Label-free cell activity detection method, device, computer device and storage medium

Abstract

The application relates to the technical field of cell activity detection, and discloses a label-free cell activity detection method and device, computer equipment and a storage medium.The method comprises the following steps: acquiring a to-be-detected bright field cell image; and performing cell activity detection based on the to-be-detected bright field cell image and a trained deep learning model to obtain a classification result of the to-be-detected bright field cell image, wherein the deep learning model comprises a first convolution module, a second convolution module, a third convolution module, a fourth convolution module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first splicing layer, a second splicing layer, a third splicing layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head and a third detection head; the trained deep learning model can be used to analyze the to-be-detected bright field cell image, and the activity of each cell in the image can be detected.

Description

Technical Field

[0001] This invention relates to the field of cell viability detection technology, and in particular to a label-free cell viability detection method, apparatus, computer equipment, and storage medium. Background Technology

[0002] Digital microfluidics (DMF), based on EWoD technology, is a technique for manipulating droplets at the micrometer scale using electric fields, pressure, or other physical methods. This technology holds immense potential in the biomedical field because it enables efficient and precise control of the movement, splitting, fusion, and mixing of minute amounts of liquids. Compared to traditional continuous microfluidics, DMF offers greater flexibility in droplet manipulation and enables high-throughput analysis. Therefore, it is widely used in fields such as biological sample processing, chemical synthesis, drug screening, and DNA sequencing.

[0003] In biological experiments, accurate detection of cell viability is a key factor for experimental success, especially in single-cell studies, where the activity and state of each cell directly affect the experimental results and conclusions. Currently, the commonly used method for detecting cell viability in digital microfluidic systems is fluorescence detection. However, this method has drawbacks, including altering cell biological behavior, increasing experimental complexity and operating costs, and requiring a highly sensitive fluorescence microscope. Summary of the Invention

[0004] Based on this, it is necessary to address the technical problem of poor cell activity detection performance in existing technologies by proposing a label-free cell activity detection method, device, computer equipment, and storage medium.

[0005] In a first aspect, a label-free method for detecting cell viability is provided, the method comprising:

[0006] Acquire images of the bright-field cells to be detected;

[0007] Cell activity detection is performed based on the bright-field cell image to be detected and the trained deep learning model to obtain the classification result of the bright-field cell image to be detected. The deep learning model includes a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head.

[0008] Secondly, a label-free cell viability detection device is provided, the device comprising:

[0009] The acquisition module is used to acquire images of bright-field cells to be detected;

[0010] The classification module is used to perform cell activity detection based on the bright-field cell image to be detected and the trained deep learning model, and obtain the classification result of the bright-field cell image to be detected. The deep learning model includes a first convolution module, a second convolution module, a third convolution module, a fourth convolution module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head.

[0011] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the label-free cell viability detection method described above.

[0012] Fourthly, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described label-free cell viability detection method.

[0013] The label-free cell viability detection method proposed in this invention acquires a bright-field cell image to be detected, and then performs cell viability detection based on the bright-field cell image and a trained deep learning model to obtain the classification result of the bright-field cell image. The deep learning model includes a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head. By analyzing the bright-field cell image using the trained deep learning model, the viability of each cell in the image can be detected. Data acquisition is simple and fast, and analysis efficiency is high. This not only reduces the influence of the experimental environment and significantly improves experimental speed, but also greatly reduces experimental costs, eliminates the potential influence of fluorescent dyes on cells, and ensures the physiological integrity of the cells. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] in:

[0016] Figure 1 This is a diagram illustrating the application environment of a label-free cell viability detection method in one embodiment;

[0017] Figure 2 This is a flowchart of a label-free cell viability detection method in one embodiment;

[0018] Figure 3 This is a schematic diagram of the convolutional module structure in a label-free cell viability detection method of one embodiment;

[0019] Figure 4 This is a schematic diagram of the coding layer structure of a label-free cell viability detection method in one embodiment;

[0020] Figure 5 This is a schematic diagram of the Jamba model structure for a label-free cell viability detection method in one embodiment;

[0021] Figure 6 This is a schematic diagram of the detection head structure of a label-free cell viability detection method in one embodiment;

[0022] Figure 7 This is a schematic diagram of the deep learning model structure for a label-free cell viability detection method in one embodiment;

[0023] Figure 8 This is a structural block diagram of a label-free cell viability detection device in one embodiment;

[0024] Figure 9 This is a structural block diagram of a computer device in one embodiment;

[0025] Figure 10 This is a structural block diagram of a computer device in another embodiment. Detailed Implementation

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The label-free cell viability detection method provided in this invention can be applied to, for example... Figure 1In this application environment, client 110 communicates with server 120 via a network. Server 120 can receive the bright-field cell image to be detected through client 110, and then server 120 performs cell activity detection based on the bright-field cell image and a trained deep learning model to obtain the classification result of the bright-field cell image. The deep learning model includes a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head. This invention can analyze the bright-field cell image to be detected using a trained deep learning model, thereby detecting the activity of each cell in the image. Data acquisition is simple and fast, and analysis efficiency is high. It not only reduces the influence of the experimental environment and significantly improves the experimental speed, but also greatly reduces experimental costs, eliminates the potential influence of fluorescent dyes on cells, and ensures the physiological integrity of cells. The client 110 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server 120 can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will now be described in detail through specific embodiments.

[0030] Please see Figure 2 As shown, Figure 2 A schematic flowchart of a label-free cell viability detection method according to an embodiment of the present invention includes the following steps:

[0031] Step S101: Obtain the bright-field cell image to be detected;

[0032] Among them, the bright-field cell image to be detected can be a bright-field image taken under a regular microscope.

[0033] Step S102: Based on the bright-field cell image to be detected and the trained deep learning model, cell activity detection is performed to obtain the classification result of the bright-field cell image to be detected. The deep learning model includes a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head.

[0034] Among them, reference Figure 3The model structure of the first, second, third, and fourth convolutional modules can be identical. Specifically, each convolutional module includes a depthwise separable convolutional layer, a batch normalization layer, and a SILU activation layer connected sequentially. The input of the depthwise separable convolutional layer and the output of the SILU activation layer are connected via a skip connection. The depthwise separable convolutional layer consists of depthwise convolution and pointwise convolution. Depthwise convolution is used to extract spatial features, while pointwise convolution is used to extract channel features. Using depthwise separable convolution can significantly reduce the number of parameters, decrease computational cost, and improve feature representation capabilities.

[0035] Batch Normalization Layer: This layer normalizes the output of each layer, making the input of the next layer more stable. This helps reduce the risk of overfitting and improves the model's generalization ability.

[0036] SILU activation layer: Using the SILU activation function as a separate layer, it can adaptively scale the input values, helping to reduce overfitting. Skip connection: Connecting the input directly to the output, this connection method can alleviate the gradient vanishing problem and help the network learn better.

[0037] The model structures of the first, second, and third upsampling layers can be the same. The upsampling layer uses bilinear interpolation to increase the size of the feature map, recovering the spatial location information of the target and improving the detection accuracy for cells (small targets). Similarly, the model structures of the first, second, and third concatenation layers can be the same. The concatenation layer connects two tensors along the channel dimension to better capture the relationships between different features in subsequent layers.

[0038] refer to Figure 4 The model structure of the first, second, and third encoding layers can be the same. Specifically, the encoding layers can include sequentially connected image patch embedding and positional encoding layers. Image patch embedding: Divides the image into fixed-size image patches and flattens them into one-dimensional vectors to adapt to the input of the next layer. Positional encoding layer: Adds positional encoding to each image patch to preserve the positional information of the image patch, enabling the network to understand the relative positional relationships between image patches.

[0039] refer to Figure 5The first, second, and third Jamba models can have the same model structure. A Jamba Module consists of a series of Mamba layers, Mamba MoE layers, Transformer layers, Mamba MoE layers, Mamba layers, and Mamba MoE layers connected sequentially. Each Mamba MoE layer's input includes the number of target classes. The Transformer layer, used in traditional Transformer models, primarily handles the self-attention mechanism of the input data, allowing the model to consider all positions in the sequence when generating the output. The Mamba layer is a state-space model that, compared to Transformer layers, can be trained in parallel and is more efficient in processing long sequences, while also exhibiting superior performance in capturing long-range dependencies. The Mamba MoE layer is a hybrid expert layer that uses multiple "expert" networks to process specific parts or tasks of the input data. A gating mechanism then determines the degree or probability of activation for each expert in the output, thereby improving the model's accuracy in identifying cells. Number of target categories: The number of cell categories to be predicted is encoded and fed into the network as a sequence for training, ultimately predicting the category and location of the detection box for each target.

[0040] refer to Figure 6 The model structures of the first, second, and third detection heads can be identical. Each detection head consists of two feedforward neural networks: one feedforward neural network uses the multi-scale output for target category prediction, and the other feedforward neural network uses the multi-scale output for target bounding box prediction.

[0041] Specifically, refer to Figure 7 The image of the bright-field cells to be detected is input into the first convolution module to obtain a first feature; the first feature is input into the second convolution module to obtain a second feature; the second feature is input into the third convolution module to obtain a third feature; and the third feature is input into the fourth convolution module to obtain a fourth feature.

[0042] The fourth feature is input to the third upsampling layer to obtain the fifth feature. The third feature and the fifth feature are then input to the third splicing layer to obtain the sixth feature.

[0043] The sixth feature is input to the second upsampling layer to obtain the seventh feature. The second feature and the seventh feature are then input to the second splicing layer to obtain the eighth feature.

[0044] The eighth feature is input to the first upsampling layer to obtain the ninth feature. The first feature and the ninth feature are then input to the first splicing layer to obtain the tenth feature.

[0045] The tenth feature is input into the first encoding layer, the first Jamba model, and the first output head, which are connected in sequence, to obtain the first target category prediction and the first target bounding box prediction.

[0046] The eighth feature is input into the second encoding layer, the second Jamba model, and the second output head, which are connected in sequence, to obtain the second target category prediction and the second target bounding box prediction.

[0047] The sixth feature is input into the third encoding layer, the third Jamba model, and the third output head, which are connected in sequence, to obtain the third target category prediction and the third target bounding box prediction.

[0048] The first target category prediction, the first target bounding box prediction, the second target category prediction, the second target bounding box prediction, the third target category prediction, and the third target bounding box prediction are processed using bilinear interpolation to obtain the target category prediction and the target bounding box prediction.

[0049] In one embodiment, the step of performing cell viability detection based on the bright-field cell image to be detected and the trained deep learning model to obtain the classification result of the bright-field cell image to be detected includes:

[0050] Step S201: Acquire bright-field cell image set and fluorescent-field cell image;

[0051] In one embodiment, the bright-field cell image and the fluorescence-field cell image are obtained by taking pictures of the cell sample using a fluorescence microscope. The cell sample is obtained by washing with PBS buffer and centrifuging, followed by incubation with cell viability detection reagent at room temperature in the dark.

[0052] As an example, firstly, cell samples are prepared: a cell suspension is prepared, washed with PBS buffer, centrifuged, and then incubated with Calcein / PI cell viability assay reagent at room temperature in the dark to ensure thorough staining. Next, cell images are acquired: the cell samples are photographed using a fluorescence microscope, resulting in numerous high-resolution bright-field and fluorescent-field cell images. These images contain physical information such as cell morphology, size, and transparency, which reflects the health and activity of the cells. Live cells typically have well-defined boundaries and a uniform morphology, while apoptotic or dead cells may appear deformed or have blurred outlines.

[0053] As another example, 1. A CHO-S cell suspension was prepared, washed with phosphate buffer, centrifuged, and resuspended in the Calcein / PI cell viability assay kit from Beyotime Biotechnology Co., Ltd. The suspension was then incubated at room temperature in the dark for 30 minutes. 2. The incubated CHO-S cell suspension was injected into a digital microfluidic chip. 3. Cell samples were imaged using a Nikon Ni-U fluorescence microscope, and bright-field and fluorescent-field cell images were recorded.

[0054] Step S202: Preprocess the bright-field cell image set;

[0055] In one embodiment, the preprocessing includes image scaling, noise reduction, normalization, and image enhancement. Specifically, image preprocessing: To improve data quality, the acquired bright-field cell images are preprocessed to make them more suitable for training deep learning models. Commonly used image preprocessing methods include: ① Image scaling: The image size is changed using methods such as Nearest Neighbor Interpolation, Bilinear Interpolation, and Bicubic Interpolation to adapt to the input of the deep learning model. ② Noise reduction: Noise in the image is removed using methods such as Gaussian Blur, Mean Filtering, and Median Filtering to improve image clarity. ③ Normalization: The pixel values ​​of the image are adjusted to a certain range (e.g., [0, 1] or [-1, 1]) using methods such as Normalization and Z-Score Normalization to reduce the impact of brightness differences between different images on the model. ④ Image enhancement: Increase the quantity and diversity of training data through methods such as image rotation, image cropping, and image flipping to improve the performance and robustness of deep learning models.

[0056] As another example, for the preprocessed bright-field cell image, Gaussian filtering is first performed to improve the image clarity. Then, normalization is used to adjust the pixel values ​​of the image to [0, 1] to reduce the impact of brightness differences between different images on the model. Finally, image rotation, image cropping, and image flipping are used to increase the amount and diversity of data.

[0057] Step S203: Based on the fluorescence field cell image as a reference, the background, live cells, and dead cells in each image of the preprocessed bright field cell image set are labeled to obtain the target bright field cell image set;

[0058] In this embodiment, the acquired fluorescent field cell image is used as a reference standard. The LabelImg annotation tool can be used to manually or semi-automatically annotate the preprocessed bright field cell image, and each pixel in the preprocessed bright field cell image is divided into three categories: background, live cells, and dead cells.

[0059] Step S204: Based on the target bright-field cell image set and the initial deep learning model, a model is trained to obtain a trained deep learning model. The trained deep learning model is used to classify each pixel in the bright-field cell image to be detected and output the classification result of the bright-field cell image to be detected. The deep learning model includes a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head.

[0060] The deep learning model can be a convolutional neural network, and the bright-field cell image to be detected can be a bright-field image taken under a regular microscope.

[0061] In this embodiment, to quickly distinguish the activity of each cell from bright-field cell images, a single-stage initial deep learning model is used for iterative training. To ensure the accuracy and robustness of the model, the following methods are typically used to improve its performance: ① Using a loss function that measures the difference between the model's predictions and the actual results. Common loss functions include: Mean Squared Error (MSE), Cross-Entropy (CE), Logarithmic Loss, Categorical Cross-Entropy (CCES), Kullback-Leibler Divergence (KLD), etc. ② Optimization algorithms that continuously adjust the internal parameters to minimize the loss function are commonly used, including: Gradient Descent (GD), Stochastic Gradient Descent (SGD), Mini-batch Gradient Descent, Momentum, and Adam (adaptive learning rate algorithm). ③ Regularization techniques to prevent the model from overfitting the training data are commonly used, including: L1 regularization, L2 regularization, Dropout regularization, and Elastic Net regularization.

[0062] Specifically, an overall error function can be used to train the initial deep learning model. The output of the initial deep learning model includes the target category and the target bounding box. The error function for the target category uses the Categorical Cross-Entropy (CCES) function, and the error function for the target bounding box uses a weighted function of Mean Absolute Error (L1 Loss) and Cross-Union Error (GIoU Loss). The overall error function is as follows: Target bounding box error function = 0.4 × Mean Absolute Error + 0.6 × Cross-Union Error; Detector head error function = 0.7 × Target Category Error Function + 0.3 × Target Bounding Box Error Function; Overall error function = 0.6 × First Detector Head Error Function + 0.3 × Second Detector Head Error Function + 0.1 ×

[0063] Error function of the third detection head.

[0064] In one embodiment, the images in the target bright-field cell image set are divided into a training set, a validation set, and a test set in a ratio of 7:1:1, wherein the number of cell samples in the training set is 21,060, the number of cell samples in the validation set is 3,122, and the number of cell samples in the test set is 3,234.

[0065] In one embodiment, a model is trained based on a target bright-field cell image set and an initial deep learning model, and the parameters of the initial deep learning model are optimized based on the cross-entropy loss function (CE), stochastic gradient descent (SGD), and dropout regularization; when the number of training iterations of the initial deep learning model meets a preset number, the initial deep learning model is used as the trained deep learning model.

[0066] As an example, after obtaining the trained deep learning model, it can be used to detect bright-field cells in images. Specifically, bright-field cell images acquired under a regular optical microscope are used as the target bright-field cell images. These images are input into the deployed deep learning model for feature extraction. The extracted features include morphological features (such as cell shape and size), optical features (such as brightness and transparency), and texture features (such as cell surface structure). These features are then used in the deep learning model for automatic reasoning and detection, classifying each pixel of the input bright-field cell image and outputting the detection results.

[0067] As an example, the trained deep school model is exported to ONNX format and deployed as a C++ library into a digital microfluidic system. Bright-field cell images from the actual working scene are collected as input to the model, and the activity of each cell can be obtained without adding fluorescent staining reagents.

[0068] Please see Figure 8 As shown, in one embodiment, a label-free cell viability detection device is provided, the device comprising:

[0069] Acquisition module 10 is used to acquire images of bright-field cells to be detected;

[0070] The prediction module 20 is used to perform cell activity detection based on the bright-field cell image to be detected and the trained deep learning model, and obtain the classification result of the bright-field cell image to be detected. The deep learning model includes a first convolution module, a second convolution module, a third convolution module, a fourth convolution module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head.

[0071] In one embodiment, the prediction module 20 is used for:

[0072] The image of the bright-field cells to be detected is input into the first convolution module to obtain a first feature; the first feature is input into the second convolution module to obtain a second feature; the second feature is input into the third convolution module to obtain a third feature; and the third feature is input into the fourth convolution module to obtain a fourth feature.

[0073] The fourth feature is input to the third upsampling layer to obtain the fifth feature. The third feature and the fifth feature are then input to the third splicing layer to obtain the sixth feature.

[0074] The sixth feature is input to the second upsampling layer to obtain the seventh feature. The second feature and the seventh feature are then input to the second splicing layer to obtain the eighth feature.

[0075] The eighth feature is input to the first upsampling layer to obtain the ninth feature. The first feature and the ninth feature are then input to the first splicing layer to obtain the tenth feature.

[0076] The tenth feature is input into the first encoding layer, the first Jamba model, and the first output head, which are connected in sequence, to obtain the first target category prediction and the first target bounding box prediction.

[0077] The eighth feature is input into the second encoding layer, the second Jamba model, and the second output head, which are connected in sequence, to obtain the second target category prediction and the second target bounding box prediction.

[0078] The sixth feature is input into the third encoding layer, the third Jamba model, and the third output head, which are connected in sequence, to obtain the third target category prediction and the third target bounding box prediction.

[0079] The first target category prediction, the first target bounding box prediction, the second target category prediction, the second target bounding box prediction, the third target category prediction, and the third target bounding box prediction are processed using bilinear interpolation to obtain the target category prediction and the target bounding box prediction.

[0080] In one embodiment, the label-free cell viability detection device is used for:

[0081] Acquire bright-field cell image sets and fluorescent-field cell images;

[0082] The bright-field cell image set is preprocessed;

[0083] Using fluorescent field cell images as a reference, the background, live cells, and dead cells in each image of the preprocessed bright field cell image set are labeled to obtain the target bright field cell image set.

[0084] Based on the target bright-field cell image set and the initial deep learning model, a model is trained to obtain a trained deep learning model. The trained deep learning model is used to classify each pixel in the bright-field cell image to be detected and outputs the target category prediction and target bounding box prediction of the bright-field cell image to be detected.

[0085] In one embodiment, the label-free cell viability detection device is used for:

[0086] The preprocessing includes image scaling, noise reduction, normalization, and image enhancement.

[0087] In one embodiment, the label-free cell viability detection device is used for:

[0088] The bright-field cell images and fluorescence-field cell images were obtained by taking pictures of the cell samples using a fluorescence microscope. The cell samples were washed with PBS buffer and centrifuged, and then incubated with cell viability detection reagent at room temperature in the dark.

[0089] In one embodiment, the label-free cell viability detection device is used for:

[0090] The model was trained based on the target bright-field cell image set and the initial deep learning model, and the parameters of the initial deep learning model were optimized based on the cross-entropy loss function, stochastic gradient descent, and regularization.

[0091] If the initial deep learning model has been trained a certain number of times, then the initial deep learning model is considered a trained deep learning model.

[0092] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a label-free cell viability detection method on the server side.

[0093] In one embodiment, a computer device is provided, which may be a client, and its internal structure diagram may be as follows: Figure 10 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When executed by the processor, the computer program implements the functions or steps of a label-free cell viability detection method on the client side.

[0094] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, performs the following steps:

[0095] Acquire images of the bright-field cells to be detected;

[0096] Cell activity detection is performed based on the bright-field cell image to be detected and the trained deep learning model to obtain the classification result of the bright-field cell image to be detected. The deep learning model includes a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head.

[0097] This invention can analyze bright-field cell images of the target cells using a trained deep learning model, thereby detecting the activity of each cell in the image. Data acquisition is simple and fast, and analysis efficiency is high. It can not only reduce the influence of the experimental environment and significantly improve the experimental speed, but also greatly reduce the experimental cost, eliminate the potential influence of fluorescent dyes on cells, and ensure the physiological integrity of cells.

[0098] In one embodiment, a computer-readable storage medium is provided that stores a computer program, which, when executed by a processor, performs the following steps:

[0099] Acquire images of the bright-field cells to be detected;

[0100] Cell activity detection is performed based on the bright-field cell image to be detected and the trained deep learning model to obtain the classification result of the bright-field cell image to be detected. The deep learning model includes a first convolutional module, a second convolutional module, a third convolutional module, a fourth convolutional module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first stitching layer, a second stitching layer, a third stitching layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head.

[0101] This invention can analyze bright-field cell images of the target cells using a trained deep learning model, thereby detecting the activity of each cell in the image. Data acquisition is simple and fast, and analysis efficiency is high. It can not only reduce the influence of the experimental environment and significantly improve the experimental speed, but also greatly reduce the experimental cost, eliminate the potential influence of fluorescent dyes on cells, and ensure the physiological integrity of cells.

[0102] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0103] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0104] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0105] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A label-free cell activity detection method, characterized by, The label-free cell activity detection method comprises: obtaining a bright field cell image to be detected; based on the bright field cell image to be detected and the trained deep learning model, the cell activity detection is carried out, and the classification result of the bright field cell image to be detected is obtained, wherein the deep learning model comprises a first convolution module, a second convolution module, a third convolution module, a fourth convolution module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first splicing layer, a second splicing layer, a third splicing layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head, and a third detection head; The step of detecting cell activity based on the bright field cell image to be detected and the trained deep learning model to obtain the classification result of the bright field cell image to be detected comprises: inputting the bright field cell image to be detected into the first convolution module to obtain first features; inputting the first features into the second convolution module to obtain second features; inputting the second features into the third convolution module to obtain third features; and inputting the third features into the fourth convolution module to obtain fourth features; the fourth features are input into the third upsampling layer to obtain fifth features, and the third features and the fifth features are input into the third splicing layer to obtain sixth features; the sixth features are input into the second upsampling layer to obtain seventh features, and the second features and the seventh features are input into the second splicing layer to obtain eighth features; the eighth features are input into the first upsampling layer to obtain ninth features, and the first features and the ninth features are input into the first splicing layer to obtain tenth features; the tenth features are input into the first encoding layer, the first Jamba model and the first output head connected in sequence to obtain first target category prediction and first target bounding box prediction; the eighth features are input into the second encoding layer, the second Jamba model and the second output head connected in sequence to obtain second target category prediction and second target bounding box prediction; the sixth features are input into the third encoding layer, the third Jamba model and the third output head connected in sequence to obtain third target category prediction and third target bounding box prediction; using a bilinear interpolation method, the first target category prediction, the first target bounding box prediction, the second target category prediction, the second target bounding box prediction, the third target category prediction, and the third target bounding box prediction are processed to obtain target category prediction and target bounding box prediction; The Jamba model comprises Mamba layers, Mamba MoE layers, Transformer layers, Mamba MoE layers, Mamba layers, and Mamba MoE layers connected in sequence. The input of each Mamba MoE layer further comprises a target category number.

2. The label-free cell activity detection method according to claim 1, wherein, The step of detecting cell activity based on the to-be-detected bright-field cell image and the trained deep learning model to obtain a classification result of the to-be-detected bright-field cell image comprises the following steps: obtaining a bright-field cell image set and a fluorescence field cell image; preprocessing the bright-field cell image set; annotating the background, living cells and dead cells in each image in the preprocessed bright-field cell image set based on the fluorescence field cell image as a reference to obtain a target bright-field cell image set; training a model based on the target bright-field cell image set and an initial deep learning model to obtain a trained deep learning model, wherein the trained deep learning model is used to classify each pixel point in a to-be-detected bright-field cell image and output a target class prediction and a target bounding box prediction of the to-be-detected bright-field cell image.

3. The label-free cell activity detection method according to claim 2, wherein, The preprocessing includes image scaling, noise reduction processing, normalization processing and image enhancement processing.

4. The label-free cell activity detection method according to claim 3, wherein, The bright-field cell image and the fluorescence field cell image are obtained by photographing a cell sample using a fluorescence microscope, and the cell sample is obtained by adding a cell activity detection reagent to the cell sample washed with PBS buffer and centrifuged under room temperature conditions and incubating in the dark.

5. The label-free cell activity detection method according to claim 4, wherein, The step of training a model based on the target bright-field cell image set and an initial deep learning model to obtain a trained deep learning model comprises the following steps: training a model based on the target bright-field cell image set and an initial deep learning model, and optimizing the parameters of the initial deep learning model based on a cross-entropy loss function, a stochastic gradient descent method and regularization; when the number of model training of the initial deep learning model meets a preset number, the initial deep learning model is used as the trained deep learning model.

6. A label-free cell activity detection device, characterized by, The label-free cell activity detection device comprises: an acquisition module configured to acquire a to-be-detected bright-field cell image; a prediction module configured to detect cell activity based on the to-be-detected bright-field cell image and a trained deep learning model to obtain a classification result of the to-be-detected bright-field cell image, wherein the deep learning model comprises a first convolution module, a second convolution module, a third convolution module, a fourth convolution module, a first upsampling layer, a second upsampling layer, a third upsampling layer, a first splicing layer, a second splicing layer, a third splicing layer, a first encoding layer, a second encoding layer, a third encoding layer, a first Jamba model, a second Jamba model, a third Jamba model, a first detection head, a second detection head and a third detection head; The step of detecting cell activity based on the to-be-detected bright-field cell image and the trained deep learning model to obtain a classification result of the to-be-detected bright-field cell image comprises the following steps: inputting the to-be-detected bright-field cell image into the first convolution module to obtain a first feature; inputting the first feature into the second convolution module to obtain a second feature; inputting the second feature into the third convolution module to obtain a third feature; and inputting the third feature into the fourth convolution module to obtain a fourth feature; The fourth feature is input to the third up-sampling layer to obtain a fifth feature, and the third feature and the fifth feature are input to the third concatenation layer to obtain a sixth feature; The sixth feature is input to the second up-sampling layer to obtain a seventh feature, and the second feature and the seventh feature are input to the second concatenation layer to obtain an eighth feature; The eighth feature is input to the first up-sampling layer to obtain a ninth feature, and the first feature and the ninth feature are input to the first concatenation layer to obtain a tenth feature; The tenth feature is input to the first encoding layer, the first Jamba model and the first output head connected in sequence to obtain a first target category prediction and a first target bounding box prediction; The eighth feature is input to the second encoding layer, the second Jamba model and the second output head connected in sequence to obtain a second target category prediction and a second target bounding box prediction; The sixth feature is input to the third encoding layer, the third Jamba model and the third output head connected in sequence to obtain a third target category prediction and a third target bounding box prediction; The first target category prediction, the first target bounding box prediction, the second target category prediction, the second target bounding box prediction, the third target category prediction and the third target bounding box prediction are processed by using a bilinear interpolation method to obtain a target category prediction and a target bounding box prediction. The Jamba model comprises Mamba layers, Mamba MoE layers, Transformer layers, Mamba MoE layers and Mamba layers connected in sequence, and the input of each Mamba MoE layer further comprises a target category number.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the label-free cell activity detection method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, the computer-readable storage medium comprising: The computer program is executed by the processor to implement the steps of the label-free cell activity detection method according to any one of claims 1 to 5.

Images