Pathogen rapid phenotype drug sensitivity detection system and application

Abstract

A rapid phenotypic drug susceptibility detection system and application for pathogens includes: a pathogenic microorganism enrichment culture module for pathogen capture, isolation, washing, purification, and culture; a drug susceptibility reaction module located on the other side; a dynamic microscopic imaging module located relative to the drug susceptibility reaction module; and an AI phenotypic analysis module connected to the drug susceptibility reaction module. This invention can achieve rapid enrichment and accurate phenotypic drug susceptibility detection of high-purity, highly active pathogens, shorten the detection cycle, and improve the accuracy of results.

Description

Technical Field

[0001] This invention relates to a technology in the field of bioanalysis, specifically to a rapid phenotypic drug susceptibility detection system for pathogens based on broad-spectrum magnetic bead enrichment and automated microscopic imaging, and its application. Background Technology

[0002] Antimicrobial susceptibility testing is a core basis for guiding anti-infective medication and curbing bacterial resistance in clinical practice. Currently, mainstream clinical antimicrobial susceptibility testing technologies have significant shortcomings: existing technologies employ fractional enrichment and static counting methods, relying on single-point bacterial counts for susceptibility assessment. This fails to continuously track pathogen morphology, division status, and dynamic growth trends, resulting in low sensitivity and a lack of standardized AI-based phenotypic interpretation systems. Furthermore, it involves significant human intervention and poor result consistency. Existing neural network-based image-assisted processing technologies only segment and classify static antimicrobial susceptibility plate images, failing to incorporate dynamic temporal imaging, pathogen growth rate quantification, automatic MIC calculation, and addressing issues such as interference from impurities in blood samples and enrichment of low-load pathogens. Summary of the Invention

[0003] This invention addresses the shortcomings of existing drug susceptibility testing methods, such as long detection cycles, significant imaging interference, distorted results, and inability to adapt to low-load blood samples. It proposes a rapid phenotypic drug susceptibility testing system and its application, enabling rapid enrichment and accurate phenotypic drug susceptibility testing of high-purity, highly active pathogens, thereby shortening the detection cycle and improving the accuracy of results.

[0004] This invention is achieved through the following technical solution:

[0005] This invention relates to a rapid phenotypic drug susceptibility detection system for pathogens, comprising: a pathogenic microorganism enrichment culture module for pathogen capture, isolation, washing, purification, and culture; a drug susceptibility reaction module located on the other side; a dynamic microscopic imaging module located relative to the drug susceptibility reaction module; and an AI phenotypic analysis module connected to the drug susceptibility reaction module. The AI ​​phenotypic analysis module integrates a cell image segmentation unit, a morphological feature extraction unit, a growth rate analysis unit, a MIC calculation unit, and a drug susceptibility result grading and interpretation unit. The cell image segmentation unit performs foreground-background separation and impurity removal on time-series microscopic images, accurately segmenting single cells and microbial community regions. The morphological feature extraction unit extracts multi-dimensional phenotypic features such as cell outline, size, division state, and aggregation morphology. The growth rate analysis unit quantifies the cell number growth rate, community area expansion rate, and division activity based on time-series images. The MIC calculation unit automatically calculates the minimum inhibitory concentration (MIC) by combining the growth inhibition rate at different drug concentrations. The drug susceptibility result grading and interpretation unit automatically determines the three levels of sensitivity (S), intermediate (I), and resistance (R) according to the CLSI standard.

[0006] The pathogenic microorganism enrichment and culture module includes: a linear motion module for horizontal movement, and several rows of pathogen capture chambers, washing chambers, purification chambers, culture chambers arranged thereon, a temperature control device located outside the culture chamber, and a magnetic separation device arranged vertically above the culture chamber.

[0007] The pathogen capture cavity is pre-embedded with broad-spectrum antibody magnetic beads.

[0008] This invention relates to an application based on the above-mentioned system, namely, rapid phenotypic drug susceptibility detection of pathogens through broad-spectrum magnetic bead enrichment and automated microscopic imaging. The sample is proliferated through a pathogen enrichment culture module and incubated in a drug susceptibility system through a drug susceptibility reaction module. Dynamic microscopic imaging is then performed through a dynamic microscopic imaging module, and the continuous image data is processed by an AI phenotypic analysis module to extract core parameters such as bacterial cell division rate, community area, and bacterial morphology. The minimum inhibitory concentration (MIC) value is calculated, and the pathogen drug susceptibility result (S / I / R) is determined according to the CLSI standard, thus completing the phenotypic drug susceptibility detection.

[0009] The AI ​​phenotypic analysis and automatic drug sensitivity assessment method, combining dynamic time-series image features and microbial growth patterns, achieves automated image recognition, phenotypic activity quantification, and intelligent drug sensitivity assessment, specifically including:

[0010] Step 1: Image preprocessing and precise bacterial cell segmentation are performed using the bacterial cell image segmentation unit, specifically including:

[0011] 1.1 Receive continuous time-series images uploaded by the dynamic microscopic imaging module, perform grayscale conversion, noise reduction, and illumination equalization preprocessing to eliminate imaging spots, background impurities, and interference from residual blood cells / proteins;

[0012] 1.2 An improved semantic segmentation network is used to segment the image at the pixel level, distinguishing the background region, magnetic bead region, dead bacterial body region, live bacterial body and microbial community region; invalid interference regions such as residual magnetic beads and cell debris are removed, and only the target live pathogen region is retained;

[0013] 1.3 Perform morphological opening and closing operations on the segmentation results, repair the breakpoints of the bacterial cell outline, and output the binarized segmentation images of single bacterial cells and bacterial communities as the base images for subsequent feature extraction.

[0014] Step 2: Extract multidimensional bacterial morphological features using the morphological feature extraction unit: Based on the segmented image, extract geometric features: major axis, minor axis, perimeter, area, circularity, and aspect ratio of a single bacterial cell; identify bacterial cell division morphology, budding morphology, and chain / cluster aggregation morphology; mark the proportion of cells in the division phase and the degree of bacterial community aggregation; count the total number of effective viable cells, the number of bacterial communities, and the total area of ​​the bacterial community in a single frame image to establish a single frame feature dataset.

[0015] Step 3: Quantitatively calculate dynamic growth parameters using the dynamic growth rate analysis unit: Using the acquisition time as the axis, perform time-series matching on the number of bacteria and the area of ​​the bacterial community in multiple consecutive frames of images to construct a time-growth characteristic curve; calculate the growth rate of the number of bacteria and the expansion rate of the bacterial community area, and calculate the comprehensive growth activity value G. The value of G ranges from 0 to 1. The closer G is to 1, the stronger the growth activity of the pathogen. The closer G is to 0, the more significantly growth is inhibited.

[0016] Step 4: Automatically calculate the minimum inhibitory concentration (MIC) using the MIC calculation unit: Extract the growth activity values ​​G corresponding to different gradient antimicrobial drug concentrations on the drug susceptibility test plate and compare them with the preset growth inhibition threshold G0. In this system, G0 is set to 0.1. Growth activity below the threshold is considered growth inhibition. The G values ​​at each drug concentration are compared sequentially, and the lowest drug concentration that can completely inhibit the normal growth of the pathogen is determined as the minimum inhibitory concentration (MIC) of the antimicrobial drug corresponding to that pathogen. If the pathogen still maintains high growth activity at the highest drug concentration, it is marked as the MIC exceeding the detection range.

[0017] Step 5: Automatic grading and interpretation of drug susceptibility results via the drug susceptibility result grading and interpretation unit: Based on the clinical CLSI drug susceptibility judgment standard, the calculated MIC value is compared with the corresponding clinical critical concentration threshold of the drug to complete the three-level result judgment: Sensitive (S): MIC ≤ clinical sensitive critical concentration, the antimicrobial drug is determined to be effective against the target pathogen; Intermediate (I): clinical sensitive critical concentration < MIC < clinical resistance critical concentration, the drug efficacy is determined to be limited, and comprehensive clinical medication is required; Resistant (R): MIC ≥ clinical resistance critical concentration, the target pathogen is determined to be resistant to the antimicrobial drug, and the drug is ineffective; The system automatically generates a test report and simultaneously outputs the original image, growth curve, MIC value, and S / I / R grading results.

[0018] Technical effect

[0019] This invention designs an array-type cavity + magnetic separation automated structure for blood samples. Without damaging the biological activity of pathogens, it thoroughly removes interfering impurities such as blood cells, proteins, and lipids. It can improve the detection rate of low-load pathogens by proliferating in just 2 hours. It solves the problems of cluttered background and missed detection of low-load samples in existing microscopic imaging. At the same time, it constructs a dedicated AI model for the entire chain of segmentation, morphology extraction, dynamic growth quantification, MIC calculation and drug sensitivity grading. It uses time-series images to track the real growth and division state of pathogens and uses growth activity as the core indicator to quantify the drug inhibition effect, rather than relying on single-point quantity statistics. The algorithm combines microbial phenotypic characteristics and industry CLSI standards to realize fully automated determination of MIC and S / I / R results, eliminating the error of human interpretation.

[0020] Compared with existing technologies, this invention has the following significant technical advantages: 1. Improved imaging quality and recognition accuracy: After enrichment and impurity removal by pre-mounted magnetic beads, the background noise of microscopic imaging is extremely low. Combined with AI-based precise segmentation algorithms, the bacterial cell recognition accuracy can reach over 99%, completely solving the industry pain point of blood sample impurities interfering with imaging; 2. Significantly reduced detection cycle: Existing phenotypic drug sensitivity testing cycles are 18-24 hours. This invention, through 2 hours of short-term proliferation + 3-4 hours of drug incubation + real-time AI analysis, controls the overall detection cycle to within 6 hours, meeting the clinical needs of rapid drug selection for critically ill patients; 3. Reliable and accurate results, avoiding false bias: Based on pathogen dynamics... Phenotypic assessment based on growth activity and morphological changes directly reflects the true antibacterial effect of drugs, overcoming the limitation of gene-based drug sensitivity testing which only detects drug resistance genes and cannot reflect the actual phenotype; 4. High degree of automation and good repeatability: The entire process from sample enrichment, drug sensitivity reaction, imaging to AI analysis and drug sensitivity interpretation is automated, eliminating the need for manual counting and interpretation, reducing human error, and ensuring strong consistency of test results, making it suitable for batch clinical sample testing; 5. Suitable for low-bacterial-load clinical samples: Targeting the low pathogen load in clinical blood samples, specific magnetic bead enrichment combined with short-term proliferation increases bacterial load, resulting in a significantly higher detection rate of low-bacterial-load pathogens than existing direct microscopic detection methods. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the invention;

[0022] Figure 2 This is a schematic diagram of the overall structure of the present invention;

[0023] Figure 3 A schematic diagram of a pathogen enrichment culture module;

[0024] Figure 4 This is a schematic diagram of the drug sensitivity reaction module;

[0025] Figure 5 This is a schematic diagram of a dynamic microscopic imaging module;

[0026] In the figure: pathogen enrichment culture module 10, drug sensitivity reaction module 20, dynamic microscopic imaging module 30, pathogen capture chamber 101, washing chamber 102, purification chamber 103, culture chamber 104, temperature control device 105, linear motion module 106, magnetic separation device 107, drug sensitivity reaction plate 201, constant temperature incubation unit 202, CMOS image acquisition unit 301, autofocus module 302, objective lens 303. Detailed Implementation

[0027] like Figure 2As shown, this embodiment relates to a rapid phenotypic drug susceptibility detection system for pathogens, including: a pathogenic microorganism enrichment culture module 10 for pathogen capture, separation, washing, purification and culture, a drug susceptibility reaction module 20 disposed on the other side, and a dynamic microscopic imaging module 30 disposed relative to the drug susceptibility reaction module 20.

[0028] like Figure 3 As shown, the pathogenic microorganism enrichment culture module 10 includes: a linear motion module 106 for horizontal movement and several rows of pathogen capture chambers 101, washing chambers 102, purification chambers 103, culture chambers 104 arranged thereon, a temperature control device 105 located outside the culture chamber 104, and a magnetic separation device 107 arranged vertically above the culture chamber 104.

[0029] The pathogen capture cavity 101 is pre-embedded with broad-spectrum antibody magnetic beads.

[0030] like Figure 4 As shown, the drug sensitivity reaction module 20 includes: a constant temperature incubation unit 202 and a drug sensitivity reaction plate 201 disposed thereon.

[0031] like Figure 5 As shown, the dynamic microscopic imaging module 30 is disposed above the drug sensitivity reaction module 20, and includes: an autofocus module 302, a CMOS image acquisition unit 301 disposed thereon, and an objective lens 303 located at the end of the CMOS image acquisition unit 301.

[0032] like Figure 1 As shown, this embodiment relates to the application of the above system, including:

[0033] 1) After sample collection, the sample is added to the capture chamber of the pathogen enrichment culture module pre-embedded with broad-spectrum antibody magnetic beads, in coordination with the linear motion module 106 and the magnetic separation device 107. After incubation at room temperature with shaking for 10-30 minutes, all types of target pathogens in the blood sample are captured through antigen-antibody specific binding. Using the magnetic separation device, the magnetic bead-pathogen complex is transferred to a washing and purification chamber pre-filled with washing solution to remove free blood components and other soluble impurities. The washed and purified magnetic bead-pathogen complex is then transferred to a culture chamber pre-filled with liquid culture medium and incubated at 37°C for approximately 2 hours to allow the pathogens to proliferate exponentially.

[0034] 2) After the pathogens have completed proliferation, they are transferred to the drug sensitivity reaction module 20. The bacterial solution is added to the microwell reaction plate 201 containing gradient concentrations of antibacterial drugs. After the plate is closed, it is placed in the constant temperature incubation unit 202 for continuous incubation for 3-4 hours.

[0035] 3) Using the dynamic microscopic imaging module 30, the micro-aperture field of view is automatically focused and image acquired every 15-30 minutes, recording the morphology, quantity, and dynamic changes of pathogen cell growth throughout the process; the continuous image data is processed by the AI ​​phenotypic analysis algorithm to extract core parameters such as cell division rate, community area, and cell morphology, calculate the minimum inhibitory concentration (MIC) value, and determine the pathogen drug susceptibility results (S / I / R) according to the CLSI standard to complete the phenotypic drug susceptibility detection.

[0036] Practical application scenarios were tested using simulated blood samples containing Staphylococcus aureus: a dynamic microscopic imaging module continuously acquired data at a frequency of 20 minutes / frame for 4 hours (12 time points in total). The AI ​​phenotypic analysis module was deployed on an industrial computer equipped with an NVIDIA RTX4090 GPU, automatically completing image analysis and drug sensitivity interpretation. The total time for single-field full-time data analysis was less than 2 minutes. The experimental data obtained showed that the average accuracy of Staphylococcus aureus cell segmentation was as high as 96%, indicating extremely low background interference and accurate AI segmentation; the MIC value (ciprofloxacin) measurement deviation was ±1 concentration gradient, and the results were accurate and consistent with the CLSI standard reference value.

[0037] This invention addresses key technical bottlenecks in existing microscopic drug susceptibility testing, such as severe background interference from impurities in blood samples, low detection rates of low-load pathogens, and long testing cycles (requiring 24 hours of pure culture). This system utilizes pre-filters with broad-spectrum magnetic beads to enrich and remove impurities, thoroughly eliminating interfering substances such as cell debris, proteins, and lipids from blood samples. This results in a clear imaging field with extremely low background, achieving a bacterial identification accuracy close to 100%, fundamentally solving the problems of "chaotic background and difficult-to-interpret results" in existing microscopic drug susceptibility testing. Through a 2-hour short-term culture and proliferation process, the bacterial concentration is increased to a detectable level while fully preserving the pathogen's biological phenotypic activity, significantly improving the detection rate of low-load samples. Dynamic microscopic imaging and phenotypic analysis directly observe the true growth / inhibition status of pathogens under antibiotic treatment. This retains the core advantage of minimum inhibitory concentration (MIC) in phenotypic drug susceptibility testing (results more closely match clinical efficacy, without the false-positive bias of genotypic drug susceptibility testing), while drastically reducing the drug susceptibility testing cycle from 24 hours to less than 6 hours. This provides a reliable basis for rapid and accurate drug selection in severe infections, potentially reducing mortality.

[0038] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.

Claims

1. A rapid phenotypic drug susceptibility detection system for pathogens, characterized in that, include: The system includes a pathogenic microorganism enrichment culture module for pathogen capture, isolation, washing, purification, and culture; an antimicrobial susceptibility testing module located on the other side; a dynamic microscopic imaging module located opposite the antimicrobial susceptibility testing module; and an AI phenotypic analysis module connected to it. The AI ​​phenotypic analysis module incorporates a cell image segmentation unit, a morphological feature extraction unit, a growth rate analysis unit, a MIC calculation unit, and an antimicrobial susceptibility result grading and interpretation unit. The built-in cell image segmentation unit performs foreground-background separation and impurity removal on time-series microscopic images, accurately segmenting single cells and microbial community regions. The morphological feature extraction unit extracts bacterial cell outlines, size, division state, and aggregation morphology; the growth rate analysis unit quantifies the bacterial cell number growth rate, community area expansion rate, and division activity based on time-series images; the MIC calculation unit automatically calculates the minimum inhibitory concentration by combining the growth inhibition rate under different drug concentrations; and the drug susceptibility result grading and interpretation unit automatically determines the three levels of susceptibility (S), intermediate (I), and resistance (R) according to the CLSI standard.

2. The rapid phenotypic drug susceptibility detection system for pathogens according to claim 1, characterized in that, The pathogenic microorganism enrichment and culture module includes: a linear motion module for horizontal movement and several rows of pathogen capture chambers, washing chambers, purification chambers, culture chambers arranged thereon, a temperature control device located outside the culture chamber, and a magnetic separation device arranged vertically above the culture chamber. The pathogen capture cavity is pre-embedded with broad-spectrum antibody magnetic beads.

3. The rapid phenotypic drug susceptibility detection system for pathogens according to claim 1, characterized in that, The drug sensitivity reaction module includes: a constant temperature incubation unit and a drug sensitivity reaction plate disposed thereon; the dynamic microscopic imaging module is disposed above the drug sensitivity reaction module and includes: an autofocus module and a CMOS image acquisition unit disposed thereon and an objective lens located at the end of the CMOS image acquisition unit.

4. An application based on the drug susceptibility testing system according to any one of claims 1-3, characterized in that, This rapid phenotypic drug susceptibility detection for pathogens, using broad-spectrum magnetic bead enrichment and automated microscopic imaging, involves proliferating samples through a pathogen enrichment culture module and incubating them in a drug susceptibility reaction module. Dynamic microscopic imaging is then performed using a dynamic microscopic imaging module, followed by processing of the continuous image data by an AI phenotypic analysis module. This process extracts bacterial cell division rate, community area, and bacterial morphology parameters, calculates the minimum inhibitory concentration (MIC), and determines the pathogen drug susceptibility results (S / I / R) according to the CLSI standard, thus completing the phenotypic drug susceptibility detection.

5. The application according to claim 4, characterized in that, The AI ​​phenotypic analysis and automatic drug sensitivity assessment method, combining dynamic time-series image features and microbial growth patterns, achieves automated image recognition, phenotypic activity quantification, and intelligent drug sensitivity assessment, specifically including: Step 1: Image preprocessing and precise bacterial segmentation are performed using the bacterial cell image segmentation unit; Step 2: Extract multidimensional bacterial cell morphological features using the morphological feature extraction unit: Based on the segmented image, extract geometric features: major axis, minor axis, perimeter, area, circularity, and aspect ratio of a single bacterial cell; identify bacterial cell division morphology, budding morphology, and chain / cluster aggregation morphology; mark the proportion of cells in the division phase and the degree of bacterial community aggregation; count the total number of effective viable cells, the number of bacterial communities, and the total area of ​​the bacterial community in a single frame image to establish a single frame feature dataset; Step 3: Quantitatively calculate dynamic growth parameters using the dynamic growth rate analysis unit: Using the acquisition time as the axis, perform time-series matching on the number of bacteria and the area of ​​the bacterial community in multiple consecutive frames of images to construct a time-growth characteristic curve; calculate the growth rate of the number of bacteria and the expansion rate of the bacterial community area, and calculate the comprehensive growth activity value G. The value of G ranges from 0 to 1. The closer G is to 1, the stronger the growth activity of the pathogen. The closer G is to 0, the more significantly growth is inhibited. Step 4: Automatically calculate the minimum inhibitory concentration (MIC) using the MIC calculation unit: Extract the growth activity values ​​G corresponding to different gradient antimicrobial drug concentrations on the drug susceptibility test plate and compare them with the preset growth inhibition threshold G0. In this system, G0 is set to 0.

1. Growth activity below the threshold is considered growth inhibition. The G values ​​at each drug concentration are compared sequentially, and the lowest drug concentration that can completely inhibit the normal growth of the pathogen is determined as the minimum inhibitory concentration (MIC) of the antimicrobial drug corresponding to that pathogen. If the pathogen still maintains high growth activity at the highest drug concentration, it is marked as the MIC exceeding the detection range. Step 5: Automatic grading and interpretation of drug susceptibility results via the drug susceptibility result grading and interpretation unit: Based on the clinical CLSI drug susceptibility judgment standard, the calculated MIC value is compared with the corresponding clinical critical concentration threshold of the drug to complete the three-level result judgment: Sensitive (S): MIC ≤ clinical sensitive critical concentration, the antimicrobial drug is determined to be effective against the target pathogen; Intermediate (I): clinical sensitive critical concentration < MIC < clinical resistance critical concentration, the drug efficacy is determined to be limited and requires comprehensive clinical medication; Resistant (R): MIC ≥ clinical resistance critical concentration, the target pathogen has developed resistance to the antimicrobial drug and the drug is ineffective; The system automatically generates a test report and synchronously outputs the original image, growth curve, MIC value, and S / I / R grading results.

6. The application according to claim 5, characterized in that, Step 1 specifically includes: 1.1 Receive continuous time-series images uploaded by the dynamic microscopic imaging module, perform grayscale conversion, noise reduction, and illumination equalization preprocessing to eliminate imaging spots, background impurities, and interference from residual blood cells / proteins; 1.2 An improved semantic segmentation network is used to perform pixel-level segmentation of the image, distinguishing the background region, magnetic bead region, dead bacterial body region, live bacterial body and microbial community region; residual magnetic beads and cell debris regions are removed, and only the target live pathogen region is retained; 1.3 Perform morphological opening and closing operations on the segmentation results, repair the breakpoints of the bacterial cell outline, and output the binarized segmentation images of single bacterial cells and bacterial communities as the base images for subsequent feature extraction.

Images