Microfluidic drug screening system, method and storage medium

Abstract

The application relates to a microfluidic drug screening system, method and storage medium, the system comprising: a microfluidic flow channel chip for generating a hydrogel droplet, curing in a crosslinking area to form an organoid microsphere; a drug administration module for forming the organoid microsphere into a to-be-tested microsphere array; an optical detection module for optical imaging of each to-be-tested microsphere and obtaining a corresponding microsphere image; a calculation and control module for analyzing the microsphere image, determining size information and fluorescence intensity information corresponding to each to-be-tested microsphere, generating a drug screening result, and generating a parameter control instruction when the size information and the fluorescence intensity information meet preset adjustment conditions; and a parameter adjustment module for adjusting fluid parameters and / or drug administration parameters in response to the parameter control instruction. According to the application, a closed loop is constructed among preparation, drug administration, imaging, identification and feedback, an integrated screening process is realized, microsphere size variation is reduced, and fluorescence quantitative comparability and drug efficacy discrimination accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of biotechnology, and in particular to a microfluidic drug screening system, method, and storage medium. Background Technology

[0002] Tumor organoids, because they retain the genetic and phenotypic characteristics of patient-derived tissues, are considered to be closer to clinical reality than traditional two-dimensional cell lines and some animal models, making them important tools for personalized drug efficacy evaluation and precision medicine research. Microfluidic droplet technology has been extensively studied in organoid construction and drug screening. This technology can precisely control fluid behavior at the micrometer scale, generating uniform droplets or hydrogel microspheres through the confluence of continuous and dispersed phases, achieving the encapsulation and high-throughput preparation of cells and organoids.

[0003] Existing organoid culture and detection methods are still largely based on a linear workflow of "preparation-drug application-imaging-offline analysis". In drug screening, drug administration is often performed offline after microspheres or organoids are prepared, with the imaging and data analysis processes being disconnected from the preparation process. Furthermore, existing organoid culture and detection methods suffer from uneven size distribution of organoids or microspheres, significant batch-to-batch variability, and reliance on manual experience, resulting in poor reproducibility and low throughput. Summary of the Invention

[0004] In view of this, it is necessary to provide a microfluidic drug screening system, method and storage medium to solve the technical problems of low size variation, low comparability of fluorescence quantification and low accuracy of drug efficacy discrimination of organoid microspheres in the prior art.

[0005] To address the aforementioned problems, in a first aspect, the present invention provides a microfluidic drug screening system, comprising: Microfluidic channel chips are used to generate hydrogel droplets containing organoids or cells under the action of continuous and dispersed phases, and solidify in the cross-linking region to form organoid microspheres; A drug delivery module is used to treat the organoid microspheres with at least one drug at multiple dose concentrations to form a test microsphere array, the test microsphere array comprising a plurality of test microspheres; The optical detection module is used to perform optical imaging on each of the microspheres to be tested and obtain the corresponding microsphere images; The calculation and control module is communicatively connected to the optical detection module. It is used to analyze the microsphere image, determine the size information and fluorescence intensity information of each microsphere to be tested, generate drug screening results, and generate parameter control instructions when the size information and fluorescence intensity information meet the preset adjustment conditions. A parameter adjustment module, connected to the calculation and control module, the microfluidic channel chip, and the drug delivery module, is used to adjust the fluid parameters of the microfluidic channel chip and / or the drug delivery parameters of the drug delivery module in response to the parameter control command.

[0006] In one possible implementation, the calculation and control module includes a PID controller or a model predictive controller; The PID controller is used to perform proportional, integral, and derivative operations based on the deviation between the size information and the preset size threshold, and / or the deviation between the fluorescence intensity information and the preset intensity threshold, to generate the parameter control command. The model prediction controller is used to construct a prediction model based on the size information and / or the fluorescence intensity information, and generate the parameter control instructions through rolling optimization solution.

[0007] One possible implementation also includes: The calibration and normalization module, connected to the calculation and control module, is used to store calibration data of reference microspheres with predetermined fluorescence intensity, and to establish a calibration model between fluorescence intensity and imaging parameters based on the calibration data. The calibration and normalization module is also used to perform cross-batch and cross-device normalization processing on the fluorescence intensity information of each microsphere to be tested according to the calibration model, and to store and trace the microsphere image, the size information of each microsphere to be tested, the fluorescence intensity information of each microsphere to be tested, the drug screening results, the fluid parameters and the drug administration parameters.

[0008] In one possible implementation, the computing and control module is configured with a trained deep learning model for microsphere analysis; The deep learning model for microsphere analysis receives the microsphere image as input and performs instance segmentation and intensity regression processing on the microsphere image to determine the size and fluorescence intensity information of each microsphere to be tested; and Based on the fluorescence intensity information corresponding to each microsphere to be tested, the drug screening results are predicted and output, including IC50 values ​​or sensitivity typing results.

[0009] In one possible implementation, the drug delivery module includes multiple drug delivery inlets disposed inside the microfluidic channel chip and a microchannel network communicating with the drug delivery inlets; The administration inlet is used to introduce the drug at different concentrations; The microfluidic network is used to form the plurality of dose concentrations within the microfluidic channel chip; And / or, The drug delivery module includes an automated drug delivery unit connected downstream of the microfluidic channel chip; The automated drug delivery unit is used to form the multiple dose concentrations on the culture plate after the organoid microspheres leave the chip and are transferred to the culture plate, so that the organoid microspheres transferred to the culture plate receive drug treatment of the corresponding concentration.

[0010] In one possible implementation, the optical detection module includes: A multi-band excitation light source is used to emit excitation light of different bands to excite the microspheres under test to generate fluorescence signals; An emission filter array is used to selectively transmit the fluorescence signal in a preset wavelength band; An image sensor is used to receive fluorescence signals transmitted through the emission filter group to obtain an image of the original microspheres; An image correction unit is used to perform flat field correction and dark field correction on the original microsphere image to eliminate uneven illumination and dark current noise, thereby generating the microsphere image.

[0011] In one possible implementation, the computing and control module includes: The preprocessing unit is used to preprocess the microsphere image, and the preprocessing includes grayscale conversion, filtering and noise reduction, and contrast enhancement. The segmentation and recognition unit is used to identify and extract the target region of a single microsphere under test using an image segmentation algorithm; The feature extraction unit is used to calculate the size information and fluorescence intensity information of each microsphere to be tested based on the target region. A normalization unit is used to normalize the fluorescence intensity information; The drug efficacy fitting unit is used to fit the dose-response curve using a four-parameter Logistic equation based on the normalized fluorescence intensity information, and obtain the IC50 value or area under the curve as the drug screening result.

[0012] In one possible implementation, the microfluidic channel chip adopts a confluence focusing or T-junction structure, with the nozzle having an equivalent narrowest width of 50–150 μm to form organoid microspheres with a diameter of 30–60 μm. The downstream side of the microfluidic channel chip is provided with an optical window, the crosslinking region and a sorting or collection branch; The dispersed phase is a crosslinkable hydrogel solution containing sodium alginate, and the continuous phase is a biocompatible oil phase containing surfactant. The crosslinking region is used to provide chemical or photocrosslinking conditions for the solidification of the hydrogel droplets.

[0013] In a second aspect, the present invention also provides a microfluidic drug screening method, applied to the microfluidic drug screening system described in any one of the first aspects, comprising: Within the microfluidic channel chip, hydrogel droplets containing organoids or cells are generated through the confluence shearing action of the continuous phase and the dispersed phase, and the hydrogel droplets are solidified in the cross-linking region to form organoid microspheres. The drug delivery module allows the organoid microspheres to be treated with at least one drug at multiple dose concentrations to form a test microsphere array, which includes multiple test microspheres. The optical detection module performs optical imaging on each of the microspheres to be tested to obtain the corresponding microsphere images. The calculation and control module analyzes the microsphere images to determine the size and fluorescence intensity information of each microsphere to be tested, and generates drug screening results. When the size information and fluorescence intensity information meet the preset adjustment conditions, the calculation and control module generates parameter control instructions, and adjusts the fluid parameters of the microfluidic channel chip and / or the drug delivery parameters of the drug delivery module in real time through the parameter adjustment module.

[0014] Thirdly, the present invention also provides a computer-readable storage medium for storing a computer-readable program or instructions, which, when executed by a processor, can implement the steps in the microfluidic drug screening method described in the second aspect above.

[0015] The beneficial effects of this invention are: The microfluidic drug screening system provided by this invention includes a microfluidic channel chip for generating hydrogel droplets embedding organoids or cells under the action of a continuous phase and a dispersed phase, and solidifying them in the cross-linking region to form organoid microspheres, thereby improving the uniformity of microsphere size and the stability of morphology; a drug delivery module for treating the organoid microspheres with at least one drug at multiple dose concentrations to form a test microsphere array, the test microsphere array including multiple test microspheres, thereby improving the throughput and data acquisition efficiency of drug screening; an optical detection module for optical imaging of each test microsphere to acquire the corresponding microsphere image, ensuring the synchronous acquisition of microsphere morphological information and fluorescence response signal; and a calculation and control module, which is integrated with the optical... The detection module is a communication connection used to analyze microsphere images, determine the size and fluorescence intensity information of each microsphere to be tested, and generate drug screening results. This improves the automation level and consistency of screening data processing, and generates parameter control commands when the size and fluorescence intensity information meet preset adjustment conditions. The parameter adjustment module is connected to the calculation and control module, the microfluidic channel chip, and the drug delivery module. It is used to respond to parameter control commands and adjust the fluid parameters of the microfluidic channel chip and / or the drug delivery parameters of the drug delivery module. This enables the system to correct process deviations in real time, improving the stability of microsphere size between different preparation batches and the batch-to-batch comparability of drug screening results. Attached Figure Description

[0016] Figure 1 A schematic diagram of an embodiment of the microfluidic drug screening system provided by the present invention; Figure 2 This is a flowchart illustrating the microsphere image recognition algorithm provided by the present invention; Figure 3 A schematic diagram of the dose-response curve provided by the present invention; Figure 4 This is a schematic diagram of a multi-drug, multi-dose screening heatmap provided by the present invention; Figure 5 This is a schematic flowchart of an embodiment of the microfluidic drug screening method provided by the present invention. Detailed Implementation

[0017] 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 a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0018] In the description of the embodiments of the present invention, unless otherwise stated, "a plurality of" means two or more.

[0019] 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 their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature.

[0020] 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 the invention. 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.

[0021] This invention provides a microfluidic drug screening system, method, and storage medium, which are described below.

[0022] Figure 1 This is a schematic diagram of an embodiment of the microfluidic drug screening system provided by the present invention, as shown below. Figure 1 As shown, the microfluidic drug screening system includes: The microfluidic channel chip 10 is used to generate hydrogel droplets containing organoids or cells under the action of a continuous phase and a dispersed phase, and solidify in the cross-linking region to form organoid microspheres.

[0023] Among them, the microfluidic channel chip 10 refers to a flat-plate device with micron-level fluid channels processed inside, and its materials can be selected from polydimethylsiloxane, polymethyl methacrylate, glass or cyclic olefin copolymer, etc.

[0024] Hydrogel droplets are tiny droplets formed from hydrogel precursor solutions under continuous phase shear, containing organoids or cells. The droplet size is typically on the order of tens to hundreds of micrometers. Under cross-linking conditions, the hydrogel precursor solution can transform from a liquid to a gel state, thereby fixing the embedded material within a three-dimensional network structure.

[0025] Specifically, the microfluidic channel chip 10 has a dispersed phase inlet and a continuous phase inlet. The dispersed phase is a hydrogel precursor solution containing organoids or cells, and the continuous phase is a fluid immiscible with the dispersed phase. The two fluid phases meet in the confluence region of the chip. Under shear force, the dispersed phase is divided into uniformly sized hydrogel droplets, and the organoids or cells are embedded inside the droplets. The droplets move downstream with the fluid and enter the cross-linking region. Under external cross-linking conditions (such as ionic cross-linking, photo-initiated cross-linking, or temperature-responsive cross-linking), they solidify, transforming from a liquid state to a gel state, forming structurally stable and regularly shaped organoid microspheres, thus improving the uniformity of microsphere size and morphological stability.

[0026] The drug delivery module 20 is used to treat the organoid microspheres with at least one drug at multiple dose concentrations to form a test microsphere array, the test microsphere array comprising multiple test microspheres.

[0027] The test microspheres refer to organoid microspheres processed by the drug delivery module. After entering the drug delivery module 20, the organoid microspheres generated by the microfluidic channel chip 10 are treated with at least one drug at multiple dose concentrations, and after drug incubation, they become the test microsphere array. The test microsphere array contains multiple test microspheres, and each test microsphere or its region corresponds to a different drug type and / or drug concentration.

[0028] Specifically, the drug delivery module 20 is used to treat the prepared organoid microspheres with at least one drug at multiple dose concentrations, thereby forming a test microsphere array. In one implementation, after microsphere preparation, the drug delivery module can guide the organoid microspheres to different microchambers or microchannel branches downstream of the chip. Each chamber or branch is pre-prepared with a drug solution of different concentrations, allowing the microspheres to incubate in the corresponding environment. In another implementation, the drug delivery module can collect the organoid microspheres from the chip and transfer them to a multi-well culture plate, adding different types or concentrations of drug to each well using a liquid handling device, thereby constructing the test microsphere array. The formation of the test microsphere array allows for the parallel acquisition of microsphere response data at multiple drugs and doses in the same batch of experiments.

[0029] In this embodiment, the drug delivery module forms an array containing multiple microspheres to be tested and enables parallel processing at multiple dose concentrations, thereby improving the throughput and data acquisition efficiency of drug screening.

[0030] The optical detection module 30 is used to perform optical imaging on each of the microspheres to be tested and obtain the corresponding microsphere images.

[0031] Specifically, the optical detection module 30 includes an excitation light source, a filter assembly, and an image acquisition device. The excitation light source emits excitation light of a specific wavelength to illuminate the area where the microspheres to be tested are located, exciting the fluorescent material labeled within the microspheres to generate a fluorescence emission signal. The fluorescence signal is selectively transmitted through the target wavelength by the filter assembly and is received by the image acquisition device and converted into a digital microsphere image. Through the optical imaging process, a bright-field image reflecting the morphology of the microspheres and a microsphere image reflecting the biological response can be acquired simultaneously.

[0032] In this embodiment, optical imaging of each microsphere under test is performed by an optical detection module, ensuring the synchronous acquisition of microsphere morphological information and fluorescence response signals.

[0033] The calculation and control module 40 is communicatively connected to the optical detection module. It is used to analyze the microsphere image, determine the size information and fluorescence intensity information of each microsphere to be tested, generate drug screening results, and generate parameter control instructions when the size information and fluorescence intensity information meet preset adjustment conditions.

[0034] The computing and control module 40 can be implemented using various hardware devices or combinations thereof with data processing and program execution capabilities. It may include one or more processors and a coupled memory. The processor executes computer program instructions stored in the memory to perform functions such as analyzing and processing microsphere images, determining size and fluorescence intensity information, generating drug screening results, determining preset adjustment conditions, and generating parameter control instructions. The processor may be, for example, a central processing unit, graphics processing unit, digital signal processor, application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or microcontroller unit, all integrated circuit chips with logic operation capabilities.

[0035] The computing and control module 40 may also include a communication interface for exchanging data with the optical detection module 30 and the parameter adjustment module. The communication interface can receive microsphere image data acquired by the optical detection module and send generated parameter control commands to the parameter adjustment module. The communication interface may be, for example, a universal serial bus interface, a serial peripheral interface, an Ethernet interface, a wireless communication module, or other interface forms suitable for data transmission between devices.

[0036] Size information reflects the preparation quality indicators such as diameter distribution and morphological consistency of the current batch of organoid microspheres. Fluorescence intensity information reflects the biological response of each microsphere at the corresponding drug concentration.

[0037] The preset adjustment conditions refer to the criteria used by the calculation and control module 40 to determine whether intervention in the system operating parameters is necessary. These criteria are correlated with the microsphere image recognition and analysis results. The specific form of the preset adjustment conditions can be set according to the actual application scenario. For example, they can be set based on statistical indicators, comparing the coefficient of variation or standard deviation of the diameter of the same batch of microspheres with a preset threshold; or they can compare the average fluorescence intensity of the microspheres under test at a specific drug concentration with historical reference values ​​or internal reference standard values.

[0038] Parameter control instructions refer to the control signals or control data generated by the calculation and control module 40 after determining that preset adjustment conditions are met. These instructions contain information to guide the parameter adjustment module in adjusting the system's operating parameters.

[0039] Specifically, after analyzing the microsphere image, the calculation and control module 40 obtains the size information and fluorescence intensity information of each microsphere to be tested, and compares them with a preset reference range or target value. When the comparison result shows that the size information deviates from the preset size target range, or the fluorescence intensity information deviates from the preset response reference range, or both deviate simultaneously and the degree of deviation reaches the set judgment standard, it is determined that the preset adjustment conditions are met.

[0040] The calculation and control module 40 generates drug screening results based on the size and fluorescence intensity information of each microsphere to be tested. For example, it can plot dose-response relationships, calculate pharmacodynamic parameters, or output sensitivity classification conclusions, thereby improving the automation level of screening data processing and the consistency of results.

[0041] It should be noted that the analysis of the microsphere images by the calculation and control module 40 can be achieved using various image processing techniques. For example, color images can be converted to single-channel grayscale images through grayscale conversion, imaging noise can be suppressed through Gaussian filtering or median filtering, and the discernibility of microsphere boundaries can be enhanced through contrast-limited adaptive histogram equalization. For microsphere recognition, traditional segmentation methods based on thresholding, edge detection, or region growing can be used, or a deep learning-based instance segmentation model can be employed to handle complex situations such as microsphere adhesion, overlap, or irregular shapes.

[0042] Furthermore, the calculation and control module 40 is also used to generate parameter control instructions when the determined size information and fluorescence intensity information meet the preset adjustment conditions, so as to indicate that the system operating parameters need to be adjusted.

[0043] The parameter adjustment module 50 is connected to the calculation and control module, the microfluidic channel chip, and the drug delivery module, and is used to adjust the fluid parameters of the microfluidic channel chip and / or the drug delivery parameters of the drug delivery module in response to the parameter control command.

[0044] The parameter adjustment module includes an instruction receiving interface, a signal conversion unit, and an execution drive unit. The instruction receiving interface communicates with the calculation and control module to receive parameter control instructions. The signal conversion unit converts the received instructions into drive signals that the execution drive unit can recognize. The execution drive unit acts on the controlled object to achieve physical adjustment of fluid parameters or drug delivery parameters. When adjusting the fluid parameters of the microfluidic channel chip, the parameter adjustment module may include a fluid drive device and a fluid control valve. The fluid drive device may be, for example, an injection pump, a peristaltic pump, a pressure pump, or a micro air pump, which adjusts the flow rate of the dispersed and continuous phases by changing the drive pressure or flow rate. The fluid control valve may be, for example, a solenoid valve, a pneumatic valve, or a microvalve array, which adjusts the flow distribution ratio of each branch in the channel or the timing of crosslinking agent supply by changing the valve opening degree or on / off state. When adjusting the drug delivery parameters of the drug delivery module, the parameter adjustment module may include a liquid dispensing device and a concentration adjustment device. Liquid dispensing devices, such as multichannel syringe pump systems, automated pipetting workstations, or piezoelectrically driven microdroplet dispensers, regulate the amount and ratio of different drug solutions added by controlling the aspiration and dispensing volumes of each channel. Concentration adjustment devices, such as gradient generators or mixing valves, achieve dynamic adjustment of the target concentration by changing the mixing ratio of drug solutions of different concentrations.

[0045] Fluid parameters include the flow rate of the dispersed phase, the flow rate of the continuous phase, the flow ratio of the two-phase fluids, or the operating conditions of the crosslinking zone.

[0046] Dosing parameters include the concentration of the drug solution, the flow rate of drug introduction, or the duration of treatment of the test microspheres in the drug environment.

[0047] Specifically, the parameter adjustment module 50 is connected to the calculation and control module 40, the microfluidic channel chip 10, and the drug delivery module 20. After receiving the parameter control command, it drives the corresponding execution unit to adjust at least one of the fluid parameters and drug delivery parameters, so that the subsequent microsphere preparation and drug processing can run with the corrected parameters, thereby improving the stability of microsphere size between different preparation batches and the batch comparability of drug screening results.

[0048] Through the coordinated operation of the microfluidic channel chip 10, drug delivery module 20, optical detection module 30, calculation and control module 40, and parameter adjustment module 50, integrated operation of organoid microsphere preparation, multi-dose drug processing, optical imaging detection, image analysis, and process parameter adjustment is achieved. When the microsphere images acquired by the optical detection module are analyzed by the calculation and control module, if the microsphere size distribution deviates from the target range or the fluorescence response signal deviates from the expected level, the calculation and control module generates corresponding parameter control commands. The parameter adjustment module then executes the adjustment action to correct the preparation parameters of the microfluidic channel chip and / or the processing parameters of the drug delivery module. The corrected parameters are applied to subsequent batches of microsphere preparation and drug processing, causing the system output to gradually converge to the target state.

[0049] By generating and transmitting parameter control commands, the system transmits the quality feedback information obtained from the downstream image analysis stage to the upstream microsphere preparation stage and drug processing stage, realizing the reverse transmission of information flow from the detection end to the execution end, thereby enabling the entire system to automatically correct process deviations.

[0050] Through the above adjustment process, the calculation and control module 40, based on the statistical results of microsphere phenotypes such as fluorescence intensity information and microsphere size information, implements closed-loop feedback of the system through the parameter adjustment module 50, realizing an integrated screening process of preparation, drug loading, imaging, identification, and feedback. This reduces the variation in microsphere size between different preparation batches, improves the comparability of fluorescence quantitative results between different experimental batches, and enhances the accuracy of drug efficacy discrimination in drug screening results.

[0051] In summary, the microfluidic drug screening system provided by this invention improves the uniformity of microsphere size and morphological stability by generating hydrogel droplets and solidifying them into organoid microspheres through a microfluidic channel chip under the interaction of a continuous phase and a dispersed phase. The drug delivery module forms an array containing multiple microspheres for testing and enables parallel processing at multiple dose concentrations, improving the throughput and data acquisition efficiency of drug screening. The optical detection module performs optical imaging on each microsphere, ensuring the synchronous acquisition of microsphere morphological information and fluorescence response signals. The calculation and control module analyzes the microsphere images and generates drug screening results, improving the automation of screening data processing and the consistency of results. The parameter adjustment module adjusts fluid and drug delivery parameters in response to parameter control commands, enabling the system to correct process deviations in real time, improving the stability of microsphere size between different preparation batches and the batch-to-batch comparability of drug screening results.

[0052] In some embodiments of the present invention, the calculation and control module includes a PID controller or a model predictive controller; the PID controller is used to perform proportional, integral, and derivative operations based on the deviation between the size information and a preset size threshold, and / or the deviation between the fluorescence intensity information and a preset intensity threshold, to generate the parameter control command; the model predictive controller is used to construct a predictive model based on the size information and / or the fluorescence intensity information, and generate the parameter control command through rolling optimization solution.

[0053] A PID controller is a feedback control unit based on proportional, integral, and derivative operations. The PID controller performs proportional amplification, integral accumulation, and derivative prediction operations based on the deviation between the actual value and the target value of the controlled variable. The results of these three operations are linearly superimposed to output the control quantity, thereby eliminating the deviation and stabilizing the system at the target state. Proportional operation is used for rapid response to the current deviation, integral operation is used to eliminate steady-state error, and derivative operation is used to suppress overshoot and accelerate the system response speed.

[0054] Among them, the Model Predictive Controller (MPC) is a control unit based on a predictive model and a rolling optimization strategy to achieve closed-loop feedback control.

[0055] Specifically, when a PID controller is used, the calculation and control module compares the acquired size information of each microsphere to be tested with a preset size threshold to obtain the size deviation, and / or compares the fluorescence intensity information with a preset intensity threshold to obtain the intensity deviation. The PID controller receives the above deviation signals, performs proportional, integral, and derivative operations respectively, and generates parameter control commands based on the linear superposition of the operation results. The specific value of the parameter control command is related to the magnitude, duration, and trend of the deviation. The speed and smoothness of the control response can be adjusted by adjusting the proportional coefficient, integral time constant, and derivative time constant.

[0056] When a model predictive controller (MMC) is used, the computation and control module constructs the MMC based on historical operating data or system identification methods. This MMC describes the influence of changes in fluid or drug delivery parameters on microsphere size and fluorescence intensity. Within each control cycle, based on the currently measured size and fluorescence intensity information, the MMC uses a predictive model to extrapolate the microsphere state for several future steps and solves for an optimal control sequence that minimizes the combined size and intensity deviations within the prediction time domain, while satisfying constraints. The MMC then converts the first control variable in this control sequence into a parameter control command and outputs it to the parameter adjustment module. The rolling optimization process is repeated in the next cycle.

[0057] It should be noted that both PID controllers and model predictive controllers can be implemented using hardware circuits, embedded microprocessors, or programmable logic devices. Model predictive controllers are typically implemented by running numerical optimization solutions on industrial PCs, embedded processors, or FPGAs.

[0058] Understandably, this embodiment enables the system to perform precise closed-loop correction of parameter deviations during microsphere preparation and drug processing by configuring a PID controller or model predictive controller in the calculation and control module.

[0059] In some embodiments of the present invention, the invention further includes: a calibration and normalization module connected to the calculation and control module, used to store calibration data of reference microspheres with predetermined fluorescence intensity, and to establish a calibration model between fluorescence intensity and imaging parameters based on the calibration data; the calibration and normalization module is also used to perform cross-batch and cross-device normalization processing on the fluorescence intensity information of each microsphere to be tested according to the calibration model, and to store and trace the microsphere images, the size information of each microsphere to be tested, the fluorescence intensity information of each microsphere to be tested, the drug screening results, the fluid parameters, and the drug administration parameters.

[0060] Among them, reference microspheres refer to standard microsphere samples with known and stable fluorescence intensity, whose fluorescence intensity values ​​can be used as a benchmark for signal comparison between different imaging batches or different imaging devices.

[0061] Calibration data refers to the fluorescence intensity measurements obtained from the acquisition of reference microspheres under different imaging parameter conditions, as well as the corresponding imaging parameter records.

[0062] Imaging parameters refer to the equipment operating parameters that affect the intensity of fluorescence signal acquisition, including but not limited to the intensity of the excitation light source, the exposure time of the image sensor, the gain value of the image sensor, the wavelength of the light source, or the filter combination.

[0063] A calibration model is a mathematical mapping relationship between fluorescence intensity measurements and imaging parameters established based on calibration data. It is used to uniformly convert fluorescence intensity values ​​measured under different imaging conditions to the same reference conditions.

[0064] Specifically, the calibration and normalization module pre-stores calibration data for the reference microspheres. This calibration data is acquired as follows: before the system's formal operation or during periodic calibration, a reference microsphere with known fluorescence intensity is placed within the imaging field of view of the optical detection module. Fluorescence images of the reference microsphere are acquired under multiple different combinations of imaging parameters, and the exposure time, gain value, and measured fluorescence intensity value are recorded for each acquisition. By fitting these multiple sets of data, a calibration model is established between fluorescence intensity and exposure time / gain value.

[0065] As a preferred approach, the calibration model employs a three-dimensional calibration curve. The three-dimensional calibration curve uses exposure time as the first dimension, gain value as the second dimension, and fluorescence intensity measurement value as the third dimension, constructing a continuous functional relationship through interpolation or fitting algorithms. For any given set of exposure time and gain values, the calibration model can output a corresponding fluorescence intensity correction coefficient, which is used to correct the actually measured fluorescence intensity value to an equivalent value under standard imaging parameter conditions.

[0066] When normalizing each microsphere under test, the calculation and control module provides the calibration and normalization module with the fluorescence intensity information corresponding to each microsphere, as well as the exposure time and gain value used in this imaging. The calibration and normalization module calculates the correction coefficient corresponding to the current imaging parameters based on the calibration model, and divides or multiplies the measured fluorescence intensity value by this correction coefficient to obtain the normalized fluorescence intensity value. After normalization, fluorescence intensity values ​​measured under different batches of experiments, different imaging devices, or different imaging parameter conditions are unified to the same reference scale.

[0067] Furthermore, the calibration and normalization module also handles data storage and traceability. This module categorizes and stores microsphere images, size information of each microsphere under test, fluorescence intensity information of each microsphere under test, drug screening results, fluid parameters, and dosing parameters, and establishes an index linking these data entries. Specifically, the microsphere preparation parameters, dosing concentration configuration, imaging parameter settings, original image files, analyzed size and intensity data, and final drug screening results for each experiment are all recorded and assigned a unique batch identifier or timestamp. When it is necessary to trace or verify a specific experimental result, the complete conditions and analysis process of that experiment can be reproduced by querying the associated data stored in the calibration and normalization module.

[0068] Understandably, this embodiment establishes a calibration model between fluorescence intensity and imaging parameters in advance by setting up a calibration and normalization module. During each analysis, the fluorescence intensity information of each microsphere is normalized, eliminating signal fluctuations introduced by differences in imaging parameters and improving the comparability of quantitative fluorescence results between different experimental batches and different imaging devices. Simultaneously, through unified storage and associated management of microsphere images, size information, fluorescence intensity information, drug screening results, and various process parameters, full traceability of experimental data is achieved, facilitating subsequent data verification and quality control.

[0069] In some embodiments of the present invention, the calculation and control module is configured with a trained deep learning model for microsphere analysis; the deep learning model for microsphere analysis is used to receive the microsphere image as input, perform instance segmentation and intensity regression processing on the microsphere image to determine the size information and fluorescence intensity information corresponding to each microsphere to be tested; and based on the fluorescence intensity information corresponding to each microsphere to be tested, predict and output drug screening results, the drug screening results including IC50 value or sensitivity typing results.

[0070] Instance segmentation involves separating each individual microsphere in the image from the background and other microspheres, assigning each microsphere an independent identifier and boundary region, and simultaneously acquiring the pixel-level fluorescence intensity distribution within that region. Intensity regression involves quantifying and analyzing the fluorescence signal within each microsphere region and outputting a fluorescence intensity value.

[0071] Deep learning models can be implemented using various network architectures. For example, the U-Net network and its improved versions, based on an encoder-decoder structure, can be used. This network fuses multi-scale features through skip connections, making it suitable for segmentation tasks in biomedical images. The Mask R-CNN network can also be used, which adds a mask prediction branch to the Faster R-CNN object detection framework, enabling simultaneous output of object bounding boxes and instance segmentation masks. End-to-end detection and segmentation models based on the Transformer architecture can also be employed, utilizing a self-attention mechanism to capture global feature dependencies in microsphere images, improving the recognition accuracy of adhered or irregularly shaped microspheres. Model development can be implemented using open-source deep learning frameworks such as PyTorch or TensorFlow.

[0072] Specifically, the training process of the deep learning model for microsphere analysis is as follows: First, a training dataset is constructed by acquiring bright-field and fluorescence images containing organoid microspheres. Each microsphere in the images is then manually or semi-automatically labeled, including its boundary mask and corresponding fluorescence intensity label. To improve the model's generalization ability, data augmentation techniques can be used to expand the training samples, including random rotation, mirror flipping, scaling, and brightness and contrast adjustments. As a preferred approach, generative adversarial networks can be used to synthesize simulated microsphere images with different morphological, texture, and fluorescence distribution characteristics. These synthesized images are then mixed with real images and used as training data to enhance the model's robustness in complex scenarios.

[0073] During model training, the loss function can be jointly optimized by combining segmentation loss and regression loss. Dice Loss can be used for segmentation loss to measure the overlap between the predicted and labeled masks. L1 Loss or mean squared error loss can be used for regression loss to measure the deviation between the predicted and labeled fluorescence intensity values. The network parameters are iteratively updated using the backpropagation algorithm until the model's performance on the validation set reaches the preset convergence criteria.

[0074] After training, the deep learning model for microsphere analysis is deployed in the computation and control module. During the inference phase, the model receives microsphere images acquired in real time by the optical detection module. After forward propagation calculations, it synchronously outputs the instance segmentation mask, bounding box, and corresponding fluorescence intensity value for each microsphere under test. Based on the fluorescence intensity information of each microsphere under test at different drug concentrations, the model further predicts the IC50 value of the drug for organoid microspheres or outputs sensitivity classification results, such as sensitive, moderately sensitive, or resistant, through fully connected layers or temporal modeling branches.

[0075] Understandably, this embodiment, through its end-to-end processing method, reduces information loss in intermediate steps and improves the efficiency of drug efficacy evaluation compared to the traditional step-by-step process of first quantifying fluorescence intensity and then performing curve fitting.

[0076] In some embodiments of the present invention, the drug delivery module includes a plurality of drug delivery inlets disposed inside the microfluidic channel chip and a microfluidic network communicating with the drug delivery inlets; the drug delivery inlets are used to introduce drugs of different concentrations; the microfluidic network is used to form the plurality of dose concentrations within the microfluidic channel chip; and / or, the drug delivery module includes an automated drug delivery unit connected downstream of the microfluidic channel chip; the automated drug delivery unit is used to form the plurality of dose concentrations on the culture plate when the organoid microspheres are transferred to the culture plate after leaving the chip, so that the organoid microspheres transferred to the culture plate receive drug treatment of corresponding concentrations.

[0077] Specifically, in the method of achieving dose concentration within a microfluidic channel chip, the microfluidic channel chip is equipped with multiple independent drug delivery inlets. Each drug delivery inlet is supplied with a solution of the same drug at different concentrations, or with solutions of different types of drugs. Downstream of each drug delivery inlet is a microfluidic network. This network can adopt a tree-like or mesh-like structure with hierarchical branching and merging. By controlling the flow resistance or flow rate ratio of each branch, a concentration variation along or perpendicular to the flow direction can be formed within the main channel or the microsphere capture area. For example, multiple drug delivery branches can be symmetrically arranged on both sides of the main channel of the chip, each branch supplying a drug solution with a gradient of concentration. Through laminar diffusion, a concentration gradient distribution perpendicular to the flow direction can be formed within the main channel. Alternatively, multiple drug delivery inlets can be sequentially arranged on the main channel, each inlet supplying a drug solution of different concentrations, creating a step-like concentration variation along the flow direction within the main channel. When organoid microspheres flow through or remain in different concentration regions, they receive drug treatment of corresponding concentrations, thereby enabling parallel screening of multiple dose concentrations within the same chip.

[0078] As a preferred approach, microchannel networks can form a logarithmically scaled concentration gradient. By setting the correspondence between the flow resistance ratio of each branch and the target concentration gradient distribution, the drug concentration ratio of adjacent regions within the main channel can be kept essentially constant, thereby more reasonably covering the concentration range required for drug screening.

[0079] In the method of achieving dose concentration externally using a microfluidic channel chip, the drug delivery module includes an automated drug delivery unit connected downstream of the microfluidic channel chip. After the organoid microspheres prepared by the microfluidic channel chip are collected from the chip outlet, they are transferred to a multi-well culture plate. The automated drug delivery unit automatically dispenses drug solutions into each well of the culture plate, adding different types or concentrations of drug solutions to different wells, thereby creating multiple dose concentrations on the culture plate. The organoid microspheres receive the corresponding drug concentration in each well. This facilitates longer incubation periods and multiple time-point imaging detection.

[0080] The two implementation methods described above can be used individually or simultaneously to form a dual-modal working mode combining in-chip gradient screening and screening on the outer chip substrate. The array configuration on the outer chip substrate facilitates subsequent long-term incubation, in-situ imaging, and high-throughput drug efficacy analysis.

[0081] Understandably, the use of the two methods alone or in combination improves the applicability and experimental flexibility of the drug screening system in different application scenarios, while ensuring the throughput and reproducibility of the screening process under multiple drug and multiple dose conditions.

[0082] In some embodiments of the present invention, the optical detection module includes: a multi-band excitation light source for emitting excitation light of different bands to excite the microsphere under test to generate a fluorescence signal; an emission filter group for selectively transmitting the fluorescence signal of a preset band; an image sensor for receiving the fluorescence signal transmitted through the emission filter group to obtain an original microsphere image; and an image correction unit for performing flat field correction and dark field correction on the original microsphere image to eliminate uneven illumination and dark current noise, thereby generating the microsphere image.

[0083] A multi-band excitation source refers to a light source device capable of emitting excitation light with multiple different center wavelengths. Multi-band excitation sources can be implemented using light-emitting diode arrays, laser arrays, or broadband light sources with excitation filter wheels.

[0084] An emission filter array is a combination of optical filtering elements used to selectively transmit fluorescence signals in a preset wavelength band. The emission filter array is placed in the receiving optical path of the fluorescence signal to filter out reflected and stray light from the excitation light, allowing only the target fluorescence wavelength band to pass through.

[0085] An image sensor is a detection device that converts received light signals into digital image signals. It can employ complementary metal-oxide-semiconductor (CMOS) image sensors, scientific-grade CMOS image sensors, or charge-coupled device (CCD) image sensors. Image sensors possess exposure control functions that are synchronously triggered with the excitation light source to adapt to the exposure time requirements of different fluorescence channels.

[0086] Flat-field correction refers to the process of compensating for brightness deviations in an image caused by uneven illumination, while dark-field correction refers to the process of subtracting dark current noise from the image sensor. Specifically, flat-field correction uses an image of a uniformly fluorescent reference sample as a flat-field image, and divides the intensity value of each pixel in the original microsphere image by the normalized intensity value of the corresponding pixel in the flat-field image, thereby eliminating spatial brightness differences caused by uneven illumination. Dark-field correction, on the other hand, acquires a dark-field image with the excitation source turned off, and subtracts the intensity value of the corresponding pixel in the dark-field image from the intensity value of each pixel in the original microsphere image, thereby eliminating background signals caused by dark current and readout noise from the image sensor.

[0087] Specifically, in a multi-channel fluorescence imaging scenario, multi-band excitation sources emit excitation light of different bands sequentially or simultaneously, and the emission filter group synchronously switches the emission filters of the corresponding bands. The image sensor acquires raw microsphere images in each channel. The image correction unit performs correction processing on the raw images of each channel, generating corrected microsphere images for each channel. The corrected multi-channel microsphere images can reflect the signal intensity of the microspheres under different fluorescent labels, such as live cell fluorescent labeling signals, apoptotic cell fluorescent labeling signals, or specific target protein fluorescent labeling signals, thus providing an image data basis for multi-dimensional phenotypic analysis of the microspheres under test.

[0088] Understandably, this embodiment achieves multi-channel fluorescence imaging of the microspheres under test through the combination of a multi-band excitation light source and an emission filter group, enabling the simultaneous acquisition of fluorescence signals reflecting different biological indicators. By performing flat-field and dark-field corrections on the original microsphere images through an image correction unit, the interference of illumination field inhomogeneity and dark current noise from the image sensor on the fluorescence quantification results is eliminated, improving the consistency and comparability of fluorescence intensity measurements across different fields of view and batches.

[0089] In some embodiments of the present invention, the calculation and control module includes: a preprocessing unit for preprocessing the microsphere image, the preprocessing including grayscale conversion, filtering and noise reduction, and contrast enhancement; a segmentation and recognition unit for identifying and extracting the target region of a single microsphere to be tested using an image segmentation algorithm; a feature extraction unit for calculating the size information and fluorescence intensity information corresponding to each microsphere to be tested based on the target region; a normalization unit for normalizing the fluorescence intensity information; and a drug efficacy fitting unit for fitting the dose-response curve using a four-parameter Logistic equation based on the normalized fluorescence intensity information to obtain the IC50 value or area under the curve as the drug screening result.

[0090] The features extracted by the feature extraction unit include size information and fluorescence intensity information.

[0091] The IC50 value refers to the concentration at which a drug produces the half-maximal inhibitory effect.

[0092] The area under the curve (AUC) is the area enclosed by the dose-response curve and the coordinate axes. It can be used to comprehensively evaluate the overall effect intensity of a drug at different concentrations.

[0093] Specifically, noise reduction can be achieved using a Gaussian filtering algorithm, which smooths the image through convolution kernels and suppresses random noise introduced during imaging. Contrast enhancement can be achieved using a contrast-limited adaptive histogram equalization algorithm, which enhances the contrast between the microsphere boundary and the background by equalizing the gray-level distribution in local areas of the image.

[0094] The segmentation and recognition unit performs image segmentation algorithms on the preprocessed image to identify and extract the target region of individual microspheres. As one implementation, the segmentation and recognition unit first uses a local adaptive thresholding method to binarize the image, separating the microsphere region from the background region. For adjacent or partially overlapping microspheres, a distance transform and watershed algorithm can be further combined for separation. The distance transform uses the distance from each foreground pixel to its nearest background pixel in the binary image as the grayscale value of that pixel, forming a distance map. The watershed algorithm uses the distance map as a terrain height map, generating watershed ridges at the junctions of the adhered microspheres by simulating a water immersion process, thereby achieving the separation of adhered targets. After segmentation, the segmentation results are labeled with connected components, with each connected component corresponding to the target region of a microsphere.

[0095] To further ensure the reliability of the analysis results, the segmentation and recognition unit can also introduce a quality control mechanism. Quality control is based on the morphological characteristics of the microsphere target region to filter out target regions that do not meet preset conditions. The filtering conditions may include an area threshold range, a roundness threshold range, and a boundary contact determination rule. The area threshold is used to exclude noise particles that are too small and microsphere clusters that are too large; the roundness threshold is used to exclude irregularly shaped or broken microspheres; and the boundary contact determination rule is used to exclude incomplete microspheres that are truncated at the image edge.

[0096] The normalization unit normalizes the fluorescence intensity information output by the feature extraction unit. The normalization process is based on a pre-established three-dimensional calibration model, which describes the correspondence between fluorescence intensity and exposure time and gain value. For the fluorescence intensity measurement of the microsphere under test, the normalization unit maps it to the equivalent fluorescence intensity value under standard imaging conditions according to the exposure and gain parameters used during imaging, thereby achieving signal normalization across batches and devices.

[0097] The pharmacodynamics fitting unit uses a four-parameter logistic equation to fit the response data of each drug at each dose concentration based on normalized fluorescence intensity information. The fitting process employs nonlinear least squares optimization methods such as the Levenberg-Marquardt algorithm, iteratively searching to minimize the sum of squared residuals between the fitted curve and the measured data points. After fitting, the IC50 value and area under the curve are extracted from the model parameters as drug screening results. The pharmacodynamics fitting unit can also classify drug responses into different subtypes such as sensitive, moderately sensitive, or resistant based on the comparison between the IC50 value and a preset threshold, providing a reference for clinical medication decisions.

[0098] like Figure 2The diagram shows the flowchart of the microsphere image recognition algorithm. The microsphere image goes through background correction, noise reduction and enhancement, segmentation and recognition, quality control, intensity quantification, normalization processing and drug efficacy fitting in sequence, and finally outputs the IC50 value and sensitivity classification results.

[0099] like Figure 3 The figure shown is a schematic diagram of the dose-response curve. Figure 3 The discrete data points represent the measured average fluorescence intensity values ​​at each concentration level. Figure 3 The continuous curve is a dose-response curve obtained by fitting the measured data points using a four-parameter Logistic equation. The curve shows a typical S-shaped trend of fluorescence intensity gradually decreasing with increasing drug concentration, reflecting the dose-dependent inhibitory effect of the drug on organoid microspheres.

[0100] like Figure 4 The figure shows a heatmap diagram of multi-drug, multi-dose screening, which represents the effect intensity of the drug-dose matrix. As can be seen from the figure, the dose concentration distribution shows that the drug concentration exhibits a gradient change at different positions in the main channel of the microfluidic chip. Therefore, it can be used to expose organoid microspheres in different regions to different dose environments.

[0101] In some embodiments of the present invention, the microfluidic channel chip adopts a confluence focusing or T-shaped confluence structure, with the nozzle having an equivalent minimum width of 50-150 μm to form organoid microspheres with a diameter of 30-60 μm; the downstream of the microfluidic channel chip is provided with an optical window, the crosslinking region, and a sorting or collecting branch; the dispersed phase is a crosslinkable hydrogel solution containing sodium alginate, and the continuous phase is a biocompatible oil phase containing surfactant; the crosslinking region is used to provide chemical or photocrosslinking conditions for solidifying the hydrogel droplets.

[0102] Specifically, the microfluidic channel chip employs a confluence focusing or T-junction structure. The chip's nozzle has an equivalent minimum width of 50–150 μm. By adjusting the flow rate ratio of the dispersed phase and the continuous phase, hydrogel droplets with diameters of 30–60 μm can be stably generated within this size range. The generated droplets move downstream with the continuous phase, undergo a solidification reaction upon entering the cross-linking zone, and form organoid microspheres. Downstream of the chip, after the cross-linking zone, are sequentially arranged optical windows and sorting or collecting branches. The optical windows correspond to the imaging position of the optical detection module, and the sorting or collecting branches are used to export the prepared organoid microspheres from the chip or to divert them as needed.

[0103] The dispersed phase is a crosslinkable hydrogel solution containing sodium alginate, and its mass-volume fraction can be adjusted within the range of 0.5%–2% (w / v) according to the mechanical strength and cell compatibility of the target microspheres. The continuous phase is a biocompatible oil phase containing surfactants. When the two fluid phases converge in the nozzle area, the dispersed phase is broken into uniform droplets under the shearing action of the continuous phase. After the droplets enter the crosslinking zone, they come into contact with the crosslinking agent or are exposed to light, undergoing chemical or photocrosslinking reactions, transforming from liquid hydrogel droplets into solid gel microspheres. Chemical crosslinking can be achieved by introducing calcium ion solution into the crosslinking zone, while photocrosslinking can be achieved by setting an ultraviolet light source in the crosslinking zone and selecting a photosensitive hydrogel material. By using a hydrogel solution containing sodium alginate as the dispersed phase and a biocompatible oil phase containing surfactants as the continuous phase, the stability of the droplet generation process and the activity of the embedded cells or organoids are ensured. By sequentially setting optical windows, cross-linking regions, and sorting or collection branches downstream of the chip, droplet generation, solidification, imaging detection, and sample collection can be completed continuously within the same chip, reducing sample transfer steps and improving operational efficiency and data repeatability.

[0104] like Figure 5 As shown, the present invention also provides a microfluidic-based drug screening method. Figure 5 This is a schematic flowchart of an embodiment of the microfluidic drug screening method provided by the present invention. The method includes: S501. Within the microfluidic channel chip, hydrogel droplets containing organoids or cells are generated through the confluence shearing action of the continuous phase and the dispersed phase, and the hydrogel droplets are solidified in the cross-linking region to form organoid microspheres. S502. The organoid microspheres are treated with at least one drug at multiple dose concentrations through the drug delivery module to form a test microsphere array, wherein the test microsphere array includes multiple test microspheres. S503. Optical imaging is performed on each of the microspheres to be tested through the optical detection module to obtain the corresponding microsphere image; S504. The calculation and control module analyzes the microsphere image to determine the size information and fluorescence intensity information of each microsphere to be tested, and generates drug screening results. S505. When the size information and fluorescence intensity information meet the preset adjustment conditions, the calculation and control module generates parameter control instructions, and adjusts the fluid parameters of the microfluidic channel chip and / or the drug delivery parameters of the drug delivery module in real time through the parameter adjustment module.

[0105] It should be noted that in the embodiments of the microfluidic drug screening method described above, please refer to the corresponding description in the microfluidic drug screening system above, and the beneficial effects that can be achieved should also refer to the beneficial effects described above, which will not be repeated here.

[0106] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.), and the computer program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0107] The microfluidic drug screening system, method, and storage medium provided by the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A microfluidic drug screening system, characterized in that, include: Microfluidic channel chips are used to generate hydrogel droplets containing organoids or cells under the action of continuous and dispersed phases, and solidify in the cross-linking region to form organoid microspheres; A drug delivery module is used to treat the organoid microspheres with at least one drug at multiple dose concentrations to form a test microsphere array, the test microsphere array comprising a plurality of test microspheres; The optical detection module is used to perform optical imaging on each of the microspheres to be tested and obtain the corresponding microsphere images; The calculation and control module is communicatively connected to the optical detection module. It is used to analyze the microsphere image, determine the size information and fluorescence intensity information of each microsphere to be tested, generate drug screening results, and generate parameter control instructions when the size information and fluorescence intensity information meet the preset adjustment conditions. A parameter adjustment module, connected to the calculation and control module, the microfluidic channel chip, and the drug delivery module, is used to adjust the fluid parameters of the microfluidic channel chip and / or the drug delivery parameters of the drug delivery module in response to the parameter control command.

2. The microfluidic drug screening system according to claim 1, characterized in that, The calculation and control module includes a PID controller or a model predictive controller; The PID controller is used to perform proportional, integral, and derivative operations based on the deviation between the size information and the preset size threshold, and / or the deviation between the fluorescence intensity information and the preset intensity threshold, to generate the parameter control command. The model prediction controller is used to construct a prediction model based on the size information and / or the fluorescence intensity information, and generate the parameter control instructions through rolling optimization solution.

3. The microfluidic drug screening system according to claim 1, characterized in that, Also includes: The calibration and normalization module, connected to the calculation and control module, is used to store calibration data of reference microspheres with predetermined fluorescence intensity, and to establish a calibration model between fluorescence intensity and imaging parameters based on the calibration data. The calibration and normalization module is also used to perform cross-batch and cross-device normalization processing on the fluorescence intensity information of each microsphere to be tested according to the calibration model, and to store and trace the microsphere image, the size information of each microsphere to be tested, the fluorescence intensity information of each microsphere to be tested, the drug screening results, the fluid parameters and the drug administration parameters.

4. The microfluidic drug screening system according to claim 1, characterized in that, The computing and control module is equipped with a trained deep learning model for microsphere analysis; The deep learning model for microsphere analysis receives the microsphere image as input and performs instance segmentation and intensity regression processing on the microsphere image to determine the size and fluorescence intensity information of each microsphere to be tested; and Based on the fluorescence intensity information corresponding to each microsphere to be tested, the drug screening results are predicted and output, including IC50 values ​​or sensitivity typing results.

5. The microfluidic drug screening system according to claim 1, characterized in that, The drug delivery module includes multiple drug delivery inlets disposed inside the microfluidic channel chip and a microfluidic network connected to the drug delivery inlets; The administration inlet is used to introduce the drug at different concentrations; The microfluidic network is used to form the plurality of dose concentrations within the microfluidic channel chip; And / or, The drug delivery module includes an automated drug delivery unit connected downstream of the microfluidic channel chip; The automated drug delivery unit is used to form the multiple dose concentrations on the culture plate after the organoid microspheres leave the chip and are transferred to the culture plate, so that the organoid microspheres transferred to the culture plate receive drug treatment of the corresponding concentration.

6. The microfluidic drug screening system according to claim 1, characterized in that, The optical detection module includes: A multi-band excitation light source is used to emit excitation light of different bands to excite the microspheres under test to generate fluorescence signals; An emission filter array is used to selectively transmit the fluorescence signal in a preset wavelength band; An image sensor is used to receive fluorescence signals transmitted through the emission filter group to obtain an image of the original microspheres; An image correction unit is used to perform flat field correction and dark field correction on the original microsphere image to eliminate uneven illumination and dark current noise, thereby generating the microsphere image.

7. The microfluidic drug screening system according to claim 1, characterized in that, The computing and control module includes: The preprocessing unit is used to preprocess the microsphere image, and the preprocessing includes grayscale conversion, filtering and noise reduction, and contrast enhancement. The segmentation and recognition unit is used to identify and extract the target region of a single microsphere under test using an image segmentation algorithm; The feature extraction unit is used to calculate the size information and fluorescence intensity information of each microsphere to be tested based on the target region; A normalization unit is used to normalize the fluorescence intensity information; The drug efficacy fitting unit is used to fit the dose-response curve using a four-parameter Logistic equation based on the normalized fluorescence intensity information, and obtain the IC50 value or area under the curve as the drug screening result.

8. The microfluidic drug screening system according to claim 1, characterized in that, The microfluidic channel chip adopts a confluence focusing or T-shaped confluence structure, with the nozzle having an equivalent narrowest width of 50-150 μm to form organoid microspheres with a diameter of 30-60 μm; The downstream side of the microfluidic channel chip is provided with an optical window, the crosslinking region and a sorting or collection branch; The dispersed phase is a crosslinkable hydrogel solution containing sodium alginate, and the continuous phase is a biocompatible oil phase containing surfactant. The crosslinking region is used to provide chemical or photocrosslinking conditions for the solidification of the hydrogel droplets.

9. A microfluidic drug screening method, characterized in that, The microfluidic drug screening system applied to any one of claims 1 to 8 comprises: Within the microfluidic channel chip, hydrogel droplets containing organoids or cells are generated through the confluence shearing action of the continuous phase and the dispersed phase, and the hydrogel droplets are solidified in the cross-linking region to form organoid microspheres. The drug delivery module allows the organoid microspheres to be treated with at least one drug at multiple dose concentrations to form a test microsphere array, which includes multiple test microspheres. The optical detection module performs optical imaging on each of the microspheres to be tested to obtain the corresponding microsphere images. The calculation and control module analyzes the microsphere images to determine the size and fluorescence intensity information of each microsphere to be tested, and generates drug screening results. When the size information and fluorescence intensity information meet the preset adjustment conditions, the calculation and control module generates parameter control instructions, and adjusts the fluid parameters of the microfluidic channel chip and / or the drug delivery parameters of the drug delivery module in real time through the parameter adjustment module.

10. A computer-readable storage medium, characterized in that, Used to store computer-readable programs or instructions that, when executed by a processor, can implement the steps in the microfluidic drug screening method described in claim 9.

Images