Fecal parasite detection and AI training methods and infection assessment parameters
By controlling the infection process of experimental animals and AI training, the sensitivity and efficiency issues of fecal parasite detection were solved, accurate identification of multiple parasites and provision of multi-dimensional infection information were achieved, and the accuracy and speed of detection were improved.
Patent Information
- Application Number
- CN202410490567.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-23
- Publication Date
- 2025-10-24
AI Technical Summary
Existing fecal parasite detection technologies have the following problems: low sensitivity, high missed diagnosis rate, strong dependence on operation, low efficiency of manual interpretation, difficulty in large-scale application of AI training, and inability to accurately identify parasite types and infection stages.
By controlling the parasitic infection process of experimental animals, collecting fecal samples of different types and infection stages, diluting and pre-treating them, taking pictures for AI training, establishing a fecal parasite feature dataset, and using AI algorithms to identify and output the proportion of parasites during the infection period.
It improves the efficiency and accuracy of fecal parasite detection, reduces the risk of misdiagnosis and missed diagnosis, realizes the classification and identification of multiple parasites and the provision of multi-dimensional infection information, and improves the detection speed and accuracy.
Smart Images
Figure CN120833602A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the technical field of microscopic magnified image-based analysis of formed elements, and particularly relates to a fecal parasite infection evaluation parameter, a fecal parasite detection method, and an AI training method and computing processing and storage device thereof. BACKGROUND
[0002] In the prior art, microscopic observation and detection of fecal parasites (including eggs, oocysts, cysts, and worm bodies) is a method for diagnosing intestinal parasite infections. The brief steps are as follows: ① Collect fresh feces ② Prepare smears ③ Stain and examine under a microscope to identify and count different types of eggs. ④ Combine clinical manifestations to judge the results. Microscopic examination can detect parasitic species. Different parasites infect animals, and corresponding eggs, oocysts, cysts, and worm bodies appear in feces. Conventional microscopic examination of feces uses manual identification of the number of target objects under high-power microscopy, which is difficult to detect and has no quantitative standard. Many substances in fecal parasites cannot be identified and counted, and the technology is outdated. Direct smear examination under a microscope, although this method is low in cost and easy to implement, has problems such as low sensitivity, high rate of missed diagnosis, and strong dependence on operation.
[0003] Manual interpretation has the following limitations: high complexity of microscopic identification: traditional fecal parasite detection relies on manual observation and identification under a microscope, and the size, shape, color, and texture of eggs, oocysts, cysts, and worm bodies are different, even the same type of parasite has great differences in different developmental stages, which increases the difficulty and complexity of identification. Time-consuming and labor-intensive: the detection of each sample often involves the observation of multiple fields and multiple sections, plus the need for careful comparison, analysis, and recording, the interpretation time of each sample is longer. When facing a large number of samples to be tested, the human consumption is huge, and the work efficiency is low. Subjective factors: due to visual fatigue, experience differences, and other factors, different testers may have some subjectivity in identifying eggs, oocysts, cysts, and worm bodies, which may lead to misdiagnosis or missed diagnosis, which also reflects the problem of low efficiency of manual interpretation.
[0004] The applicant has proposed a series of Chinese patents, such as
[0005] 1. CN2020112669290, "Cell analysis method and system and quantitative method and system";
[0006] 2. CN2020112669182, "Cell suspension sample imaging method and system and kit";
[0007] 3. CN2022104799126, "Microscopic image acquisition device rapid focusing method and microscopic image acquisition method";
[0008] 4、CN2023110496905, "A urine feces (multi-parameter) formed element detection device, chip and method"
[0009] A new technical solution is used to measure the content of target substances in feces.
[0010] Fecal parasite detection can be combined with an understanding of the host and the characteristic morphological structures observed in feces, as well as the forms of many parasites, to help diagnose specific parasite species and accurately administer medication. It can reveal the parasite infection status of multiple systems in the body, and early detection and treatment can help with health management and monitoring and prevention of zoonotic diseases.
[0011] There are many types of parasites in feces, and different parasite infections require different clinical intervention methods. Accurate identification of different parasite species is necessary for accurate intervention measures. In addition, the state of parasites may vary greatly at different stages of infection, and precise intervention measures need to be taken according to the different stages of parasite infection.
[0012] However, in actual large-scale applications, collecting different types of parasite infection samples is time-consuming and labor-intensive, and the efficiency is low; it is very difficult to collect parasite samples at different infection stages, making large-scale AI training difficult. It is impossible to accurately identify the species of parasites or accurately identify the infection stage of parasites. SUMMARY
[0013] In this application, the inventor's technical solution collects different types of parasites and different infection stages of parasites during fecal collection and pretreatment, greatly improving the efficiency of the fecal parasite detection AI training method. By controlling the parasite infection process of experimental animals, accurate pictures of various parasites at each growth stage are obtained, making the AI model training accurate; different infection stages of fecal parasites are detected and identified. The proportion of the number of parasites during the infection period provides multi-dimensional fecal parasite infection evaluation parameters for fecal analysis.
[0014] A fecal parasite detection AI training method includes: collecting feces of parasite-infected animals; pretreating the feces to obtain a fecal sample; the fecal pretreatment includes diluting the feces with a diluent; allowing the fecal parasites to assume a natural state in the liquid; taking pictures of the fecal sample; using the obtained pictures for AI training to obtain a fecal parasite feature dataset A; the above-mentioned collection of feces of parasite-infected animals includes the steps of experimental animal selection, experimental animal infection, and fecal sample collection.
[0015] The fecal parasite detection AI training method includes any one of the following technical features: including any one of the following technical features: TC1: the above-mentioned parasites include helminths; the helminths include any one or more of flukes, tapeworms and nematodes; TC2: the above-mentioned parasites include protozoa; TC3: the above-mentioned fecal parasites include various growth stages of parasites, and the helminths include eggs and worms; TC4: the above-mentioned fecal parasites include various growth stages of parasites, and the protozoa include oocysts, sporozoites, cysts and trophozoites; TC5: the above-mentioned experimental animal selection includes a step of deworming the experimental animals, and a step of immunosuppression and infection prevention of the experimental animals; TC6: the above-mentioned experimental animal selection includes a step of deworming the experimental animals, and a step of immunosuppression and infection prevention of the experimental animals, and the above-mentioned step of immunosuppression and infection prevention of the experimental animals includes a step of using dexamethasone to suppress immunity or using antibiotics to prevent infection in the experimental animals; TC7: the above-mentioned animals include rats, mice, dogs or hamsters; TC8: the experimental animal infection is achieved by culturing an intermediate host, and the intermediate host is used to infect the experimental animals with any one of the selected growth stages of eggs, oocysts, cysts or worms; TC9: the experimental animal infection is achieved by injecting the selected growth stages of target parasites into the experimental animals through intraperitoneal injection, subcutaneous inoculation, gavage or intragastric injection
[0016] The experimental animal infection includes any one of the following technical features: TA10: a model of experimental animals infected with H. nana, 10 infectious cercariae per animal; TA20: a model of experimental animals infected with S. stercoralis, 500-1000 infectious filariform larvae per animal; TA30: a model of experimental animals infected with G. lamblia, 1×10^4 infectious cysts per animal.
[0017] The fecal parasite feature data set A includes any one or more of the following parasites: TB10: G. lamblia; TB20: H. nana; TB30: Clonorchis sinensis; TB40: S. stercoralis; TB50: Cryptosporidium.
[0018] The above-mentioned fecal parasite detection AI training method includes: collecting feces at different infection stages after the experimental animals are infected, and the above-mentioned fecal parasite feature data set A includes a parasite infection stage feature data set A; the infection stage is in units of days or in units of hours.
[0019] The parasite infection stage feature data set A includes any one or more of the following features: TB10: N-day G. lamblia; TB20: N-day H. nana; TB30: N-day Clonorchis sinensis; TB40: N-day S. stercoralis; TB50: N-day Cryptosporidium.
[0020] The fecal parasite detection AI training method includes any one of the following technical features: TD10: The fecal pretreatment includes fecal sample dyeing treatment, and the fecal sample dyeing treatment adds a dyeing agent to the feces; TD20: The fecal pretreatment includes fecal sample dyeing treatment, and the fecal sample dyeing treatment adds a dyeing agent to the feces; the dyeing agent is dissolved in the diluent; TD30: The dilution treatment is performed with reference to a turbidity card, and the feces is diluted to a set concentration range; TD40: The dilution treatment is performed with reference to a turbidity card, and the feces is diluted to a set concentration range; a dyeing diluent of a set proportion is added to the fecal sample obtained by turbidity dilution, and the dyeing diluent includes a dyeing agent and / or an insecticide; TD50: The obtained fecal sample is subjected to enrichment, and the enrichment is sedimentation enrichment and / or floating enrichment; after the enrichment, a suspension liquid of a layer in which the fecal parasites are located is taken as the fecal sample; TD60: The AI training includes image preprocessing and AI training processing, the image preprocessing outputs the identified fecal parasites, and the AI training processing outputs a fecal parasite feature data set A according to the identified fecal parasites; TD70: The fecal pretreatment includes live worm shaping treatment.
[0021] The live worm shaping treatment in step TD70 includes any one of the following technical features: TE10: The live worm shaping is performed by refrigeration, the temperature after the refrigeration is restored to normal temperature, the refrigeration temperature is higher than 2 degrees and lower than 8 degrees; TE20: The live worm shaping is performed by heating, the temperature after the heating is restored to normal temperature, the heating temperature is higher than 39 degrees and lower than 60 degrees; TE30: The live worm shaping treatment is performed on live worms, worm eggs, egg capsules or cysts by using a medicament; TE40: The live worm shaping treatment is performed on live worms, worm eggs, egg capsules or cysts by using a medicament, and the worm shaping medicament includes any one or more of formic acid, acetic acid, propionic acid, malonic acid, butyric acid, methanol, ethanol, formaldehyde, acetaldehyde, propyl aldehyde, butyl aldehyde, n-pentanal and glutaraldehyde.
[0022] The fecal parasite feature data set A is used as a feature data set, one or more pictures are identified by using AI software, a marked picture is obtained, the marked picture includes marked fecal parasites, manual review is performed on the marked picture to form a manually reviewed marked picture, AI training is performed by using the manually reviewed marked picture, and a fecal parasite feature data set B is obtained.
[0023] A fecal parasite detection method, which includes the following steps: obtaining a fecal sample by pretreating feces; the fecal pretreatment includes dilution treatment of the feces by using a diluent; the fecal parasites are suspended or precipitated in the liquid to present a natural state; a picture is obtained by taking a picture of the fecal sample; an AI recognition algorithm is used to identify the parasites in the image; and the AI recognition algorithm identifies the parasites according to a parasite infection stage feature data set.
[0024] The parasite infection stage feature data set is obtained by training after artificially labeling the parasite images; or the parasite infection stage feature data set is obtained by training after obtaining the fecal images of the experimental animals infected for N days.
[0025] The parasite infection stage feature data set includes any one or more of protozoa, nematodes, tapeworms, and flukes.
[0026] The fecal parasite detection method includes any one of the following technical features: TH10: outputting the number of parasites in different infection stages; TH20: outputting the total number of selected parasites; TH30: outputting the proportion of the number of parasites in the infection period.
[0027] A fecal parasite infection evaluation parameter, the proportion of the number of parasites in the infection period = the number of parasites in the infection period / the total number of parasites.
[0028] The unit of the infection period is days N, and 1≤N≤30.
[0029] The fecal parasite infection evaluation parameter is an array, which includes infection period parasite number proportion value 1 and infection period parasite number proportion value 2; the infection period parasite number proportion value 1 corresponds to the infection period 1, and the infection period parasite number proportion value 2 corresponds to the infection period 2.
[0030] A computing processing device includes any one of the following technical features: TEA1: for running all or part of the fecal parasite detection AI training method; TEA2: the memory of the computing processing device includes the fecal parasite feature data set A; TEA3: the memory of the computing processing device includes the fecal parasite feature data set B; TEA4: for running all or part of the fecal parasite detection method; TEA5: the memory of the computing processing device includes the fecal parasite infection evaluation parameter.
[0031] A data storage device includes any one of the following technical features: TEB1: storing program code for executing all or part of the fecal parasite detection AI training method; TEB2: storing the fecal parasite feature data set A; TEB3: storing the fecal parasite feature data set B; TEB4: storing program code for executing all or part of the fecal parasite detection method; TEB5: storing the fecal parasite infection evaluation parameter.
[0032] A detection device includes any one of the following technical features: TEC1: for running all or part of the fecal parasite detection AI training method; TEC2: for running all or part of the fecal parasite detection method.
[0033] The technical effects of the above technical solutions include: allowing fecal parasites to be in a liquid and in a natural state; allowing their morphological characteristics to fully expand to form effective samples, so that the training has sufficient recognition rate; without natural expansion of samples, manual annotation cannot be standardized, and AI training work cannot be carried out. Taking pictures of fecal samples corresponds to the same detection application scenario, and the recognition rate of AI training is high.
[0034] The technical effects of the above technical solutions include: multiple parasites can be classified and identified, and multiple types of parasites can be identified at the same time.
[0035] The technical effects of the above technical solutions include: controllable cultivation of various growth stages of parasites, phased identification, and obtaining more deep-dimensional parasite information.
[0036] The technical effects of the above technical solutions include: the steps of deworming and immunosuppression of experimental animals can accurately control the desired parasite infection samples, and improve the pertinence and effectiveness of training.
[0037] The technical effects of the above technical solutions include: multiple categories of experimental animals are convenient to obtain.
[0038] The technical effects of the above technical solutions include: multiple infection methods of experimental animals are convenient for obtaining parasite samples.
[0039] The technical effects of the above technical solutions include: the amount of infection of experimental animals can be accurately controlled, which further facilitates the acquisition of target types of parasites.
[0040] The technical effects of the above technical solutions include: the fecal parasite feature data set A includes multiple parasites, including how many types of parasites, that is, the number of types that can be identified by AI application. AI training and application greatly improve the efficiency of parasite identification. Through the application of AI artificial intelligence in the field of fecal component analysis, AI artificial intelligence can be used to replace manual interpretation, greatly improving efficiency.
[0041] The technical effects of the above technical solutions include: using AI recognition to improve accuracy and consistency: excellent AI models can achieve high recognition accuracy and stability after sufficient training, reducing the risk of misdiagnosis and missed diagnosis. AI will not miss tiny eggs or oocysts or cysts or worm bodies, or mistakenly identify other substances as eggs or oocysts or cysts or worm bodies due to human visual errors, experience differences or distraction. Improved efficiency: AI algorithms can process a large number of smear images in a short period of time, and are not affected by factors such as fatigue and emotions, which improves the speed and efficiency of detection and can effectively solve the detection bottleneck caused by time-consuming manual interpretation. Precision and sensitivity: AI algorithms can recognize tiny details and subtle color differences, and may be more sensitive to certain types of eggs or oocysts or cysts or worm bodies that have large morphological variations and are difficult to detect.
[0042] The technical benefits of the above-mentioned technical solution include: fecal parasite feature dataset A includes parasite infection stage feature dataset A, with infection stages measured in days or hours. Differentiating infection stages allows for deeper parasite analysis, providing clinicians with information on the time dimension of infection. The integration of multi-dimensional information on infection time, infection type, and infection quantity can provide clinicians with more accurate reference information.
[0043] The technical effects of the above technical solution include: dilution and simultaneous dyeing, pre-configuring the dye and diluent into a dye diluent of a certain concentration, which can effectively control the concentration and uniformity of the dye; the parasites after dyeing are easier to identify.
[0044] The technical effects of the above technical solution include: diluting the stool to within the set concentration range by referring to the turbidimetric card, so that the impurity distribution of different pictures trained by AI is the same or similar.
[0045] The technical benefits of the above-mentioned solution include: enriching the obtained fecal sample, either by sedimentation or flotation; after enrichment, the suspension of the fecal parasite layer is taken as the fecal sample, thus reducing the difficulty of finding the target. Enrichment can also improve detection sensitivity and specificity: the density of parasite eggs, oocysts, cysts, or parasite bodies in fecal samples is generally low, and direct testing may lead to missed diagnoses.
[0046] The technical effects of the above technical solution include: through effective insect egg or oocyst or cyst or insect body enrichment technology, the concentration of target biomarkers (such as insect eggs) in the sample can be significantly increased, thereby improving the sensitivity and positive detection rate of detection, ensuring the accuracy of clinical diagnosis, and reducing false negative results.
[0047] The technical effects of the above technical solution include: pre-processing the feces including the shaping of live worms to prevent the live worms from interfering with the quality of the taken pictures.
[0048] The technical effects of the above technical solutions include: the live insects are shaped by refrigeration or heating, which is low in cost and convenient to process.
[0049] The technical effects of the above technical solutions include: the live insects are shaped by refrigeration or heating, which does not affect the characteristics of the proteins in a specific temperature range and changes the external morphology.
[0050] The technical effects of the above technical solutions include: the live insects or insect eggs or egg capsules or cysts or insect bodies are shaped by medicaments, which is simple and effective to operate.
[0051] The technical effects of the above technical solutions include: based on the fecal parasite feature data set A, the fecal parasite feature data set B is packaged into an independent data, so that the AI recognition ability can be iterated and improved. The rolling training of the AI recognition model will make the data set more and more rich, and the recognition more and more accurate. After the model update after iterative training, a closed loop feedback is formed, so that the instrument can continuously improve the recognition ability in daily use.
[0052] The technical effects of the above technical solutions include: according to the parasite infection stage feature data set, the AI recognition algorithm is used to identify that the parasites in the image have the infection time dimension feature.
[0053] The technical effects of the above technical solutions include: outputting the number of parasites in different infection stages; outputting the total number of selected parasites; outputting the proportion of the number of parasites in the infection period, which can provide multi-dimensional parasite infection information.
[0054] The technical effects of the above technical solutions include: the proportion of the number of parasites in the infection period = the number of parasites in the infection period / the total number of parasites, which provides another dimension of parasite infection information, which can be used to evaluate the degree of fecal parasite infection.
[0055] The technical effects of the above technical solutions include: based on the array of fecal parasite infection evaluation parameters, the parasite infection information is statistically analyzed, and more dimensional analysis data can be output for clinical reference. BRIEF DESCRIPTION OF DRAWINGS
[0056] Figure 1 It is an AI training method for detecting fecal parasites Figure 1 ;
[0057] Figure 2 It is a schematic diagram of collecting feces of animals infected with parasites Figure 1 ;
[0058] Figure 3 It is a schematic diagram of collecting feces of animals infected with parasites Figure 2 ;
[0059] Figure 4is a schematic of collecting feces from a parasite-infected animal Figure 3 ;
[0060] Figure 5 is a schematic of feces pre-treatment Figure 1 ;
[0061] Figure 6 is a schematic of feces pre-treatment Figure 2 ;
[0062] Figure 7 is a schematic of feces pre-treatment Figure 3 ;
[0063] Figure 8 is a schematic of feces pre-treatment Figure 4 ;
[0064] Figure 9 is a schematic of feces pre-treatment Figure 5 ;
[0065] Figure 10 is a schematic of feces pre-treatment Figure 6 ;
[0066] Figure 11 is a schematic of collecting feces from a parasite-infected animal Figure 4 ;
[0067] Figure 12 is a schematic of a fecal parasite detection AI training method Figure 2 ;
[0068] Figure 13 is a schematic of collecting feces from a parasite-infected animal Figure 5 ;
[0069] Figure 14 is a schematic of collecting feces from a parasite-infected animal Figure 6 ;
[0070] Figure 15 is a schematic of collecting feces from a parasite-infected animal Figure 7 ;
[0071] Figure 16 is a schematic of collecting feces from a parasite-infected animal Figure 8 ;
[0072] Figure 17 is a schematic of a fecal parasite detection AI training method Figure 3 ;
[0073] Figure 18 is a schematic of a fecal parasite detection AI training method Figure 4 ;
[0074] Figure 19A fecal parasite detection AI training method Figure 5 ;
[0075] Figure 20 A fecal parasite detection AI training method Figure 6 ;
[0076] Figure 21 A fecal parasite detection method Figure 1 ;
[0077] Figure 22 A fecal parasite picture display
[0078] Figure 23 A fecal parasite picture manual recognition and classification display
[0079] Figure 24 A recognized fecal parasite DETAILED DESCRIPTION
[0080] The application will be described in further detail below with reference to the drawings. It should be noted that the following description of the preferred embodiments of the application is merely illustrative and does not in any way limit the application. The description of the preferred embodiments of the application is merely illustrative of the general principles of the application. The numbers "first", "second", and "A", "B" in the application are merely for the convenience of description and do not represent the order of time or space. The combination of letters and numbers "TA", "TB", "H" in the application is merely for the convenience of description, and the specific meaning is determined by the specific words referred to.
[0081] As Figure 1 A fecal parasite detection AI training method, collecting feces of parasite-infected animals; pretreating the feces to obtain a fecal sample; the feces pretreatment includes diluting the feces with a diluent; allowing the fecal parasites to assume a natural state in the liquid, which can be a suspended state or a precipitated state; taking pictures of the fecal sample; using the obtained pictures for AI training to obtain a fecal parasite feature dataset A; the above-mentioned collection of feces of parasite-infected animals includes the steps of experimental animal selection, experimental animal infection, and fecal sample collection.
[0082] The above-mentioned parasites include helminths; the helminths include any one or more of trematodes, cestodes, and nematodes; the nematodes include any one or more of pinworms, hookworms, roundworms, nematodes, and whipworms; the above-mentioned parasites include protozoa; the above-mentioned fecal parasites include various growth stages of parasites, and the helminths include eggs and worms; the above-mentioned fecal parasites include various growth stages of parasites, and the protozoa include oocysts, sporozoites, cysts, and trophozoites.
[0083] As Figure 2, collecting feces from parasite-infected animals, including the selection of experimental animals, infection of experimental animals, and collection of fecal samples. The experimental animals include rats, mice, dogs, or hamsters; experimental animal infection is carried out by injecting the target parasite life cycle stage into the experimental animals through intraperitoneal injection, subcutaneous inoculation, gavage, or intragastric injection. The process of collecting fecal eggs, oocysts, cysts, or parasite bodies by establishing an animal model of parasite infection includes the following steps:
[0084] Preparation and pretreatment of experimental animals: According to the experimental requirements, select appropriate experimental animal species, such as rats, mice, dogs, hamsters, etc., ensure that the animals are in good health, and perform necessary pretreatment, including deworming, adaptive feeding, immunosuppression (such as the use of dexamethasone) or antibiotics (such as penicillin) to prevent complications.
[0085] One method for obtaining parasite life cycle stages: For parasites that require an intermediate host, first cultivate the intermediate host (such as the intermediate host of Hymenolepithecus microtiformis, Tribolium castaneum) and infect it with the corresponding eggs, oocysts, cysts, or parasite bodies, and then collect the required life stages (such as cysticerci, metacercariae, protoscolex, etc.).
[0086] The second method for obtaining parasite life cycle stages: For parasites that do not require an intermediate host, obtain eggs, oocysts, cysts, or parasite bodies directly from patients or infected animals and purify and count them by appropriate methods (such as centrifugation, filtration, and dilution).
[0087] Experimental Animal Infection: Inject the target parasite life cycle stage into experimental animals using an appropriate route of infection (e.g., intraperitoneal injection, subcutaneous inoculation, oral gavage, intragastric infusion, etc.). Control the infective dose. For example, in the Hymenolepithecus microtiformis infection model, inject each animal with 10 infective cysticerci, the infective larval stage of Hymenolepithecus microtiformis; in the TA20 (Strylephoides stercoralis) infection model, inject each animal with 500-1000 infective filariform larvae, the infective larval stage of Strongyloides stercoralis; and in the TA30 (Giardia lamblia) infection model, inject each animal with 1×10^4 infective cysts to ensure experimental efficacy and minimize animal mortality.
[0088] Monitoring and Sample Collection: Regular fecal testing for worm eggs, gravid segments, or other morphological markers begins at designated time points post-infection. The morphology of these markers in fecal samples is obtained to confirm successful infection. Symptoms, survival status, and fecal egg quantity and quality changes of infected animals are recorded. For example, in a 6-day Giardia lamblia animal model, G. lamblia is in the form of trophozoites and cysts; in an 18-day Hymenolepis nana animal model, H. nana is in the form of eggs; in a 1-month Clonorchis sinensis animal model, C. sinensis is in the form of eggs; in a 2-week Strongyloides stercoralis animal model, S. stercoralis is in the form of larvae; in a 4-day Cryptosporidium animal model, Cryptosporidium is in the form of oocysts.
[0089] Model Optimization and Long-term Maintenance: Adjust model parameters (such as infection dose, infection route, use of immunosuppressants, etc.) according to the characteristics of different parasites to optimize model stability and durability, so as to obtain fecal worm eggs or oocysts or cysts or worm samples for a long time. Maintain a good feeding environment, avoid repeated infection, and record the overall health status of animals and its impact on subsequent experiments.
[0090] Model Application: Parasite Feature Morphology Collection and AI Training: After successfully constructing various parasite animal models and confirming that the animals have been effectively infected, fecal samples from infected animals are systematically collected according to a pre-set schedule and frequency. Professional treatment of fecal samples is performed, and image data of worm eggs or oocysts or cysts or worms are extracted and classified through microscopic detection and imaging technology. The obtained high-quality worm egg image dataset is used to train artificial intelligence algorithms, and through deep learning, the ability and accuracy of the AI system to identify worm eggs or oocysts or cysts or worms of different parasites are improved. The worm egg image database is continuously expanded and improved to cover more types of parasites and worm egg morphologies at different infection stages to meet the needs of the AI system for diversified samples and improve its detection efficiency in actual application scenarios. Through this general fecal worm egg or oocyst or cyst or worm collection process, abundant sample resources can be continuously and efficiently provided to solve the problem of insufficient number of AI fecal worm egg or oocyst or cyst or worm recognition training samples.
[0091] The above process is individually adjusted according to specific experimental goals and the types of parasites used, and relevant experimental animal ethics regulations are followed. At the same time, for parts involving human samples, relevant laws, regulations, and ethical review requirements are strictly followed.
[0092] For example, Figure 3 , the experimental animal selection includes a step of deworming the experimental animals, and a step of immunosuppressing and preventing infection of the experimental animals, the step of immunosuppressing and preventing infection of the experimental animals including a step of using, for example, dexamethasone to suppress immunity or antibiotics to prevent infection.
[0093] The experimental animal infection is achieved by cultivating intermediate hosts, and the intermediate hosts are infected with corresponding worm eggs or oocysts or cysts or worm bodies. For example, the intermediate host of Hymenolepis nana is Tribolium castaneum, and the intermediate host is infected with corresponding worm eggs or oocysts or cysts or worm bodies, and then the desired life stages such as cercaria-like tail cercaria, cysticercus, and protoscolex are collected.
[0094] The experimental animal infection can be an animal model infected with Hymenolepis nana, 10 cercaria-like tail cercariae are injected into each animal; an animal model infected with Strongyloides stercoralis, 500-1000 filariform larvae are injected into each animal; and an animal model infected with Giardia lamblia, 1x10^4 cysts are injected into each animal. The amount of parasites is quantitatively input, and the degree of infection is controlled.
[0095] The fecal parasite feature data set A includes any one or more of the following parasites: TB10: Giardia lamblia; TB20: Hymenolepis nana; TB30: Clonorchis sinensis; TB40: Strongyloides stercoralis; and TB50: Cryptosporidium.
[0096] As shown in Figure 4 , after the experimental animal infection, the worm reproduction and growth are waited for, and then the degree of infection is detected, and the feces at different infection stages are collected. The above fecal parasite feature data set A includes a parasite infection stage feature data set A; the infection stage is in units of days or in units of hours.
[0097] The parasite infection stage feature data set A includes any one or more of the following features: TB10: N-day Giardia lamblia; TB20: N-day Hymenolepis nana; TB30: N-day Clonorchis sinensis; TB40: N-day Strongyloides stercoralis; and TB50: N-day Cryptosporidium.
[0098] As shown in Figure 11 , the feces of the parasite-infected animal are collected Figure 4 , the experimental animal is infected with Giardia lamblia, and the worm reproduction and growth are waited for for 6 days, and then the fecal sample collection is performed. Then, the fecal parasite detection AI training method shown in Figure 12 is used for AI training to obtain a Giardia AI feature data set. Similarly, for other parasites, the fecal parasite detection AI training method shown in Figure 17 is used for AI training to obtain an Hymenolepis nana AI feature data set, a Clonorchis sinensis AI feature data set, a Strongyloides stercoralis AI feature data set, and a Cryptosporidium AI feature data set. The data set and the parasite in the collected sample correspond.
[0099] As shown in Figure 13 , the feces of the parasite-infected animal are collected Figure 5 , the experimental animal is infected with Hymenolepis nana, and the worm reproduction and growth are waited for for 18 days, and then the fecal sample collection is performed.
[0100] As shown in Figure 14 , the feces of the parasite-infected animal are collected Figure 6 , the experimental animals are infected with Clonorchis sinensis, and the feces sample collection is performed after waiting for the worms to reproduce and grow for 30 days.
[0101] As shown in Figure 15 , the feces of the parasite-infected animal are collected Figure 7 , the experimental animals are infected with Syphacia parvum, and the feces sample collection is performed after waiting for the worms to reproduce and grow for 14 days.
[0102] As shown in Figure 16 , the feces of the parasite-infected animal are collected Figure 8 , the experimental animals are infected with Cryptosporidium, and the feces sample collection is performed after waiting for the worms to reproduce and grow for 4 days.
[0103] The AI training method for fecal parasite detection shown in Figure 18 is used for AI training to obtain AI feature data sets such as N-day Blue Giarid AI feature data set, N-day Microcystic Diphyllobothrium AI feature data set, N-day Clonorchis sinensis AI feature data set, N-day Syphacia parvum AI feature data set, and N-day Cryptosporidium AI feature data set. The data sets correspond to the parasites in the collected samples. The feature data set of each parasite at different stages can be obtained respectively, and the division of different stages can be set according to the characteristics of the specific parasite, such as in units of days, hours, or minutes.
[0104] As shown in Figure 5 , the feces pretreatment includes feces sample dyeing treatment, and a dyeing agent is added to the feces during the feces sample dyeing treatment; the dyeing agent is dissolved in the diluent. As shown in Figure 5 and Figure 6 , the dilution treatment is performed by referring to the turbidity card, and the feces is diluted to a set concentration range. As shown in Figure 6 , the dilution treatment is performed by referring to the turbidity card, and the feces is diluted to a set concentration range. The feces sample obtained by turbidity dilution is added with a set proportion of dyeing diluent, and the dyeing diluent includes a dyeing agent and / or a pesticide.
[0105] As shown in Figure 7 and Figure 8 , the feces pretreatment includes live worm shaping treatment. The live worm shaping treatment can be performed by refrigeration as shown in Figure 10 , the live worm is shaped by refrigeration, and the temperature after refrigeration is restored to normal temperature, the temperature of refrigeration is higher than 2 degrees and lower than 8 degrees. The live worm can also be shaped by heating, and the temperature after heating is restored to normal temperature, the temperature of heating is higher than 39 degrees and lower than 60 degrees.
[0106] As shown in Figure 6 andFigure 8 The live worm shaping treatment is shaping the live worm or worm egg or egg capsule or package or worm body by using a drug, and the worm shaping drug includes any one or more of formic acid, acetic acid, propionic acid, malonic acid, butyric acid, methanol, ethanol, formaldehyde, acetaldehyde, propionaldehyde, butyraldehyde, n-pentanal, glutaraldehyde.
[0107] As Figure 5 And Figure 6 And Figure 9 And Figure 10 The obtained fecal sample is enriched, and the enrichment is sedimentation enrichment and / or floating enrichment, and can also be stratification enrichment, centrifugal enrichment or other enrichment methods; after enrichment, the suspension liquid in the layer where the fecal parasite is located is taken as the fecal sample. As Figure 22 It is a picture display of fecal parasites; the left picture is the state of the parasite without enrichment, and the right is the state of the parasite after enrichment.
[0108] The AI training includes image preprocessing and AI training processing, and the image preprocessing outputs the identified fecal parasites, and the AI training processing outputs the fecal parasite feature data set A according to the identified fecal parasites; as Figure 23 It is a display of artificial identification and classification of fecal parasite pictures; different parasites are artificially labeled. Figure 24 It is according to the identified fecal parasites. Figure 24 In the upper left, upper right, lower left and lower middle pictures, the parasites are all heavy-winged trematode eggs; the lower right picture is a Taenia saginata egg.
[0109] Figure 19 The fecal parasite feature data set A is used as a feature data set, one or more of the above pictures are identified by using AI software, a marked picture is obtained, the marked picture includes the marked fecal parasites, the marked picture is artificially reviewed to form an artificial review marked picture, and the artificial review marked picture is used for AI training to obtain a fecal parasite feature data set B.
[0110] As Figure 20 A fecal parasite detection method, the fecal sample is obtained by pre-treating the feces; the fecal pre-treatment includes diluting the feces with a diluent; the fecal parasite is suspended or precipitated in the liquid to present a natural state; the fecal sample is photographed to obtain a picture; the AI recognition algorithm is used to identify the parasites in the image; the AI recognition algorithm identifies the parasites according to the parasite infection stage feature data set. The target detection object in the picture is the parasite. The parasite infection stage feature data set is obtained by training after artificially labeling the parasite image; or the parasite infection stage feature data set is obtained by training after obtaining the fecal image of the experimental animal infected for N days. The parasite infection stage feature data set includes any one or more of protozoa, nematodes, tapeworms and trematodes.
[0111] As Figure 20 A fecal parasite detection method can output the number of parasites at different infection stages. A fecal parasite detection method not shown in the figure can also output the total number of selected parasites; the proportion of the number of parasites at the infection period can also be output.
[0112] As Figure 21 The number of target detection objects corresponding to unit mass of dry matter, i.e. the number of fecal parasites corresponding to unit mass of dry matter, can also be obtained. The mass of dry matter corresponding to unit volume can be obtained by image shooting and empirical formula calculation; after obtaining the mass of dry matter corresponding to unit volume, the mass of dry matter can be calculated according to the volume of the detection sample. How to obtain the mass of dry matter corresponding to unit volume by image shooting and empirical formula calculation is clearly stated in another patent "CN2024100344351 Fecal suspension parameter detection and dry matter calculation formula acquisition method and device" submitted by the applicant.
[0113] A fecal parasite infection evaluation parameter, the proportion of the number of parasites at the infection period = the number of parasites at the infection period / the total number of parasites. The unit of the infection period is days N, 1≤N≤30. The fecal parasite infection evaluation parameter is an array, which includes the proportion of the number of parasites at the infection period value 1 and the proportion of the number of parasites at the infection period value 2; the proportion of the number of parasites at the infection period value 1 corresponds to the infection period 1, and the proportion of the number of parasites at the infection period value 2 corresponds to the infection period 2. The proportion of the number of parasites at the infection period = the number of parasites at the infection period / the total number of parasites can be calculated by a single species of parasite; it can also be calculated by combining several species of parasites. The specific species corresponding to the number of parasites at the infection period and the total number of parasites can be selected and set as needed.
[0114] A computing processing device for running all or part of the above-mentioned fecal parasite detection AI training method; the memory of the above-mentioned computing processing device includes the above-mentioned fecal parasite feature data set A; the memory of the above-mentioned computing processing device includes the above-mentioned fecal parasite feature data set B; for running all or part of the above-mentioned fecal parasite detection method; the memory of the above-mentioned computing processing device includes the above-mentioned fecal parasite infection evaluation parameter.
[0115] A data storage device for storing program codes for executing all or part of the above-mentioned fecal parasite detection AI training method; storing the above-mentioned fecal parasite feature data set A; storing the above-mentioned fecal parasite feature data set B; storing the program codes for executing all or part of the above-mentioned fecal parasite detection method; storing the above-mentioned fecal parasite infection evaluation parameter.
[0116] A detection device comprising any one of the following technical features: all or part of the above fecal parasite detection AI training method; all or part of the above fecal parasite detection method.
[0117] The application is illustrated and described according to preferred embodiments and several alternatives, but the application is not limited by the specific description in the specification. Other additional alternative or equivalent components can also be used to practice the application.
Claims
1. A fecal parasite detection AI training method, characterized by, It comprises: Collecting the feces of animals infected with parasites; Obtaining a fecal sample by pretreating the feces; the fecal pretreatment comprises diluting the feces with a diluent; Letting the fecal parasites assume a natural state in the liquid; Taking pictures of the fecal sample to obtain pictures; Using the obtained pictures for AI training to obtain a fecal parasite feature dataset A; The collection of feces of animals infected with parasites includes the steps of experimental animal selection, experimental animal infection and fecal sample collection.
2. The fecal parasite detection AI training method of claim 1, wherein It comprises any one of the following technical features: TC1: The parasites include helminths; the helminths include any one or more of trematodes, cestodes and nematodes; TC2: The parasites include protozoa; TC3: The fecal parasites include various growth stages of the parasites, and the helminths include eggs and worms; TC4: The fecal parasites include various growth stages of the parasites, and the protozoa include oocysts, sporozoites, cysts and trophozoites; TC5: The experimental animal selection includes the steps of deworming the experimental animals, and the steps of immunosuppression and infection prevention of the experimental animals; TC6: The experimental animal selection includes the steps of deworming the experimental animals, and the steps of immunosuppression and infection prevention of the experimental animals, wherein the steps of immunosuppression and infection prevention of the experimental animals include the steps of using dexamethasone to suppress immunity or using antibiotics to prevent infection in the experimental animals; TC7: The animals include rats, mice, dogs or hamsters; TC8: The experimental animal infection is achieved by cultivating intermediate hosts, and the intermediate hosts are used to infect the experimental animals with any one of the corresponding eggs, oocysts, cysts or worms; TC9: The experimental animal infection is achieved by injecting the selected growth stage of the target parasites into the experimental animals through intraperitoneal injection, subcutaneous inoculation, gavage or intragastric injection.
3. The fecal parasite detection AI training method of claim 1, wherein The experimental animal infection comprises any one of the following technical features: TA10: The animal model infected with H. diminuta is injected with 10 infective cysticercoids per animal; TA20: The animal model infected with S. stercoralis is injected with 500-1000 infective filariforms per animal; TA30: The animal model infected with G. lamblia is injected with 1×10^4 infective cysts per animal.
4. The fecal parasite detection AI training method of claim 1, wherein The fecal parasite feature dataset A includes any one or more of the following parasites: TB10: G. lamblia; TB20: H. diminuta; TB30: Clonorchis sinensis; TB40: S. stercoralis; TB50: Cryptosporidium.
5. The fecal parasite detection AI training method of claim 1, It comprises: Collecting feces at different infection stages after the experimental animal infection, and the fecal parasite feature dataset A comprises a parasite infection stage feature dataset A; the infection stage is in units of days or in units of hours.
6. The fecal parasite detection AI training method of claim 5, wherein The parasite infection stage feature dataset A comprises any one or more of the following features: TB10: Giardia nivalis; TB20: Hymenolepis nana; TB30: Clonorchis sinensis; TB40: Strongyloides stercoralis; TB50: Cryptosporidium nivalis.
7. The fecal parasite detection AI training method according to claim 1, characterized in that, any one of the following technical features is included: TD10: the fecal sample pretreatment includes a fecal sample dyeing process, and the fecal sample dyeing process adds a dyeing agent to the fecal sample; TD20: the fecal sample pretreatment includes a fecal sample dyeing process, and the fecal sample dyeing process adds a dyeing agent to the fecal sample; the dyeing agent is dissolved in the diluent; TD30: the dilution treatment is performed with reference to a turbidity card, and the fecal sample is diluted to a set concentration range; TD40: the dilution treatment is performed with reference to a turbidity card, and the fecal sample is diluted to a set concentration range; a dyeing diluent is added to the fecal sample obtained by turbidity dilution, and the dyeing diluent includes a dyeing agent and / or an insecticide; TD50: the obtained fecal sample is subjected to enrichment, and the enrichment is sedimentation enrichment and / or floatation enrichment; after enrichment, the suspension liquid of the layer where the fecal parasites are located is taken as the fecal sample; TD60: the AI training includes image preprocessing and AI training processing, the image preprocessing outputs the identified fecal parasites, and the AI training processing outputs a fecal parasite feature dataset A according to the identified fecal parasites; TD70: the fecal sample pretreatment includes a live worm shaping process.
8. The fecal parasite detection AI training method according to claim 7, characterized in that, any one of the following technical features is included in the live worm shaping process in step TD70: TE10: the live worm is shaped by refrigeration, and the temperature is restored to normal temperature after refrigeration; the refrigeration temperature is higher than 2 degrees and lower than 8 degrees; TE20: the live worm is shaped by heating, and the temperature is restored to normal temperature after heating; the heating temperature is higher than 39 degrees and lower than 60 degrees; TE30: the live worm shaping process is performed by using a medicament to shape the live worm, worm egg, oocyst or cyst; TE40: the live worm shaping process is performed by using a medicament to shape the live worm, worm egg, oocyst or cyst; the worm shaping medicament includes any one or more of formic acid, acetic acid, propionic acid, malonic acid, butyric acid, methanol, ethanol, formaldehyde, acetaldehyde, propionaldehyde, butyraldehyde, n-pentanal and glutaraldehyde.
9. The fecal parasite detection AI training method according to any one of claims 1 to 8, characterized in that, the fecal parasite feature dataset A is used as a feature dataset, one or more pictures are identified by using AI software, a marked picture is obtained, the marked picture includes marked fecal parasites, the marked picture is manually reviewed to form a manually reviewed marked picture, the manually reviewed marked picture is used for AI training, and a fecal parasite feature dataset B is obtained.
10. A fecal parasite detection method, characterized in that, a fecal sample is obtained by pretreating the fecal sample; the fecal pretreatment includes diluting the fecal sample with a diluent; the fecal parasites are allowed to present in a natural state in the liquid; the fecal sample is photographed to obtain a picture; The AI recognition algorithm recognizes the parasites in the image according to the parasite infection stage feature data set.
11. The fecal parasite detection method according to claim 10, wherein, The parasite infection stage feature data set is obtained by training after manually annotating the parasite image; or the parasite infection stage feature data set is obtained by training after obtaining the fecal image of the experimental animal infected for N days.
12. The fecal parasite detection method according to claim 10, wherein, The parasite infection stage feature data set includes any one or more of protozoa, nematodes, tapeworms, and flukes.
13. The fecal parasite detection method according to claim 10, wherein, Any one of the following technical features is included: TH10: Output the number of parasites in different infection stages; TH20: Output the total number of selected parasites; TH30: Output the proportion of the number of parasites in the infection period.
14. A fecal parasite infection evaluation parameter, wherein, The proportion of the number of parasites in the infection period = the number of parasites in the infection period / the total number of parasites.
15. The fecal parasite infection evaluation parameter according to claim 14, wherein, The infection period is in days N, 1 ≤ N ≤ 30.
16. The fecal parasite infection evaluation parameter according to claim 14, wherein, The fecal parasite infection evaluation parameter is an array, which includes the infection period parasite number proportion value 1 and the infection period parasite number proportion value 2; the infection period parasite number proportion value 1 corresponds to the infection period 1, and the infection period parasite number proportion value 2 corresponds to the infection period 2.
17. A computing processing device, comprising any one of the following technical features: TEA1: used for running all or part of the fecal parasite detection AI training method according to any one of claims 1 to 9; TEA2: the memory of the computing processing device includes the fecal parasite feature data set A according to claims 1 to 9; TEA3: the memory of the computing processing device includes the fecal parasite feature data set B according to claim 9; TEA4: used for running all or part of the fecal parasite detection method according to any one of claims 10 to 13; TEA5: the memory of the computing processing device includes the fecal parasite infection evaluation parameter according to any one of claims 14 to 16.
18. A data storage device, comprising any one of the following technical features: TEB1: storing program code for executing all or part of the fecal parasite detection AI training method according to any one of claims 1 to 9; TEB2: storing the fecal parasite feature data set A according to claims 1 to 9; TEB3: storing the fecal parasite feature data set B according to claim 9; TEB4: storing program code for executing all or part of the fecal parasite detection method according to any one of claims 10 to 13; TEB5: storing the fecal parasite infection evaluation parameter according to any one of claims 14 to 16.
19. A detection device, characterized in that, Any one of the following technical features is included: TEC1: for running all or part of the fecal parasite detection AI training method according to any one of claims 1 to 9; TEC2: for running all or part of the fecal parasite detection method according to any one of claims 10 to 13.