Hydrogel microsphere sorting method, artificial organ preparation method, and system and medium

Abstract

A hydrogel microsphere sorting method, an artificial organ preparation method, and a system and a medium. The sorting method comprises the following steps: collecting a hydrogel microsphere image in a microfluidic chip; analyzing the hydrogel microsphere image by means of a neural network, so as to obtain an analysis result; and on the basis of the analysis result, sorting hydrogel microspheres encapsulating target cells, wherein the neural network comprises a feature extraction layer and a multi-scale joint output head, and the hydrogel microsphere image is subjected to convolution by means of the feature extraction layer and is then processed by the multi-scale joint output head, so as to obtain the analysis result. Hydrogel microspheres encapsulating target cells are accurately identified and sorted by using droplet microfluidic technology, and standardized artificial organs can be prepared on the basis of the sorted hydrogel microspheres, thereby significantly improving the functional maturity and reliability of the artificial organs.

Description

Hydrogel microsphere sorting methods, artificial organ preparation methods, systems and media Technical Field

[0001] This invention relates to the field of microfluidic cell sorting technology, specifically to a hydrogel microsphere sorting method, artificial organ preparation method, system, and medium. Background Technology

[0002] Droplet microfluidics has been widely used to prepare precisely sized and shaped hydrogel droplets, which can encapsulate drugs or cells to achieve sustained release. However, traditional methods suffer from low encapsulation efficiency and yield due to the discontinuous process during droplet formation, resulting in the loss of a large amount of unencapsulated drug and cells.

[0003] On the other hand, image analysis-based cell sorting systems suffer from model redundancy and data latency issues. Furthermore, the significant differences in the optical properties of hydrogel droplets made of different materials lead to inconsistent sorting system performance in identifying whether cells or drugs are encapsulated, making them prone to errors. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method for sorting hydrogel microspheres, a method for preparing artificial organs, a system, and a medium.

[0005] The first aspect of this invention provides a method for sorting hydrogel microspheres encapsulating target cells, comprising the following steps:

[0006] Images of hydrogel microspheres in a microfluidic chip were acquired.

[0007] The analysis results were obtained by analyzing the images of hydrogel microspheres using a neural network.

[0008] Based on the analysis results, hydrogel microspheres encapsulating target cells were sorted out.

[0009] The neural network includes a feature extraction layer and a multi-scale joint output head. The hydrogel microsphere image is convolved by the feature extraction layer and then processed by the multi-scale joint output head to obtain the analysis results.

[0010] The analysis results include the coordinates of the hydrogel microspheres in the image, as well as marking information on whether the hydrogel microspheres encapsulate target cells.

[0011] Furthermore, prior to the neural network's analysis of the hydrogel microsphere image, the following steps are included:

[0012] The hydrogel microsphere image is format-aligned to obtain a first preprocessed image;

[0013] The first preprocessed image is subjected to Gaussian filtering to obtain the second preprocessed image;

[0014] Edge enhancement is performed on the second preprocessed image to obtain the third preprocessed image;

[0015] The third preprocessed image is subjected to grayscale mapping and contrast stretching to obtain a preprocessed hydrogel microsphere image, which is then used for image analysis.

[0016] Furthermore, the feature extraction layer is formed by stacking multiple convolutional layers to extract multi-level feature information from the hydrogel microsphere image; the joint output head is used to fuse feature information at different levels, perform convolution operation on the fused feature map to obtain a prediction tensor, and convert the prediction tensor into an analysis result; the analysis result includes the coordinates of the hydrogel microspheres in the image, and the labeling information of whether the hydrogel microspheres encapsulate target cells.

[0017] Furthermore, the feature extraction layer includes four consecutively stacked depthwise separable convolutional units; each depthwise separable convolutional unit consists of a depthwise convolutional kernel, a pointwise convolutional kernel, and a nonlinear activation layer; the depthwise convolutional kernel is used to extract local grayscale changes and shape features of the input image pixel by pixel to obtain multiple single-channel feature maps; the pointwise convolutional kernel is used to perform cross-channel fusion of the multiple single-channel feature maps output by the depthwise convolutional kernel to obtain a fused feature map; the nonlinear activation layer is used to perform row-wise activation on the fused feature map.

[0018] In the four consecutively stacked depthwise separable convolutional units, the depthwise separable convolutional unit of the first layer takes the hydrogel microsphere image as input, is used to perform initial downsampling of the input image, and reduce the size of the input image to 1 / 2, converting the single-channel raw signal into 8-dimensional primary features;

[0019] In the four consecutively stacked depthwise separable convolutional units, the depthwise separable convolutional unit of the second layer takes the output image of the depthwise separable convolutional unit of the first layer as input, and is used to perform secondary downsampling on the input image, reducing the size of the input image to 1 / 4 of the original hydrogel microsphere image, and upgrading the 8-dimensional primary features to 16-dimensional intermediate features.

[0020] In the four consecutively stacked depthwise separable convolutional units, the depthwise separable convolutional unit of the third layer takes the output image of the depthwise separable convolutional unit of the second layer as input, and performs downsampling on the input image three times to reduce the size of the input image to 1 / 8 of the original hydrogel microsphere image, while maintaining 16-dimensional mid-layer features to focus on the outline and local details of the hydrogel microspheres.

[0021] In the four consecutively stacked depthwise separable convolutional units, the fourth layer depthwise separable convolutional unit takes the output image of the third layer depthwise separable convolutional unit as input, and performs downsampling on the input image four times, reducing the size of the input image to 1 / 16 of the original hydrogel microsphere image, and upscaling the 16-dimensional mid-layer features to 24-dimensional high-layer features; the output image of the fourth layer depthwise separable convolutional unit is passed to the joint output head as the output of the feature extraction layer.

[0022] Furthermore, the joint output head is used to perform dual-scale feature reconstruction on the output image of the feature extraction layer to obtain a first-size feature map and a second-size feature map; the first-size feature map and the second-size feature map are respectively used to analyze and identify hydrogel microspheres of different sizes; the first size is smaller than the second size;

[0023] The joint output head obtains a first-size feature map by performing bilinear interpolation on the output image; the joint output head obtains a second-size feature map by performing bilinear interpolation on the output image after performing a convolution.

[0024] The first and second size feature maps contain multiple candidate boxes that match the contours of the hydrogel microspheres; the predicted tensor is obtained by analyzing and identifying the candidate boxes.

[0025] Furthermore, the process of sorting the hydrogel microspheres encapsulating the target cells based on the analysis results specifically includes the following steps:

[0026] Based on the analysis results of the image analysis, generate electrode driving instructions;

[0027] The electrode driving command controls the electrode driving module in the microfluidic chip to generate an electric field according to a preset timing sequence, and the electric field separates the hydrogel microspheres containing target cells from those not containing target cells into different flow channels.

[0028] Another aspect of this invention discloses a method for preparing an artificial organ, comprising the following steps:

[0029] Prepare hydrogel microspheres encapsulating target cells in a microfluidic chip;

[0030] The above-described method for sorting hydrogel microspheres encapsulating target cells was used to sort the hydrogel microspheres encapsulating target cells.

[0031] The hydrogel microspheres containing target cells were purified and cultured to obtain artificial organs.

[0032] Furthermore, the preparation of hydrogel microspheres in the microfluidic chip specifically includes the following steps:

[0033] Prepare a buffer solution containing gel matrix components as the first aqueous phase;

[0034] Prepare a buffer solution containing target cells and a cross-linking agent as the second aqueous phase;

[0035] Preparation of microfluidic carrier oil phase and spacer oil phase;

[0036] The first aqueous phase, the second aqueous phase, the carrier oil phase, and the spacer oil phase are injected into the microfluidic chip at a preset flow rate, so that the first aqueous phase, the second aqueous phase, and the carrier oil phase undergo a cross-linking reaction in the microfluidic chip, thereby forming hydrogel microspheres in which the oil phase encapsulates the aqueous phase.

[0037] Another aspect of this invention discloses a droplet microfluidic cell sorting system for implementing the above-mentioned method for sorting hydrogel microspheres encapsulating target cells and / or the method for preparing artificial organs. The system is characterized by comprising a microfluidic chip module, an image acquisition module, a central processing unit module, and an electrode driving module; the image acquisition module and the electrode driving module are disposed within the microfluidic chip module, and the central processing unit module establishes communication connections with both the image acquisition module and the electrode driving module.

[0038] The microfluidic chip module is used for the preparation and sorting of hydrogel microspheres;

[0039] The image acquisition module is used to acquire images of the hydrogel microspheres prepared in the microfluidic chip module and transmit the acquired hydrogel microsphere images to the central processing unit module.

[0040] The central processing unit module is used to perform image analysis on the hydrogel microsphere images and generate electrode driving instructions based on the image analysis results, which are then sent to the electrode driving module.

[0041] The electrode driving module is used to generate an electric field in the microfluidic chip module according to the electrode driving command, and to separate the hydrogel microspheres containing target cells from those not containing target cells through the electric field.

[0042] In another aspect, the present invention discloses a computer-readable storage medium storing a program that is executed by a processor to implement the above-described method for sorting hydrogel microspheres encapsulating target cells and / or for preparing artificial organs.

[0043] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0044] The embodiments of the present invention have the following beneficial effects: The hydrogel microsphere sorting method, artificial organ preparation method, system, and medium provided by the present invention utilize droplet microfluidics to accurately identify and sort hydrogel microspheres encapsulating target cells, enabling the preparation of standardized artificial organs based on the sorted hydrogel microspheres, significantly improving the functional maturity and reliability of artificial organs. The embodiments of the present invention employ an ultra-lightweight neural network model with multiple depthwise separable convolutional feature extraction layers and a joint output head for image recognition of hydrogel microspheres, achieving high-speed and high-precision hydrogel microsphere sorting, greatly improving the efficiency of artificial organ preparation.

[0045] Additional aspects and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description or may be learned by practice of the invention. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 is a schematic flowchart of the steps of a method for sorting hydrogel microspheres encapsulating target cells according to the present invention.

[0048] Figure 2 is a schematic diagram of the microfluidic chip structure of the present invention;

[0049] Figure 3 is a schematic flowchart of the artificial organ preparation method of the present invention;

[0050] Figure 4 is a schematic diagram of the cell encapsulation effect of Embodiment 1, Embodiment 2, Comparative Example 1 and Comparative Example 2 of the present invention;

[0051] Figure 5 is a schematic diagram of the effect of fibrous actin from HUVEC cells of the present invention covering hydrogel microspheres.

[0052] Figure 6 is a schematic diagram of a computer-readable storage medium structure according to the present invention. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] Droplet microfluidics has been widely used to prepare hydrogel droplets with precise size and shape. The use of droplet microfluidics to encapsulate drugs and cells within hydrogel droplets for sustained-release drug preparation has also been reported. However, traditional droplet microfluidics, which encapsulate drugs or cells in droplets, typically involve simultaneously forming oil-in-water or water-in-oil droplets in an aqueous phase while pre-mixed cells and drugs in the oil or aqueous phase are encapsulated. Because the droplet formation process is not continuous, a significant amount of cells and drugs that do not remain in the aqueous or oil phase are not encapsulated, resulting in substantial losses (low encapsulation efficiency and yield).

[0055] On the other hand, image-based cell sorting systems suffer from model redundancy and data transmission latency issues. Furthermore, different materials produce hydrogel droplets with varying optical properties. Microfluidic sorting systems based on image analysis may exhibit inconsistent sorting performance for different types of hydrogel droplets. Microfluidic sorting systems also have limitations in identifying whether hydrogels of different materials encapsulate cells or drugs.

[0056] In view of this, embodiments of the present invention provide a method for sorting hydrogel microspheres encapsulating target cells, as shown in Figure 1, comprising the following steps:

[0057] S101. Acquire images of hydrogel microspheres in the microfluidic chip;

[0058] S102. Analyze the images of hydrogel microspheres using a neural network to obtain the analysis results;

[0059] S103. Based on the analysis results, sort out the hydrogel microspheres that encapsulate the target cells.

[0060] This invention aims to prepare and sort hydrogel microspheres encapsulating target cells using droplet microfluidic technology, and to prepare artificial organs based on the sorted and collected hydrogel microspheres encapsulating target cells. This invention optimizes the image algorithm of the droplet microfluidic sorting system, improving the sorting efficiency, accuracy, and enrichment efficiency of the hydrogel microspheres encapsulating target cells.

[0061] The implementation process of each step of this invention is described in detail below:

[0062] S101. Acquire images of hydrogel microspheres in the microfluidic chip.

[0063] As shown in Figure 2, in this embodiment of the invention, images of hydrogel microspheres are captured in the image acquisition area of ​​the microfluidic chip using a high-speed camera. The captured images of hydrogel microspheres can directly reflect the real-time state of the hydrogel microspheres in the microchannel.

[0064] In a preferred embodiment, before performing image analysis on the hydrogel microsphere images, preprocessing is required to obtain preprocessed images. Preprocessing includes steps such as format alignment, Gaussian filtering, edge enhancement, grayscale mapping, and contrast stretching.

[0065] In the preprocessing steps, the format conversion aims to convert the raw hydrogel microsphere images (such as RAW format or specifically encoded video streams) acquired by the camera into a common and easily processed image format (such as bitmap format). Next, a Gaussian filtering step is used to perform a weighted average on the image to remove random noise while preserving edge information. Then, edge enhancement is applied to the hydrogel image to highlight the contour edges of the hydrogel microspheres. The grayscale standardization step converts the edge-enhanced image into a standardized 8-bit grayscale image to reduce computational complexity and unify the data range. Finally, contrast stretching is performed by adjusting the image's grayscale dynamic range to enhance image contrast and make the distinction between the microspheres and the background more obvious. This preprocessing process in this embodiment maximizes the enhancement of the hydrogel microsphere features, resulting in better classification prediction results from the subsequent neural network output.

[0066] As a preferred embodiment, a SIMD (Single Instruction, Multiple Data) instruction set, such as Intel's AVX-512 instructions, can be used to simultaneously perform the dual tasks of "separate Gaussian filtering" and "edge enhancement" of the image. The instructions are accelerated using a compiler version of GCC 11 or later (compiler option -mavx512f). Then, what would normally be a 3×3 2D convolution of the image is decomposed into a 1×3 horizontal convolution (weights [0.25, 0.5, 0.25]) and a 3×1 vertical convolution (weights [0.25, 0.5, 0.25]) to denoise and smooth the image. Simultaneously, an edge detection algorithm (Sobel operator) is used to calculate the gradient at the edges of the hydrogel microspheres, amplifying the grayscale differences at the edges of the microspheres in the image, making the edges clearer and more prominent. Using the AVX-512 instructions _mm512_mul_ps and _mm512_add_ps, eight pixels can be processed in parallel at once. Tests showed that the total processing time for a 1920×1080 image after the above filtering and edge enhancement was ≤0.08ms.

[0067] On the other hand, the GPU's void quantizeEnhanceV2 kernel function can be used to quantize the image grayscale (8-bit → 6-bit, 0-255 → 0-63), reducing the data bit width; adaptive contrast stretching, based on the local grayscale histogram of the image, increases the grayscale difference between the cells inside the microspheres and the hydrogel matrix from 15-20 levels to 25-30 levels; secondary noise reduction is performed on the image using a 3×3 mean filter to smooth out small-range pixel value fluctuations. Each thread block processes a 32×32 pixel region in parallel, with a time ≤0.07ms.

[0068] As a preferred embodiment, the preprocessed image described above can employ GPU zero-copy memory technology. A memory region accessible to both the CPU and GPU can be allocated via cudaHostAlloc, and memory mapping is enabled by calling cudaSetDeviceFlags(cudaDeviceMapHost) to avoid permission issues. The CPU-filtered data is directly written to this region, allowing the GPU to start processing without waiting for data copying. Data transmission latency is ≤0.03ms, and the entire preprocessing process takes ≤0.18ms, meeting the requirements of this embodiment for high-speed sorting of hydrogel microspheres encapsulating target cells.

[0069] S102. Analyze the images of hydrogel microspheres using a neural network to obtain the analysis results.

[0070] In some embodiments, image analysis of hydrogel microsphere images is performed using a neural network.

[0071] The neural network used for image analysis in this invention is built based on the PyTorch 2.0 deep learning framework. First, pre-classified droplet images (hydrogel images encased in cells are one class, and those without cells are another) are used as samples for training to construct an image classification model. Then, the image classification model is compressed using the INT8 quantization tool and converted to INT8 format. Next, channel-level KL divergence calibration is performed to generate a quantized ONNX format classification model. Finally, the pre-trained backbone (feature extractor) of the ONNX format classification model is retained as the feature extraction part to construct the neural network.

[0072] The constructed neural network includes a feature extraction layer and a joint output head. The feature extraction layer is formed by stacking multiple convolutional layers and is used to extract multi-level feature information from hydrogel microsphere images. The joint output head is used to fuse feature information at different levels, perform convolution operations on the fused feature map to obtain a prediction tensor, and convert the prediction tensor into analysis results. The analysis results include the coordinates of the hydrogel microspheres in the image and the labeling information of whether the hydrogel microspheres encapsulate target cells.

[0073] In a preferred embodiment, the neural network consists of a backbone feature extraction layer and a joint output head as its core framework. It employs a "pooling-free downsampling" design, directly reducing the feature map size through convolution stride to avoid feature loss and inference delay caused by pooling operations. The final output is a dual-scale detection result, and the predicted structure integrates key information, eliminating the need for subsequent ROI (Region of Interest) extraction and NMS (Non-Maximum Suppression) post-processing steps.

[0074] For the Backbone feature extraction layer, it uses depthwise separable convolutional units as the basic module, and completes feature extraction and downsampling through four consecutively stacked layers. The core objective is to efficiently extract key features such as the shape and grayscale distribution of microspheres from the single-channel original image. Each depthwise separable convolutional unit consists of three cascaded sub-modules, which execute according to the logic of "feature extraction → channel fusion → activation constraint" to adapt to the feature processing requirements of single-channel input. First, spatial convolution operation is performed on the single-channel input through 3×3 depthwise convolution to extract local grayscale changes and shape features pixel by pixel. Then, cross-channel fusion is performed on the single-channel feature map output by 1×1 point convolution to convert the number of channels of the feature map into the target output dimension of the current layer, realizing feature condensation and dimensionality control. Finally, the ReLU6 activation function is used to non-linearly activate the output of the 1×1 convolution, forcibly limiting the feature value range to the [0,6] interval. This constraint can effectively reduce numerical overflow and accuracy loss in the subsequent model quantization process (such as INT8 quantization), especially adapting to the numerical distribution characteristics of single-channel grayscale images.

[0075] In this embodiment of the invention, four depthwise separable convolutional units are stacked sequentially, with the parameters of each layer precisely matching the progressive extraction requirements of single-channel features using a downsampling strategy, as detailed below:

[0076] 1) Layer 1: Input channel number = 1 (single-channel original image), output channel number = 8, convolution stride = 2, padding = 1 (ensuring that the feature map size is reduced proportionally to the stride after convolution). This layer completes the first downsampling, reducing the original image size to 1 / 2, and at the same time converting the single-channel original signal into 8-dimensional primary features.

[0077] 2) Layer 2: Input channels = 8 (output from the upper layer), output channels = 16, stride = 2, padding = 1. After secondary downsampling, the size is 1 / 4 of the original. The 8-dimensional features are upscaled to 16-dimensional through 1×1 convolution, enhancing the feature representation capability.

[0078] 3) Layer 3: Input channels = 16, Output channels = 16, Step size = 2, Padding = 1. After three downsampling operations, the size is 1 / 8 of the original, while the number of channels remains at 16 dimensions. This deepens and refines the mid-layer features, focusing on the contours and local details of the microspheres.

[0079] 4) Layer 4: Input channels = 16, output channels = 24, stride = 2, padding = 1. After four downsampling passes, the size is 1 / 16 of the original, the number of channels is increased to 24 dimensions, and the output is a high-dimensional feature map rich in global and local features of the microspheres.

[0080] For the joint output head, it constructs a dual-scale detection branch based on the 16×24×16 feature map output by the Backbone, which is adapted to microspheres of different sizes, and integrates candidate boxes and probability prediction.

[0081] To address the grayscale detail dependency of single-channel features, the joint output head employs a combination strategy of "interpolation + transposed convolution" to reconstruct dual-scale feature maps, ensuring feature continuity and spatial resolution.

[0082] For a 32×32 scale feature map: the 16×16 feature map output by Backbone is upsampled to 32×32 using bilinear interpolation. Bilinear interpolation can smoothly transition based on the gray values ​​of neighboring pixels, avoiding gray-level discontinuities in single-channel features during scaling, and better preserving the edge features of small and medium-sized microspheres.

[0083] For the 64×64 scale feature map, a "two-step upsampling" strategy is adopted: the first step is to upsample the 16×16 feature map to 32×32 through a 2×2 transposed convolution (stride=2, padding=0, number of output channels=24). The transposed convolution reconstructs feature details by learning weights, which is closer to the gray-scale distribution of a single-channel microsphere than simple interpolation. The second step is to enlarge the 32×32 feature map to 64×64 through bilinear interpolation, which takes into account both the spatial positioning accuracy and feature integrity of large-sized microspheres.

[0084] On the other hand, the joint output head, based on the circular shape of the microspheres and the single-channel detection requirements, optimizes the number of candidate boxes and output dimensions, reducing redundant calculations while improving prediction accuracy:

[0085] 1) Candidate box parameters: Only 2 candidate boxes are generated for each grid at both scales, with aspect ratios set to 0.95 and 1.05 respectively. This aspect ratio is close to 1, which accurately matches the circular outline of the microsphere and avoids invalid candidate boxes caused by large aspect ratio differences. It is especially suitable for the "quasi-circular grayscale region" feature of microspheres in single-channel images.

[0086] 2) Output Dimension and Mapping: The number of channels in the dual-scale feature map is adjusted to 12 dimensions (2 candidate boxes × 6 predicted values) through 1×1 convolution. The 6 predicted values ​​are mapped in a fixed order as follows:

[0087] 1: Microsphere center coordinates (x, y): Based on the offset of the grid coordinates, normalized representation is used (value range [0, 1]) to adapt to different input sizes.

[0088] 2: Microsphere width and height (w, h): Normalized values ​​relative to the feature map size, combined with aspect ratio constraints, to ensure that the predicted shape conforms to the circular feature.

[0089] 3: Probability of containing cells (p_cell): The candidate box contains the probability value of microspheres containing cells (value range [0,1]).

[0090] 4: Empty packet probability (p_empty): The probability value of the candidate box containing an empty microsphere (range [0,1]).

[0091] 3) Output format adjustment: The output format is changed from "batch × 12 × height × width" to "batch × height × width × 12" by dimensional rearrangement, so that each spatial location (grid) directly corresponds to the complete prediction information of two candidate boxes. There is no need to extract ROI or perform NMS. The final analysis result is directly output. The format of the final analysis result is (x, y, w, h, p_cell, p_empty).

[0092] The aforementioned microsphere monitoring model achieves ultra-lightweight deployment, featuring four depthwise separable convolutional backbone layers (backbone network feature extractor layers) and a two-scale joint output head, with ≤480,000 parameters. Each processing unit in the backbone layer performs a 3×3 depthwise convolution, a 1×1 pointwise convolution, and a ReLU6 activation function. The kernel counts for the four convolutional layers are 8, 16, 16, and 24 respectively, with a stride of 2 to generate a "feature map." The pooling-free architecture avoids feature loss and latency.

[0093] S103. Based on the analysis results, sort out the hydrogel microspheres that encapsulate the target cells.

[0094] In some embodiments, sorting can be achieved by driving an electric field based on the analysis results, specifically including the following steps:

[0095] S103-1. Generate electrode driving instructions based on the analysis results of image analysis;

[0096] S103-2. By using electrode driving commands, the electrode driving module in the microfluidic chip is controlled to generate an electric field according to a preset timing sequence, and the hydrogel microspheres containing target cells and those not containing target cells are separated into different flow channels by the electric field.

[0097] The electrode driving commands generated in this embodiment of the invention include parameters such as voltage magnitude, pulse timing, and duration. Once the neural network identifies a microsphere, the control unit predicts the precise time it will arrive at the electric field sorting zone based on the microsphere's coordinates in the current and previous frames, combined with the known fluid flow rate.

[0098] When the hydrogel microspheres reach the electric field sorting zone, the living cells encapsulated within the microspheres carry a net charge. The entire microsphere is deflected by dielectrophoretic forces or other electric forces within the electric field, thus achieving the sorting of the hydrogel microspheres. In this embodiment of the invention, contactless high-speed sorting of hydrogel microspheres is achieved by applying an electric field.

[0099] The second embodiment of the present invention provides a method for preparing artificial organs based on image recognition, as shown in Figure 3, including the following steps:

[0100] S201. Prepare hydrogel microspheres encapsulating target cells in a microfluidic chip;

[0101] S202. The hydrogel microspheres encapsulating target cells are sorted using the sorting method of the first embodiment.

[0102] S203. The hydrogel microspheres containing target cells are processed and cultured to obtain artificial organs.

[0103] This invention aims to fabricate artificial organs using droplet microfluidic technology, while optimizing the image algorithm of the droplet microfluidic sorting system to improve the sorting efficiency, accuracy, and enrichment efficiency of artificial organs. The organ fabricated in this embodiment is a pancreas, but the fabrication method provided by this invention can be used not only for the pancreas but also for other organs such as the liver, kidney, lung, and heart.

[0104] The implementation process of each step in the method of this invention is described in detail below:

[0105] S201. Prepare hydrogel microspheres encapsulating target cells in a microfluidic chip.

[0106] The microfluidic chip has aqueous microchannels and oil microchannels. The aqueous microchannels flow through a buffer solution containing cells, cross-linking agents and a gel matrix, while the oil microchannels flow through a microfluidic-specific oil containing surfactants. The aqueous and oil microchannels meet, and the two liquid phases form water-in-oil droplets. Then, as the cross-linking reaction proceeds, the droplets form hydrogel microspheres, and the target cells are encapsulated within the droplets / microspheres.

[0107] In some embodiments, as shown in FIG2, the microfluidic chip is injected into the microchannel chip by an injection device, wherein a first aqueous phase, a second aqueous phase and an oil phase are injected into the microchannel chip, the first aqueous phase is a buffer solution containing a gel matrix component; the second aqueous phase is a buffer solution containing target cells and a cross-linking agent; and the oil phase is a microfluidic-specific oil containing a surfactant, used to emulsify and form water-in-oil droplets.

[0108] In a preferred embodiment, if sodium alginate and human α-thrombin are used as the gel matrix, and human fibrinogen is selected as the cross-linking agent, the prepared hydrogel microspheres exhibit appropriate swelling and degradation properties. The cells required for the artificial organ can be pancreatic β-cells, etc. Novec 7500 oil or similar can be used for the oil phase, and HEPES buffer can be used for the aqueous buffer phase.

[0109] As a preferred embodiment, the oil phase flow rate is typically set to twice that of the water phase to ensure stable droplet formation.

[0110] S202. The hydrogel microspheres encapsulating target cells are sorted using the sorting method of the first embodiment.

[0111] In step S202, the ultra-lightweight neural network model with multiple depthwise separable convolutional feature extraction layers and a joint output head in the first embodiment is used to perform image recognition of hydrogel microspheres, achieving high-speed and high-precision sorting of hydrogel microspheres. A detailed description of step S202 can be found in the first embodiment and will not be repeated here.

[0112] S203. Purify the hydrogel microspheres containing the target cells.

[0113] In step S203, the hydrogel microspheres encapsulating the target cells are purified, specifically including the following steps:

[0114] S203-1. Place the hydrogel microspheres containing the target cells into a centrifuge tube;

[0115] S203-2. Add a demulsifier to the centrifuge tube to remove the oil phase of the hydrogel microspheres;

[0116] S203-3. Add buffer solution to the centrifuge tube to allow the hydrogel microspheres to swell and disperse;

[0117] S203-4. Use hexane to centrifuge and separate the hydrogel microspheres into layers to obtain purified hydrogel microspheres;

[0118] S203-5. Artificial organs were obtained by purifying hydrogel microspheres containing target cells.

[0119] Taking the preparation of the pancreas as an example, after collecting hydrogel microspheres encapsulating target cells, excess oil phase is first removed. Then, 20 m / m% perfluorooctanoic acid (Sigma) is added, followed by 50 v / v% Novec 7500 oil to destabilize the hydrogel microspheres. HEPES buffer is then added to swell and disperse the microspheres. The oil phase is then extracted multiple times with hexane (like dissolves like), and the purified hydrogel microspheres are obtained by centrifugation and layer separation. Finally, the hydrogel microspheres are co-cultured with HUVEC cells to obtain an artificial pancreas that can be injected into the body.

[0120] The method of this invention is applied to a droplet microfluidic cell sorting system, which includes a microfluidic chip module, an image acquisition module, a central processing unit module, and an electrode driving module. The image acquisition module and the electrode driving module are disposed in the microfluidic chip module, and the central processing unit module establishes communication connections with the image acquisition module and the electrode driving module, respectively.

[0121] The microfluidic chip module is used for the preparation and sorting of hydrogel microspheres;

[0122] The image acquisition module is used to acquire images of the hydrogel microspheres prepared in the microfluidic chip module and transmit the acquired hydrogel microsphere images to the central processing unit module.

[0123] The central processing unit module is used to perform image analysis on the hydrogel microsphere images and generate electrode driving instructions based on the image analysis results, which are then sent to the electrode driving module.

[0124] The electrode driving module is used to generate an electric field in the microfluidic chip module according to the electrode driving command, and to separate the hydrogel microspheres with target cells and those without target cells by means of the electric field.

[0125] In a preferred embodiment, the microfluidic chip module includes a liquid injection channel, a droplet formation region, a main channel, an image acquisition region, an electric field sorting region, and an outlet channel connected in sequence.

[0126] The liquid injection channel includes a first aqueous phase channel, a second aqueous phase channel, and an oil phase channel, which are used to inject the first aqueous phase, the second aqueous phase, and the oil phase, respectively.

[0127] The first aqueous phase channel, the second aqueous phase channel, and the oil phase channel converge in the droplet formation region, where water-in-oil droplets are formed.

[0128] The gel matrix in the water-in-oil droplet reacts with the crosslinking agent in the main channel, causing the water-in-oil droplet to crosslink and solidify into hydrogel microspheres;

[0129] The image acquisition module acquires images of the hydrogel microspheres in the image acquisition area;

[0130] The electrode driving module applies an electric field to the electric field sorting area to control the flow of hydrogel microspheres to different outlet channels;

[0131] The outlet channel includes a screening channel and a waste liquid channel. The screening channel is used to collect hydrogel microspheres coated with target cells, while the waste liquid channel is used to discharge hydrogel microspheres that are not coated with target cells.

[0132] The following are examples and comparative illustrations of the image recognition-based artificial organ preparation method of this invention:

[0133] Example 1:

[0134] The preparation method of the artificial pancreas is as follows (aseptic operation):

[0135] Prepare aqueous phase 1: HEPES buffer containing 1 m / m% sodium alginate and 0.5 U / mL human α-thrombin;

[0136] Preparation of aqueous phase 2: containing pancreatic β cells (concentration of 1.25 × 10⁻⁶). 6 HEPES buffer containing 1 mg / mL of human fibrinogen;

[0137] Oil phase: contains 0.5 m / m% 3M Novec 7500 electronic fluorinated liquid;

[0138] S1: Aqueous phase 1, aqueous phase 2, and oil phase are injected into the microchannel chip of the microfluidic chip module via the injection device module. The flow rate of the carrier oil phase is 200 L / h, the flow rate of the spacer oil phase is 50 L / h, and the flow rates of aqueous phase 1 and aqueous phase 2 are 100 L / h.

[0139] S2: The high-speed camera of the image acquisition module acquires image information in the chip image acquisition area at a rate of 2000 frames per second, records the formation of hydrogel microspheres, and transmits the image information to the central processing module.

[0140] S3: The central processing unit module reads the image information captured by the camera and performs preprocessing on the image information, such as desaturation, noise reduction and smoothing (Gaussian filtering), Sobel edge enhancement, gray value quantization, contrast stretching and noise suppression, to obtain a preprocessed image;

[0141] S4: The central processing unit module's neural network model determines whether hydrogel microspheres encapsulate target cells based on the analysis of preprocessed images;

[0142] S5: Based on the determination of S4, the central processing unit module control unit controls the electrode drive module to drive the microchannel chip electrode switch deflection electric field, thereby realizing the phase separation of hydrogel microspheres encapsulating target cells and unencapsulated hydrogel microspheres.

[0143] When it is determined that the hydrogel droplet encapsulates the target cell, the deflection electric field is turned on, and the microspheres are sieved into the sieving channel under the action of the electric field force; when it is determined that the hydrogel droplet does not encapsulate the target cell, the deflection electric field is turned off, and the microspheres flow away with the direction of the main liquid flow.

[0144] S6: Collect hydrogel microspheres encapsulating target cells. First, remove excess oil phase from the hydrogel microspheres. Then, add 20 m / m% perfluorooctanoic acid (Sigma), followed by 50 v / v% Novec 7500 oil to destabilize the hydrogel microspheres. Next, add HEPES buffer to swell and disperse the hydrogel microspheres. Then, extract the oil phase multiple times with hexane (extraction principle), and centrifuge to separate the layers and obtain purified hydrogel microspheres.

[0145] S7: Artificial pancreas was obtained by co-culturing hydrogel microspheres with HUVEC cells.

[0146] Example 2

[0147] The preparation method of the artificial pancreas in Example 2 is basically the same as that in Example 1, except that the concentration of pancreatic β cells in aqueous phase 2 is different.

[0148] In this embodiment, aqueous phase 2 contains pancreatic β cells (concentration of 1×10⁻⁶). 5 HEPES buffer (1 mg / mL of human fibrinogen) and 1 mg / mL of human fibrinogen.

[0149] Comparative Example 1:

[0150] The preparation method of the artificial pancreas in Comparative Example 1 is basically the same as that in Example 1, except that the electrodes are not turned on and the hydrogel microspheres encapsulating the target cells and the unencapsulated hydrogel microspheres are not separated.

[0151] Comparative Example 2:

[0152] The preparation method of the artificial pancreas in Comparative Example 2 is basically the same as that in Example 2, except that the electrodes are not turned on and the hydrogel microspheres encapsulating the target cells and the unencapsulated hydrogel microspheres are not separated.

[0153] Example of detection:

[0154] (1) Enrichment rate / encapsulation rate: The encapsulation rate of the artificial pancreas (the proportion of the number of target cell microspheres encapsulated in the obtained microspheres) was calculated by visual inspection.

[0155] Results: The encapsulation rates of Examples 1, 2, Comparative Examples 1, and 2 are shown in Figure 4. Figure 4(a) is a schematic diagram of Comparative Example 1, Figure 4(b) is a schematic diagram of Example 1, Figure 4(c) is a schematic diagram of Comparative Example 2, and Figure 4(d) is a schematic diagram of Example 2. According to Comparative Examples 1 and 2, a higher cell concentration ensures that when the oil and liquid phases form droplets, some cells are precisely encapsulated within them. Based on the results of Examples 1 and 2, a higher cell concentration results in a higher encapsulation rate.

[0156] (2) Cell viability: Cells in the microspheres of step S6 in Example 1 were cultured in RPMI 1640 medium for 48 hours. RPMI 1640 was supplemented with 10% fetal bovine serum, 0.05 mM 2-mercaptoethanol, 1% sodium pyruvate, and 1% penicillin / streptomycin. Cells were cultured under standard conditions, i.e., 37°C, 5% CO2 in a humid environment; live / dead staining of the cells in the microspheres of step S6 in Example 1 was performed according to the Cyto3D™ Live-Dead Assay Kit instructions, and analyzed on NoviSight 3D cell analysis software.

[0157] Results: The cell survival rate inside the microspheres was over 90% after 48 hours.

[0158] (3) Immunofluorescence staining: On day 3, the microspheres co-cultured with HUVECs in step S7 of Example 1 were fixed with 4% paraformaldehyde (PFA) solution and incubated overnight at 4°C. After fixation, the samples were permeabilized with 0.1% Triton X-100 (Sigma) solution for 10 minutes and then washed with phosphate-buffered saline (PBS). The fixed samples were then stained with Alexa Fluor 647 phalloidin (Invitrogen) at room temperature for 1 hour and washed again with PBS. For nuclear counterstaining, the samples were treated with 2 μg / mL Hoechst solution in PBS at room temperature for 30 minutes.

[0159] Results: As shown in Figure 5, the fibrous actin of HUVECs covers the entire microsphere, which facilitates the formation of vascular structures between the microspheres and the body after injection, providing nutrition to the pancreatic β cells within the microspheres. Insulin is released from the pancreatic β cells and flows into the body through these vascular structures.

[0160] Figure 6 is a schematic diagram of the structure of a computer-readable storage medium according to an embodiment of the present invention. The computer-readable storage medium of the fourth embodiment of the present invention stores program instructions capable of implementing the above-described image recognition-based artificial organ preparation method. These program instructions can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned computer-readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0161] The methods described in the first embodiment of the present invention are applicable to the computer-readable storage medium embodiment. The specific functions implemented by the computer-readable storage medium embodiment are the same as those in the above method embodiment, and the beneficial effects achieved are also the same as those achieved by the above method.

[0162] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to realize the image recognition-based artificial organ preparation method provided in the above embodiment.

[0163] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of the present invention are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0164] Those skilled in the art will understand that modules in the device of the embodiments of the present invention can be adaptively modified and placed in one or more devices different from those embodiments. Modules, units, or components in the embodiments of the present invention can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the corresponding claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the corresponding claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0165] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0166] Furthermore, the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to mutually. In particular, for embodiments such as apparatus and devices, since they are basically similar to the method embodiments, the relevant parts can be referred to the description of the method embodiments. The apparatus, devices, and other embodiments described above are merely illustrative, and the modules, units, etc., described as separate components may or may not be physically separate, that is, they may be located in one place or distributed in multiple places, such as nodes in a system network. Specifically, some or all of the modules and units can be selected according to actual needs to achieve the purpose of the above-described embodiment solutions. Those skilled in the art can understand and implement this without creative effort.

[0167] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0168] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0169] Furthermore, the terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying relative importance, or implicitly specifying the number of technical features indicated in this embodiment. Therefore, features defined with terms such as "first" and "second" in the embodiments of this invention can explicitly or implicitly indicate that the embodiment includes at least one of those features. In the description of this invention, the word "multiple" means at least two or more, such as two, three, four, etc., unless otherwise explicitly specified in the embodiments.

[0170] In embodiments of the present invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, components, features, and elements with the same names in different embodiments of the present invention may have the same meaning or different meanings, the specific meaning of which must be determined by its interpretation in that specific embodiment or further in conjunction with the context of that specific embodiment.

[0171] Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention. Other embodiments of the present invention will readily conceive of by considering the specification and practicing the invention. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

Claims

1. A method for sorting hydrogel microspheres encapsulating target cells, characterized in that, Includes the following steps: Images of hydrogel microspheres in a microfluidic chip were acquired. The analysis results were obtained by analyzing the images of hydrogel microspheres using a neural network; Based on the analysis results, hydrogel microspheres encapsulating target cells were sorted out. The neural network includes a feature extraction layer and a multi-scale joint output head. The hydrogel microsphere image is convolved by the feature extraction layer and then processed by the multi-scale joint output head to obtain the analysis results. The analysis results include the coordinates of the hydrogel microspheres in the image, as well as marking information on whether the hydrogel microspheres encapsulate target cells.

2. The method for sorting hydrogel microspheres encapsulating target cells according to claim 1, characterized in that, Before the neural network analyzes the hydrogel microsphere image, the following steps are included: The hydrogel microsphere image is format-aligned to obtain a first preprocessed image; The first preprocessed image is subjected to Gaussian filtering to obtain the second preprocessed image; Edge enhancement is performed on the second preprocessed image to obtain the third preprocessed image; The third preprocessed image is subjected to grayscale mapping and contrast stretching to obtain a preprocessed hydrogel microsphere image, which is then used for image analysis.

3. The method for sorting hydrogel microspheres encapsulating target cells according to claim 1, characterized in that, The feature extraction layer is formed by stacking multiple convolutional layers and is used to extract multi-level feature information from the hydrogel microsphere image. The joint output head is used to fuse the feature information at different levels, perform convolution operation on the fused feature map to obtain a prediction tensor, and convert the prediction tensor into an analysis result. The analysis result includes the coordinates of the hydrogel microspheres in the image and the labeling information of whether the hydrogel microspheres encapsulate target cells.

4. The method for sorting hydrogel microspheres encapsulating target cells according to claim 3, characterized in that, The feature extraction layer includes four consecutively stacked depthwise separable convolutional units; each depthwise separable convolutional unit consists of a depthwise convolutional kernel, a pointwise convolutional kernel, and a nonlinear activation layer; the depthwise convolutional kernel is used to extract local grayscale changes and shape features of the input image pixel by pixel to obtain multiple single-channel feature maps; The point convolution kernel is used to perform cross-channel fusion of multiple single-channel feature maps output by the depth convolution kernel to obtain a fused feature map; the nonlinear activation layer is used to perform row-wise activation on the fused feature map; In the four consecutively stacked depthwise separable convolutional units, the depthwise separable convolutional unit of the first layer takes the hydrogel microsphere image as input, is used to perform initial downsampling of the input image, and reduce the size of the input image to 1 / 2, converting the single-channel raw signal into 8-dimensional primary features; In the four consecutively stacked depthwise separable convolutional units, the depthwise separable convolutional unit of the second layer takes the output image of the depthwise separable convolutional unit of the first layer as input, and is used to perform secondary downsampling on the input image, reducing the size of the input image to 1 / 4 of the original hydrogel microsphere image, and upgrading the 8-dimensional primary features to 16-dimensional intermediate features. In the four consecutively stacked depthwise separable convolutional units, the depthwise separable convolutional unit of the third layer takes the output image of the depthwise separable convolutional unit of the second layer as input, and performs downsampling on the input image three times to reduce the size of the input image to 1 / 8 of the original hydrogel microsphere image, while maintaining 16-dimensional mid-layer features to focus on the outline and local details of the hydrogel microspheres. In the four consecutively stacked depthwise separable convolutional units, the fourth layer depthwise separable convolutional unit takes the output image of the third layer depthwise separable convolutional unit as input, and performs downsampling on the input image four times, reducing the size of the input image to 1 / 16 of the original hydrogel microsphere image, and upscaling the 16-dimensional mid-layer features to 24-dimensional high-layer features; the output image of the fourth layer depthwise separable convolutional unit is passed to the joint output head as the output of the feature extraction layer.

5. The method for sorting hydrogel microspheres encapsulating target cells according to claim 4, characterized in that, The joint output head is used to perform dual-scale feature reconstruction on the output image of the feature extraction layer to obtain a first-size feature map and a second-size feature map; the first-size feature map and the second-size feature map are used to analyze and identify hydrogel microspheres of different sizes; the first size is smaller than the second size. The joint output head obtains a first-size feature map by performing bilinear interpolation on the output image; the joint output head obtains a second-size feature map by performing bilinear interpolation on the output image after performing a convolution. The first and second size feature maps contain multiple candidate boxes that match the contours of the hydrogel microspheres; the predicted tensor is obtained by analyzing and identifying the candidate boxes.

6. The method for sorting hydrogel microspheres encapsulating target cells according to claim 1, characterized in that, The process of sorting hydrogel microspheres encapsulating target cells based on the analysis results includes the following steps: Based on the analysis results of the image analysis, generate electrode driving instructions; The electrode driving command controls the electrode driving module in the microfluidic chip to generate an electric field according to a preset timing sequence, and the electric field separates the hydrogel microspheres containing target cells from those not containing target cells into different flow channels.

7. A method for preparing an artificial organ, characterized in that, Includes the following steps: Prepare hydrogel microspheres encapsulating target cells in a microfluidic chip; The hydrogel microspheres encapsulating target cells are sorted using the sorting method for hydrogel microspheres encapsulating target cells as described in any one of claims 1-6; The hydrogel microspheres containing target cells were purified and cultured to obtain artificial organs.

8. The method for preparing an artificial organ according to claim 7, characterized in that, The preparation of hydrogel microspheres in a microfluidic chip specifically includes the following steps: Prepare a buffer solution containing gel matrix components as the first aqueous phase; Prepare a buffer solution containing target cells and a cross-linking agent as the second aqueous phase; Preparation of microfluidic carrier oil phase and spacer oil phase; The first aqueous phase, the second aqueous phase, the carrier oil phase, and the spacer oil phase are injected into the microfluidic chip at a preset flow rate, so that the first aqueous phase, the second aqueous phase, and the carrier oil phase undergo a cross-linking reaction in the microfluidic chip, thereby forming hydrogel microspheres in which the oil phase encapsulates the aqueous phase.

9. A droplet microfluidic cell sorting system for implementing the hydrogel microsphere sorting method for encapsulating target cells as described in any one of claims 1-8 and / or the artificial organ preparation method, characterized in that, It includes a microfluidic chip module, an image acquisition module, a central processing unit module, and an electrode driving module; the image acquisition module and the electrode driving module are disposed in the microfluidic chip module, and the central processing unit module establishes communication connections with the image acquisition module and the electrode driving module respectively; The microfluidic chip module is used for the preparation and sorting of hydrogel microspheres; The image acquisition module is used to acquire images of the hydrogel microspheres prepared in the microfluidic chip module and transmit the acquired hydrogel microsphere images to the central processing unit module. The central processing unit module is used to perform image analysis on the hydrogel microsphere images and generate electrode driving instructions based on the image analysis results, which are then sent to the electrode driving module. The electrode driving module is used to generate an electric field in the microfluidic chip module according to the electrode driving command, and to separate the hydrogel microspheres containing target cells from those not containing target cells through the electric field.

10. A computer-readable storage medium, characterized in that, The storage medium stores a program that is executed by a processor to implement a method for sorting hydrogel microspheres encapsulating target cells and / or a method for preparing artificial organs as described in any one of claims 1-8.

Images