field of cancer prognosis, biopsy, as the gold standard, is crucial for early disease detection, accurate staging, and treatment plan development. For a long time, cell classification and enumeration during biopsy have relied primarily on traditional methods such as flow cytometry, impedance cytometry, and technologies based on cell chemical properties. These methods classify cells by analyzing physical properties such as size and impedance, or by employing chemical labeling. However, these traditional techniques have numerous limitations in practical application. Firstly, the complex structure of the associated equipment leads to high instrument costs, limiting their widespread use in primary care settings. Secondly, these methods may damage cells during operation, affecting their original state, and their throughput is limited, making them difficult to meet the needs of large-scale clinical sample analysis. Furthermore, while traditional optical microscopes can provide cell morphological information with micrometer or even nanometer resolution, they are limited by the numerical aperture and magnification of the objective lens. Obtaining high-resolution images often requires expensive high-resolution objectives, which not only increases equipment cost but also complicates the optical system, making the equipment bulky and occupying a large amount of laboratory space. In addition, traditional microscopic imaging usually requires staining of transparent cells to enhance contrast, but the invasiveness of the stains may change the physiological state of the cells, thereby adversely affecting subsequent cell analysis.

[0003] To overcome the many drawbacks of traditional cell analysis techniques, lensless microscopy has emerged. By computationally reconstructing cellular phase information, this technology overcomes the many limitations of conventional objective lenses on imaging systems, significantly reducing equipment costs and simplifying system design, offering a new approach and method for cell analysis. In recent years, lensless microscopy has made significant progress in the field of cell imaging and analysis, becoming a research hotspot and widely recognized for its potential to play a significant role in cell classification and counting applications.

[0004] Although lensless microscopy has certain advantages, it still faces many bottlenecks in the actual application of cell classification and counting. Specifically, first, single-modality lensless images are difficult to distinguish cells with similar morphology but different functions; second, existing classification and counting methods mostly rely on manual operation or simple threshold segmentation. Manual operation is inefficient, just like manually counting items one by one, and is highly subjective. Different people have different standards, making it difficult to ensure the consistency and accuracy of the results; simple threshold segmentation has poor robustness in complex scenes such as cell adhesion and noise interference, and is prone to misjudgment and missed judgments; third, when traditional semantic segmentation models process lensless images, due to low image resolution and blurred features, the segmentation accuracy is poor and mis-segmentation is prone to occur; fourth, cell counting algorithms are mostly based on connected domain analysis, which has limited ability to separate overlapping cells and lack a joint verification mechanism based on multimodal information, resulting in high statistical errors. These defects greatly limit the promotion and application of lensless microscopy in clinical high-throughput cell analysis. Summary of the Invention

[0005] In order to solve the above problems existing in the prior art, the present invention provides a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting.

[0006] The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0007] The present invention provides a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting, comprising: a microscopic imaging subsystem, a fluorescence subsystem, a sample to be tested, and a processor;

[0008] A microscopic imaging subsystem, which is used to perform imaging processing on the sample to be tested using the optical imaging principle to obtain microscopic imaging results;

[0009] A fluorescence subsystem is used to perform imaging processing on the sample to be tested based on the fluorescent marker in the sample to be tested to obtain a fluorescence imaging result;

[0010] The processor is used to merge the microscopic imaging results and the fluorescence imaging results, and based on the merged processing results, use a pre-trained classification counting network to perform classification counting processing to obtain classification counting results; the classification counting results include: the type of cell samples in the sample to be tested and the number of cells corresponding to each cell sample.

[0011] Optionally, the microscopic imaging subsystem includes: a semiconductor laser, a microaperture, a semi-transparent and semi-reflective mirror, a beam splitter, and a signal detector;

[0012] The semiconductor laser, the microhole, the semi-transparent and semi-reflective mirror, the beam splitter and the signal detector are arranged vertically in sequence, and the center points of the semiconductor laser, the microhole, the semi-transparent and semi-reflective mirror, the beam splitter and the signal detector are located on the same vertical axis;

[0013] The sample to be tested is placed on the upper surface of the signal detector, and the center of the sample to be tested coincides with the center of the signal detector;

[0014] semiconductor lasers, used to generate coherent light sources;

[0015] The microhole is used to filter out the central light beam of the coherent light source to obtain a first filtered light source;

[0016] A semi-transparent and semi-reflective mirror and a beam splitter are used to allow the laser light of a preset wavelength band in the first screening light source to pass through to form a second screening light source;

[0017] The signal detector is used to image the second screening light source irradiated on the sample to be tested to obtain a microscopic imaging result.

[0018] Optionally, the fluorescence subsystem includes: a fluorescence excitation light source, an objective lens, a sleeve, and an industrial camera;

[0019] The fluorescence excitation light source is arranged perpendicular to the microscopic imaging subsystem, and the excitation fluorescence emitted by the fluorescence excitation light source is directed directly toward the center point of the semi-transparent and semi-reflective mirror;

[0020] The objective lens, the sleeve and the industrial camera are arranged in sequence, and the center points of the objective lens, the sleeve and the industrial camera are on the same horizontal line as the center point of the beam splitter; the industrial camera is located on a side away from the beam splitter;

[0021] A fluorescence excitation light source, used to generate excitation fluorescence;

[0022] A semi-transparent and semi-reflective mirror is used to reflect the excited fluorescence to form reflected fluorescence;

[0023] The spectroscope is used to receive the excited fluorescence generated by the sample to be tested after being excited by the reflected fluorescence from a direction perpendicular to the sample to be tested, and reflect the excited fluorescence into the objective lens;

[0024] An objective lens is used to amplify the excited fluorescence to obtain amplified excited fluorescence;

[0025] A sleeve is used for collecting, filtering and amplifying the excitation fluorescence to form filtered excitation fluorescence;

[0026] Industrial cameras are used to image filtered excited fluorescence to form fluorescence imaging results.

[0027] Optionally, the sleeve is provided with a built-in lens and a filter at the front end of the sleeve;

[0028] The built-in lens at the front end of the tube is located on the side close to the objective lens.

[0029] Optionally, the semi-transparent and semi-reflective mirror is a long-wave pass dichroic mirror.

[0030] Optionally, the signal detector is a CMOS image sensor.

[0031] Optionally, the processor includes: a merging processing unit and a classification counting unit, the processor being configured to merge the microscopic imaging results and the fluorescence imaging results, and perform classification counting processing based on the merging processing result using a pre-trained classification counting network to obtain a classification counting result, including:

[0032] a merging processing unit, configured to call a preset simulation registration software, select multiple pairs of matching points from the microscopic imaging results and the fluorescence imaging results based on the cpselect function in the preset simulation registration software, calculate a change matrix between the multiple pairs of matching points based on the Fitgeotrans function, perform affine transformation processing on the microscopic imaging results using the change matrix and the imwarp function, and obtain a merging processing result;

[0033] The classification counting unit is used to input the result of the merging process into the pre-trained classification counting network for classification counting to obtain a classification counting result.

[0034] Optionally, the preset simulation registration software uses MATLAB software, and the Fitgeotrans function, the imwarp function, and the cpselect function are all toolbox functions in the MATLAB software.

[0035] Optionally, the pre-trained classification counting network uses ResNet18 as a reference network and is provided with multiple RefineNetBlocks, and the multiple RefineNetBlocks are connected to the reference network in a cascade structure.

[0036] Optionally, the training process of the pre-trained classification counting network includes:

[0037] Obtain a training sample set;

[0038] Use the training sample set and the preset loss function to train the initial classification counting network;

[0039] The initial classification counting network that meets the preset stopping conditions is used as the pre-trained classification counting network;

[0040] The preset stopping conditions include: the value of the preset loss function is continuously less than the loss threshold or the number of iterations is greater than the iteration threshold;

[0041] The initial classification counting network has the same structure as the pre-trained classification counting network.

[0042] The present invention provides a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting, comprising: a microscopic imaging subsystem, a fluorescence subsystem, a sample to be tested, and a processor; the microscopic imaging subsystem is used to perform imaging processing on the sample to be tested using an optical imaging principle to obtain a microscopic imaging result; the fluorescence subsystem is used to perform imaging processing on the sample to be tested based on fluorescent markers in the sample to be tested to obtain a fluorescence imaging result; the processor is used to merge the microscopic imaging result and the fluorescence imaging result, and perform classification counting processing using a pre-trained classification counting network based on the merged result to obtain a classification counting result; the classification counting result includes: the type of cell sample in the sample to be tested and the number of cells corresponding to each cell sample. In the present invention, by combining the microscopic imaging subsystem and the fluorescence subsystem, since the fluorescence subsystem uses fluorescent markers for imaging processing, labeled imaging of cells with similar morphology but different functions can be achieved. At the same time, through the separate design of the microscopic imaging subsystem and the fluorescence subsystem, synchronous and alternating acquisition of dual-modal data is achieved, thereby improving the accuracy of the fluorescence imaging results. In addition, the classification and counting processing is combined with the pre-trained classification and counting network and the fluorescence imaging results, thereby improving the accuracy and robustness of cell classification and counting.

[0043] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A schematic structural diagram of a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting provided by an embodiment of the present invention;

[0045] Figure 2 The structural diagram of the pre-trained classification counting network is shown as an example. DETAILED DESCRIPTION

[0046] The present invention will be further described in detail below with reference to specific examples, but the embodiments of the present invention are not limited thereto.

[0047] In order to improve the accuracy and robustness of cell classification and counting, an embodiment of the present invention provides a fluorescence-lensless dual-mode microscopy imaging system for cell classification and counting. Figure 1 A schematic diagram of the structure of a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, including:

[0048] Microscopic imaging subsystem 1, fluorescence subsystem 2, sample to be tested 7 and processor 15;

[0049] The microscopic imaging subsystem 1 is used to perform imaging processing on the sample 7 to be tested using the optical imaging principle to obtain microscopic imaging results;

[0050] The fluorescence subsystem 2 is used to perform imaging processing on the sample 7 to obtain a fluorescence imaging result based on the fluorescent marker in the sample 7;

[0051] The processor 15 is used to merge the microscopic imaging results and the fluorescence imaging results, and based on the merged processing results, use a pre-trained classification counting network to perform classification counting processing to obtain classification counting results; the classification counting results include: the type of cell samples in the sample to be tested 7 and the number of cells corresponding to each cell sample.

[0052] In this embodiment, the sample to be tested 7 may be a microorganism and pathogen detection sample, a disease prediction sample, a cell biology research sample, or an environmental and industrial sample, etc.

[0053] An embodiment of the present invention provides a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting. By combining a microscopic imaging subsystem 1 and a fluorescence subsystem 2, since the fluorescence subsystem 2 uses fluorescent markers for imaging processing, it can achieve labeled imaging of cells with similar morphology but different functions. At the same time, through the separate design of the microscopic imaging subsystem 1 and the fluorescence subsystem 2, synchronous and alternating acquisition of dual-modal data is achieved, thereby improving the accuracy of the fluorescence imaging results. In addition, the classification and counting processing is combined with a pre-trained classification and counting network and the fluorescence imaging results, thereby improving the accuracy and robustness of cell classification and counting.

[0054] Optionally, the microscopic imaging subsystem 1 includes: a semiconductor laser 3, a microaperture 4, a semi-transparent and semi-reflective mirror 5, a beam splitter 6 and a signal detector 8;

[0055] The semiconductor laser 3, the microhole 4, the semi-transparent and semi-reflective mirror 5, the beam splitter 6 and the signal detector 8 are arranged vertically in sequence, and the center points of the semiconductor laser 3, the microhole 4, the semi-transparent and semi-reflective mirror 5, the beam splitter 6 and the signal detector 8 are located on the same vertical axis;

[0056] The sample to be tested 7 is placed on the upper surface of the signal detector 8, and the center of the sample to be tested 7 coincides with the center of the signal detector 8;

[0057] Semiconductor laser 3, used to generate a coherent light source;

[0058] Microhole 4, used to filter out the central light beam of the coherent light source to obtain a first filtered light source;

[0059] The semi-transparent and semi-reflective mirror 5 and the beam splitter 6 are used to allow the laser light of a preset wavelength band in the first screening light source to pass through to form a second screening light source;

[0060] The signal detector 8 is used to image the second screening light source irradiated on the sample to be tested 7 to obtain a microscopic imaging result.

[0061] In addition, the semi-transparent and semi-reflective mirror 5 can be installed between the microhole 4 and the beam splitter 6 at a preset tilt angle. The specific tilt angle value can be flexibly set according to the actual usage scenario, and this embodiment does not limit this.

[0062] Optionally, the fluorescence subsystem 2 includes: a fluorescence excitation light source 9, an objective lens 10, a sleeve 11 and an industrial camera 12;

[0063] The fluorescence excitation light source 9 is arranged perpendicular to the microscopic imaging subsystem 1, and the excitation fluorescence emitted by the fluorescence excitation light source 9 is directed directly toward the center point of the semi-transparent and semi-reflective mirror 5;

[0064] The objective lens 10, the sleeve 11 and the industrial camera 12 are arranged in sequence, and the center points of the objective lens 10, the sleeve 11 and the industrial camera 12 are on the same horizontal line as the center point of the beam splitter 6; the industrial camera 12 is located on a side away from the beam splitter 6;

[0065] A fluorescence excitation light source 9, for generating excitation fluorescence;

[0066] A semi-transparent and semi-reflective mirror 5 is used to reflect the excited fluorescence to form reflected fluorescence;

[0067] The spectroscope 6 is used to receive the excitation fluorescence generated by the sample 7 being excited by the reflected fluorescence from a direction perpendicular to the sample 7, and reflect the excitation fluorescence into the objective lens 10;

[0068] The objective lens 10 is used to amplify the excited fluorescence to obtain amplified excited fluorescence;

[0069] Sleeve 11 is used to collect, filter and amplify the excitation fluorescence to form filtered excitation fluorescence;

[0070] The industrial camera 12 is used to image the filtered excited fluorescence to form a fluorescence imaging result.

[0071] Optionally, the sleeve 11 is provided with a lens 13 and a filter 14 built into the front end of the sleeve 11;

[0072] A lens 13 is built into the front end of the sleeve 11 and is located on a side close to the objective lens 10 .

[0073] In this embodiment, the wavelength of the output light of the fluorescent excitation light source 9 is 488 nm, and the maximum power is 50 mW, which is maintained at the maximum power during use; the magnification of the objective lens 10 can be 4, and the numerical aperture is 0.1; the industrial camera 12 can use a 1440×1080 pixel sensor with a pixel size of 3.45 μm; the filter 14 uses a bandpass filter, and the wavelength range of light allowed to pass by the bandpass filter is between 485 nm and 565 nm.

[0074] Optionally, the semi-transparent and semi-reflective mirror 5 is a long-wave-pass dichroic mirror.

[0075] Specifically, the starting wavelength of the long-wave pass dichroic mirror is 505 nm, and the 520 nm-800 nm band allows transmission, and the 380 nm-490 nm band allows reflection.

[0076] Optionally, the signal detector 8 is a CMOS image sensor.

[0077] Specifically, the signal detector 8 uses a complementary metal oxide semiconductor sensor with a pixel size of 1.34 μm to collect signals. The imaging field of the sensor can reach 25.52 mm. 2 The distance between the signal detector 8 and the sample to be tested 7 is usually controlled within 1 mm.

[0078] Since the fields of view, scales, and directions of the images obtained by the microscopic imaging subsystem 1 and the fluorescence subsystem 2 are not consistent, in order to make the cells under the two systems correspond one to one, the images need to be registered using a registration algorithm. The specific details can be referred to the image merging processing performed by the processor 15.

[0079] Optionally, the processor 15 includes: a merging processing unit and a classification counting unit. The processor 15 is configured to merge the microscopic imaging results and the fluorescence imaging results, and perform classification counting processing based on the merged processing result using a pre-trained classification counting network to obtain a classification counting result, including:

[0080] a merging processing unit, configured to call a preset simulation registration software, select multiple pairs of matching points from the microscopic imaging results and the fluorescence imaging results based on the cpselect function in the preset simulation registration software, calculate a change matrix between the multiple pairs of matching points based on the Fitgeotrans function, perform affine transformation processing on the microscopic imaging results using the change matrix and the imwarp function, and obtain a merging processing result;

[0081] The classification counting unit is used to input the result of the merging process into the pre-trained classification counting network for classification counting processing to obtain a classification counting result.

[0082] It should be noted that the multiple pairs of matching points are generally greater than or equal to 4 pairs of matching points. In addition, during the above affine transformation process, the missing parts of the image caused by the affine transformation process can be filled with zeros.

[0083] Optionally, the preset simulation registration software uses MATLAB software, and the Fitgeotrans function, the imwarp function, and the cpselect function are all toolbox functions in the MATLAB software.

[0084] Optionally, the pre-trained classification counting network uses ResNet18 as a reference network, and is provided with multiple RefineNetBlocks, which are connected to the reference network in a cascade structure.

[0085] Figure 2 The structural diagram of the pre-trained classification counting network is exemplarily shown. In this embodiment, four RefineNetBlocks are set for exemplification. Figure 2 In the pre-trained classification counting network, ResNet18 is used as the baseline network. The input is the result of the merge process, and the output is the classification counting result. The input of the first-level RefineNetBlock is the output features of ResNet18, and the output is connected to the next-level RefineNetBlock. The intermediate RefineNetBlock is connected not only to ResNet18 but also to the RefineNetBlocks of the previous and next two levels. The output of the final RefineNetBlock is the classification counting result.

[0086] Optionally, the training process of the pre-trained classification counting network includes:

[0087] Obtain a training sample set;

[0088] Use the training sample set and the preset loss function to train the initial classification counting network;

[0089] The initial classification counting network that meets the preset stopping conditions is used as the pre-trained classification counting network;

[0090] The preset stopping conditions include: the value of the preset loss function is continuously less than the loss threshold or the number of iterations is greater than the iteration threshold;

[0091] The initial classification counting network has the same structure as the pre-trained classification counting network.

[0092] The present invention provides a fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting, which has the following technical advantages:

[0093] 1. Existing single-modality imaging technologies (such as traditional lensless microscopy or fluorescence microscopy) face fundamental limitations in cell classification: lensless imaging lacks specific labeling capabilities and cannot distinguish between morphologically similar but functionally distinct cells (e.g., K150 and ECa109); while fluorescence imaging relies on costly objective lenses and can damage cells. This invention utilizes a fluorescence-lensless dual-modality fusion system to deeply integrate the chemical specificity of fluorescence images with the morphological information of lensless images, enabling precise cell classification.

[0094] 2. Existing techniques rely on manual threshold segmentation or simple connected domain analysis, which has a low degree of automation and is easily influenced by subjective factors. This paper proposes an improved RefineNet semantic segmentation model (a pre-trained classification counting network). By combining a chained residual pooling module (capturing a wide range of contextual information) with multi-resolution feature fusion (integrating multi-level features of ResNet), it significantly enhances the ability to analyze low-contrast lensless images.

[0095] 3. The present invention abandons the traditional high numerical aperture objective lens in hardware, adopts a 4x objective lens (reducing costs by 70%), and adopts a low-cost CMOS sensor (1.34μm pixel), reducing the system volume by more than 50%, thereby significantly reducing equipment costs and maintenance difficulty while maintaining high performance.

[0096] It should be noted that the terms "first," "second," and the like are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in sequences other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention.

[0097] In the description of this specification, the reference terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" mean that the specific features or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described can be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification.

[0098] Although the present invention is described herein in conjunction with various embodiments, in the process of implementing the claimed invention, those skilled in the art can understand and implement other variations of the above-mentioned disclosed embodiments by viewing the drawings and the disclosed content. In the description of the present invention, the word "comprising" does not exclude other components or steps, "one" or "an" does not exclude multiple situations, and the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, certain measures are recorded in different embodiments, but this does not mean that these measures cannot be combined to produce good results.

[0099] The above is a further detailed description of the present invention in conjunction with specific preferred embodiments, and the specific implementation of the present invention cannot be considered to be limited to these descriptions. For those skilled in the art of the present invention, several simple deductions or substitutions can be made without departing from the concept of the present invention, and all of these should be considered to fall within the scope of protection of the present invention.

Claims

1. A fluorescence-lensless dual-mode microscopy imaging system for cell classification and counting, characterized in that: include: Microscopic imaging subsystem, fluorescence subsystem, sample to be tested, and processor; The microscopic imaging subsystem is used to perform imaging processing on the sample to be tested using the optical imaging principle to obtain microscopic imaging results; The fluorescence subsystem is used to perform imaging processing on the sample to be tested based on the fluorescent marker in the sample to be tested to obtain a fluorescence imaging result; The processor is configured to combine the microscopic imaging result and the fluorescence imaging result, and perform classification counting processing based on the combined result using a pre-trained classification counting network to obtain a classification counting result; The classification and counting result includes: the types of cell samples in the sample to be tested and the number of cells corresponding to each cell sample.

2. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 1, characterized in that: The microscopic imaging subsystem includes: a semiconductor laser, a microhole, a semi-transparent and semi-reflective mirror, a spectroscope and a signal detector; The semiconductor laser, the microhole, the semi-transparent and semi-reflective mirror, the beam splitter and the signal detector are arranged vertically in sequence, and the center points of the semiconductor laser, the microhole, the semi-transparent and semi-reflective mirror, the beam splitter and the signal detector are located on the same vertical axis; The sample to be tested is placed on the upper surface of the signal detector, and the center of the sample to be tested coincides with the center of the signal detector; The semiconductor laser is used to generate a coherent light source; The microhole is used to filter out the central light beam of the coherent light source to obtain a first filtered light source; The semi-transparent and semi-reflective mirrors and the beam splitter are used to allow the laser light of a preset wavelength band in the first screening light source to pass through, thereby forming a second screening light source; The signal detector is used to image the second screening light source irradiated on the sample to be tested to obtain the microscopic imaging result.

3. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 2, characterized in that: The fluorescence subsystem includes: a fluorescence excitation light source, an objective lens, a sleeve and an industrial camera; The fluorescence excitation light source is arranged perpendicular to the microscopic imaging subsystem, and the excitation fluorescence emitted by the fluorescence excitation light source is directed toward the center point of the semi-transparent and semi-reflective mirror; The objective lens, the sleeve, and the industrial camera are arranged in sequence, and the center points of the objective lens, the sleeve, and the industrial camera are located on the same horizontal line as the center point of the beam splitter; the industrial camera is located on a side away from the beam splitter; The fluorescence excitation light source is used to generate the excitation fluorescence; The semi-transparent and semi-reflective mirror is used to reflect the excited fluorescence to form reflected fluorescence; The spectroscope is used to receive the excitation fluorescence generated by the sample to be tested after being excited by the reflected fluorescence from a direction perpendicular to the sample to be tested, and reflect the excitation fluorescence into the objective lens; The objective lens is used to amplify the excitation fluorescence to obtain amplified excitation fluorescence; The sleeve is used to collect and filter the amplified excitation fluorescence to form filtered excitation fluorescence; The industrial camera is used to image the filtered excitation fluorescence to form the fluorescence imaging result.

4. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 3, characterized in that: The sleeve is provided with a built-in lens and a filter at the front end of the sleeve; The built-in lens at the front end of the sleeve is located on a side close to the objective lens.

5. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 2, characterized in that: The semi-transparent and semi-reflective mirror is a long-wave-pass dichroic mirror.

6. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 2, characterized in that: The signal detector is a CMOS image sensor.

7. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 1, characterized in that: The processor includes: a merging processing unit and a classification counting unit, wherein the processor is configured to merge the microscopic imaging result and the fluorescence imaging result, and perform classification counting processing based on the result of the merging processing using a pre-trained classification counting network to obtain a classification counting result, including: The merging processing unit is configured to call a preset simulation registration software, select multiple pairs of matching points from the microscopic imaging results and the fluorescence imaging results based on the cpselect function in the preset simulation registration software, calculate a change matrix between the multiple pairs of matching points based on the Fitgeotrans function, and perform affine transformation processing on the microscopic imaging results using the change matrix and the imwarp function to obtain the merging processing result; The classification counting unit is used to input the result of the merging process into the pre-trained classification counting network to perform classification counting to obtain the classification counting result.

8. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 7, characterized in that: The preset simulation registration software adopts MATLAB software, and the Fitgeotrans function, the imwarp function and the cpselect function are all toolbox functions in MATLAB software.

9. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 7, characterized in that: The pre-trained classification counting network uses ResNet18 as a reference network and is provided with multiple RefineNetBlocks, and the multiple RefineNetBlocks are connected to the reference network in a cascade structure.

10. The fluorescence-lensless dual-mode microscopic imaging system for cell classification and counting according to claim 1, characterized in that: The training process of the pre-trained classification counting network includes: Obtain a training sample set; Training the initial classification counting network using the training sample set and a preset loss function; Using the initial classification counting network that meets the preset stopping condition as the pre-trained classification counting network; The preset stop condition includes: the value of the preset loss function is continuously less than the loss threshold or the number of iterations is greater than the iteration threshold; The initial classification counting network has the same structure as the pre-trained classification counting network.

Images