OA21122A - Parasite detection method and system based on artificial intelligence, and terminal device. - Google Patents
Parasite detection method and system based on artificial intelligence, and terminal device. Download PDFInfo
- Publication number
- OA21122A OA21122A OA1202300018 OA21122A OA 21122 A OA21122 A OA 21122A OA 1202300018 OA1202300018 OA 1202300018 OA 21122 A OA21122 A OA 21122A
- Authority
- OA
- OAPI
- Prior art keywords
- image
- parasite
- détection
- detected
- training
- Prior art date
Links
- 244000045947 parasite Species 0.000 title claims abstract description 169
- 238000001514 detection method Methods 0.000 title abstract description 13
- 238000013473 artificial intelligence Methods 0.000 title description 7
- 238000003709 image segmentation Methods 0.000 claims abstract description 9
- 210000004027 cell Anatomy 0.000 claims description 77
- 238000012549 training Methods 0.000 claims description 63
- 238000000034 method Methods 0.000 claims description 58
- 201000004792 malaria Diseases 0.000 claims description 49
- 238000004590 computer program Methods 0.000 claims description 23
- 230000008569 process Effects 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 15
- 241000224489 Amoeba Species 0.000 claims description 6
- 241000797992 Richmondia Species 0.000 claims description 6
- 241000223104 Trypanosoma Species 0.000 claims description 6
- 241000224016 Plasmodium Species 0.000 claims description 5
- 241000223997 Toxoplasma gondii Species 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 5
- 208000015181 infectious disease Diseases 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 5
- 210000000601 blood cell Anatomy 0.000 claims description 2
- 210000003812 trophozoite Anatomy 0.000 claims description 2
- 230000015572 biosynthetic process Effects 0.000 claims 1
- 238000003745 diagnosis Methods 0.000 abstract description 9
- 201000010099 disease Diseases 0.000 abstract description 3
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 abstract description 3
- 244000054366 zoonotic parasite Species 0.000 abstract description 3
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 230000008030 elimination Effects 0.000 abstract 1
- 238000003379 elimination reaction Methods 0.000 abstract 1
- 238000012124 rapid diagnostic test Methods 0.000 description 12
- 238000012360 testing method Methods 0.000 description 12
- 210000004369 blood Anatomy 0.000 description 9
- 239000008280 blood Substances 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 230000034994 death Effects 0.000 description 6
- 231100000517 death Toxicity 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 238000000386 microscopy Methods 0.000 description 6
- 230000035945 sensitivity Effects 0.000 description 6
- 238000012545 processing Methods 0.000 description 5
- 239000000427 antigen Substances 0.000 description 3
- 108091007433 antigens Proteins 0.000 description 3
- 102000036639 antigens Human genes 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 241001465754 Metazoa Species 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000013145 classification model Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000010186 staining Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 208000032754 Infant Death Diseases 0.000 description 1
- 241000223821 Plasmodium malariae Species 0.000 description 1
- 241000223996 Toxoplasma Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000000339 bright-field microscopy Methods 0.000 description 1
- 238000012136 culture method Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000013399 early diagnosis Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 210000003743 erythrocyte Anatomy 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- XDDAORKBJWWYJS-UHFFFAOYSA-N glyphosate Chemical compound OC(=O)CNCP(O)(O)=O XDDAORKBJWWYJS-UHFFFAOYSA-N 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000000813 microbial effect Effects 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007899 nucleic acid hybridization Methods 0.000 description 1
- 150000007523 nucleic acids Chemical group 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 229940118768 plasmodium malariae Drugs 0.000 description 1
- 238000003752 polymerase chain reaction Methods 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 102220096713 rs876659643 Human genes 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000011120 smear test Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Abstract
The present invention relates to the technical field of disease diagnosis and image detection. Provided are a parasite (for example, a zoonotic parasite) detection system, detection device and detection method. The parasite detection system comprises an image collection module, an interference elimination module, an image segmentation module, an image classification module and a classification result determination module. By means of the present invention, automatic detection can be directly realized on site, timely on-site detection can be realized by means of common manual photographing, and the detection accuracy rate is greatly improved, thereby reducing the work intensity of detection personnel, improving the detection efficiency, and having a wide application value. Moreover, by means of the present invention, costs can be saved on, and the complexity of parasite detection can be reduced, which is beneficial for popularization and application thereof.
Description
DESCRIPTION
PARASITE DETECTION METHOD AND SYSTEM BASED ON ARTIFICIAL
INTELLIGENCE, AND TERMINAL DEVICE
Technical field
The present invention relates to the field of disease diagnosis and microbial morphology image récognition technology, as well as the field of artificial intelligence, providing a parasite identification, détection method and détection system and détection equipment.
Background technology
Common zoonotic parasites include Plasmodium malaria, Babesiella, amoeba species, Richmondia donovali, Toxoplasma gondii, Trypanosoma irhelmeria, etc. Malaria is currently endemic in a broadband near the equator, in the Americas, in many parts of Asia and much of Africa; in sub-Saharan Africa, 85-90% of malaria deaths occur. According to the world malaria report of 2016, there were about 214 million malaria cases and 438 000 malaria deaths worldwide in 2016. This burden is heaviest in Africa, where an estimated 92% of malaria deaths occur and more than two-thirds of ail deaths among children under 5 years of âge. WHO estimâtes that there were 228 million new cases of malaria in 2018, resulting in 405 000 deaths. Most cases (65%) occur in children under 15 years of âge. About 125 million prégnant women are at risk of infection each year; in sub-Saharan Africa, maternai malaria is associated with up to 200 000 infant deaths each year. Early diagnosis and treatment of malaria reduces the spread of the disease and prevents death. However, for low-income countries, there is a severe shortage of experienced parasitelogists in sanitation. Therefore, one of the obstacles to the décliné in mortality in these régions or countries is inaccurate malaria testing. For parasites, détection is crucial. At present, there are obvious defects in the détection methods of parasites, mainly including:
First, ït îs a relatively traditional microscopie inspection method. The sample is stained into a slide and placed under a microscope to observe the morphological characteristics to détermine the presence of parasites. However, this détection method needs to rely on the professional level of the inspecter, so there are probiems such as low efficiency, high work intensîty, and easy fatigue of personnel. One is the culture propagation détection method. That is, under certain nutritional conditions, the microorganisms in the sample are cultured, and the growth characteristics of the propagation and expansion are observed to détermine whether there are parasites. However, the culture method requires the parasite to be grown on the medium for 18-24 hours, which is longer and therefore inefficient. The other is antigen or antibody détection i
methods. The presence of parasites is determined by injecting spécifie antibodies or antigens into the sample and observing whether a spécifie reaction occurs. This obviously requîtes a clear antigen-antibody response, and a clear antigen-antibody response is required for different parasites. In practice, it is often the case that there is a relatively clear suspicion of parasites, and the corresponding antigen-antibody reaction is selected for détection. Relatively advanced is the method of genetic testing. Nucleic acid hybridization, gene chips, polymerase réactions and other technologies are used to determine the presence of parasites based on whether spécifie nucleic acid sequences of parasites are detected. This method is relatively costly, and its application in remote mountainous and underdeveloped areas îs not very common.
Malaria is usually diagnosed by microscopie examination of the blood by taking a blood smear or rapid diagnostic tests (RDTs). Methods for detecting plasmodium DNA using polymerase chain reaction hâve been developed, but due to cost and complexity, they are not widely used in areas where malaria is common. The use of bright/bright field microscopy to visualize Jimsa-stained thick and/or thin blood smears îs currently the gold standard and main method for the évaluation of blood samples for malaria testing. A thin blood smear may provide a positive or négative screening test to determine the presence of parasites in the blood smear. High-magnifîcation and high-resolution white light microscopy imaging was used to identify and morphological évaluation of parasite species in blood smears. However, in addition to trained microscopy techniciens, tradîtional white light microscopy often requires clinical laboratory structures, which are rare in ail areas where malaria is most prévalent. In addition, based on the training of expert microscope techniciens and the equipment used, current microscopy techniques lead to subjective measurements, which are reported to vary widely.
Rapid diagnostic tests (RDTs) are widely used around the world, providing a cheaper and more time-consuming alternative to dîagnosing malaria with blood drawn from fingers. However, RDTs currently used for malaria testing include antigen-based détection protocols. Therefore, the performance of RDT in the tropics has been reported as degraded due to sensing chemistry. In addition, existing RDTs détection limits are often much higher than current gold-standard tools for detecting early infection. In addition, these devices do not provide quantitative parasitaemia results and are inconsistent in dîagnosing strain-spécifie malaria infection. In addition, RDTs are not effective in dîagnosing low parasite density. Clearly, the lack of experienced and skilled microscopes is a key challenge that needs to be addressed immediately by Afrîca's malaria prévention goals. Recentiy, there are some new technologies to apply parasite détection, such as with the development of deep learning, there is also a détection method using deep îearning, but its data collection equipment is électron microscopy, which îs expensive, hindering the general use of électron microscopy, thereby limîting the application of deep learning în malaria détection
Systems. In addition, the proposed method adopts two deep learnîng models, which overutilizes computing resources, thereby limiting the popularity of the method. For example, the Chinese patent application for CNl 10807426A involves a deep learning-based parasite détection system, which involves image processing technology. However, these prior art methods still lack efficient 5 means of direct détection on site, and the détection accuracy needs to be improved, the present invention is developed based on this purpose.
Invention content
In view of this, embodiments of the present invention provide a parasite détection method, 10 détection system and terminal equipment to solve the prior art în high cost and high complexity, in low-income people, or countries or problems in remote areas, which simply cannot be widely applied. The parasite détection method and the détection model in the system in the present invention is based on the VGG (Visual Geometry Group) framework, optimized artificial intelligence (AI) with skip connections diagnostic (Diagnosis) mode), known as the AIDMAN 15 model. Specifically, the parasite détection model of the present invention îs optimized and modified based on the VGG framework, by changing the convoiutional layer in the encoder to a deep space convolutional layer, in order to reduce the training parameters of the model. thereby reducing the running time; by connecting the encoder and the décoder, the features of different scales are fused, and the features learned by the model are increased, and the accuracy of the 20 system is increased.
Thus, the present invention first provides a parasite détection System, wherein the parasite is not limited to Plasmodium malaria, Babesiella, amoeba parasites, Richmondia donovali, Toxoplasma gondii, Trypanosoma irhelmeria and other zoonotic parasites, and these parasites form a cyclic and / or trophozoites in or around the blood cells in the host after infection, the 25 system comprises:
Image acquisition module for acquiring images to be inspected;
The de-interference module is used to de-interfere with the image to be detected, and obtains the image to be detected after de-interference;
Image segmentation module, for segmenting the image to be detected after de-interference, 30 to obtain a plurality of cell images to be detected;
The image classification module îs used to înput the plurality of cell images to be detected into the trained parasite détection model respectively, and obtains the classification results corresponding to the plurality of cell images to be detected;
The classification resuit détermination module is used to détermine the absence of parasites 35 in the image to be detected if the classification results corresponding to the plurality of cell images ίο be detected are ail free of parasites; if any one of the multiple cell images to be detected corresponds to the classification resuit of the presence of parasites, it is determined that there are parasites in the images to be tested.
Preferably, the resuit display module is further included.
In one embodiment, the image to be detected is de-interfered to obtain the image to be detected after de-interférence, comprising; obtaining a circular image of the inscrîbed square image of the image to be inspected as the image to be detected after de-înterference.
In a preferred embodiment, the image to be detected after de-interference îs segmented, and a plurality of cell images to be detected is obtained, comprising:
The image to be detected after de-interference is grayscale to obtain a grayscale image;
According to the grayscale image, the grayscale value statistical chart is obtained; wherein, the abscîssa of the gray value statistical chart is the gray value, and the ordinate is the number of occurrences of the corresponding gray value in the grayscale image;
In the gray value chart, the target gray value is obtained; wherein, between the first and the second gray value, the ordinate value corresponding to the target gray value is the smallest, and the first gray value corresponds the ordinate and the ordinate corresponding to the second gray value are the two peaks in the gray value chart;
According to the target gray value, the image to be detected after de-interference is segmented to obtain a plurality of cell images to be detected.
In one spécifie embodiment, the parasite détection model is obtained by the foilowing method:
Obtain a set of training samples comprising a plurality of training images;
The de-interference operation is performed on each training image to obtain the de-jamming training image.
The training images after de-interference are segmented separately, and multiple training cell images corresponding to each training image are obtained. Wherein, each training cell image has been labeled as parasites present or absent;
Based on a plurality of training cell images corresponding to each training image, the pre-constructed parasite détection model is trained, and the trained parasite détection model is obtained.
More preferably, the parasite détection model comprises thirteen deep spatial convolutional layers and five maximum pooling layers; in the decoding process, the parasite détection model converts the features of each dimension into features of the same size through a fully connected layer, and adds the features of the same size in turn to obtain a plurality of features sum, and by flattening layer and fully connected layer, the features of sums of each feature are extracted;
during the decoding process. the parasite détection model classifies the image of the cells to be detected by means of a Softmax activation function.
The parasite détection model is based on the VG G (Visual Geometry Group) framework, optimized for artificial intelligence (AI) diagnosis (Diagnosis) with skip connections model, called the AlDMAN model. Specifically, the parasite détection model of the present invention is optimized and modified based on the VGG framework, by changing the convolutional layer in the encoder to a deep space convolutional layer, in order to reduce the training parameters of the model, thereby reducing the running time; by connecting the encoder and the décoder, the features of different scales are fused, and the features learned by the model are increased, and the accuracy ofthe system is increased.
Wherein the detected parasite is one of the malaria parasites, Babesiella, amoeba parasites, Richmondia donovali, Toxoplasma gondü, Trypanosoma irhelmeria respectively trained to obtain the training model of the corresponding parasites, preferably a variety of parasites training models to constitute a training model set, automatically detecting the détection object or providing a choice of which parasite to detect at the time of détection.
In one embodiment, the image acquisition module comprises a direct reading or wireless method to obtain an image, preferably the image îs acquired by a mobile terminal such as a mobile phone.
In a spécifie embodiment, the classification resuit display module displays the results through a remote wireless upload network, or displays the results dîrectly on the screen via a display device.
The present invention also provides a détection device comprising the above-mentioned parasite détection system, which comprises a memory, a processor, and a computer program stored in the memory and may run on the processor, the processor exécutés the computer program, and the computer program encodes to implement the parasite détection system , preferably further comprises a supportîng image caméra device such as a mobile phone, and a display device such as a screen or remote resuit dîsplay.
At the same time, the present invention also provides a method for detecting parasites using the above-mentioned détection equipment, and includes the following steps or processes when detecting:
Acquîre the image to be inspected;
The de-interférence operation is carried out on the detected image to obtain the image to be detected after de-interference;
The image to be detected after de-interference is segmented to obtain multiple cell images to be detected;
Multiple cell images to be detected were input into the trained malaria parasite détection model. and the classification results corresponding to multiple cell images to be detected were obtained.
If the classification results corresponding to multiple cell images to be tested are the absence of malaria parasites, it is determined that there is no malaria parasite in the images to be tested;
If any one of the multiple cell images to be tested corresponds to a classification resuit of the presence of malaria parasites, the presence of malaria parasites in the images to be tested is determined.
The test may be non-diagnostic and refers to a test that is not performed on a iiving human or animal, such as takîng environmental samples to detect the presence of malaria parasites, environmental samples such as soil, water samples, and other environmental samples that may present malaria parasites, such as dead animal bodies.
Further. the present invention also provides a computer-readable storage medium, computer-readable storage medium stores computer programs, computer programs are executed by one or more processors when implemented such as the first aspect of malaria parasite détection system.
The bénéficiai effect of the embodiment of the present invention compared with the prior art is: the present invention first obtains the image to be detected, performs de-interférence operation on the detected image, obtains the image to be detected after de-interférence, can reduce the size of the image to be detected, thereby reducing the subséquent amount of calculation; then, the image to be detected after de-interference is segmented to obtain multiple cell images to be detected, and multiple cell images to be detected are separated. It is input into the trained malaria parasite détection model to obtain the classification results corresponding to multiple cell images to be detected. According to the above classification results, détermine whether there is malaria parasites în the images to be detected, and the malaria parasite détection mode! can automatically ciassify the detected images, which can reduce human labor; the invention can save costs, reduce the complexity of malaria parasite détection, and can be widely used in low-income countries.
Description of the drawings
In order to more clearly illustrate the technical solution of the present invention, the following will be embodiments or prior art description of the need to use the drawings briefly introduced, obviously, the following description of the drawings are only some embodiments of the present invention, for those of ordinary skill in the art, without the premise of créative labor, may also obtain other drawings according to these drawings.
FIG l is a schematic diagram of the implémentation process of the malaria parasite détection method of the present invention;
FIG 2 is a schematic diagram of each image in the image processing process of the present invention (application embodiment l);
FIG 3 is a schematic diagram of the structure of the malaria parasite détection model of the present invention;
FIG 4 is a schematic diagram of the structure ofthe malaria parasite détection system of the present invention;
FIG 5 is a schematic view of the structure of the terminal apparatus ofthe present invention.
FIG. 6 R.OC curve in application embodiment l.
FIG. 7a Some pictures and R.DT diagnostic results of uninfected persons in application embodiment 2.
F IG.7b The pîcture and R.DT diagnostic results of the infected person in application embodiment 2.
Spécifie embodiment
In the following description, for illustrative rather than qualifying, spécifie details such as a particular system structure, technology and the like are proposed in order to thoroughly understand embodiments of the present invention. However, those skilled in the art should be clear that the present invention may also be implemented in other embodiments without these spécifie details. In other cases, the weil-known System, apparatus, circuit, and method description îs omitted, so as not to interfère with the description of the present invention with unnecessary details.
In order to illustrate the technical solution described in the present invention, the following is illustrated by spécifie embodiments.
FIG l is a schematic diagram of the implémentation process of a parasite détection method provided by an embodiment of the present invention, for illustrative purposes, only a portion related to an embodiment of the présent invention is shown. The subject of execution of embodiments of the present invention may be a terminal device testing equipment. Wherein the parasites are, for example, Plasmodium malariae, Babesiella, amoeba parasites, Richmondia donovali, Toxoplasma gondii, Trypanosoma irhelmeria, i.e., the present invention is suitable for détection of these parasites.
As shown in Figure i, the above parasite détection method may include the following steps:
S101 : Acquîre the image to be inspected.
In one embodiment of the present invention, the above S101 may include the following steps:
The mobile terminal caméra and macro lens are used to acquire the image to be inspected.
Among them, the mobile terminal can be a mobile device with a caméra such as a mobile phone. Images can be acquired by mobile phones and macro lenses, which can reduce the cost of image acquisition. When acquiring images, place the target in the center of the lens.
Optionally, macro lenses can be replaced with optical microscopes.
S102: The de-interference operation is performed on the detected image to obtain the image to be detected after de-interference.
In one embodiment of the present invention, the image to be detected is a circular image; The above S102 can include the following steps:
Obtain the image of the circle to be inscribed with the image to be inspected as the image to be detected after de-interference.
S103: Segment the image to be detected after de-interference to obtain a plurality of cell images to be detected.
In one embodiment of the present invention, the S103 may include the following steps:
The image to be detected after de-interference is grayscale to obtain a grayscale image;
Grayscale value statistical chart obtained according to grayscale image; among them, the abscissa of the gray value chart is the gray value, and the ordînate is the number of occurrences of the corresponding gray value in the grayscale image;
In the gray value chart, obtain the target gray value; among them, between the first and the second gray value, the ordînate value corresponding to the target gray value is the smallest, and the ordînate corresponding to the first gray value and the ordînate corresponding to the second gray value are the two peaks in the gray value statistical chart;
According to the target gray value, the image to be detected after de-interference is segmented to obtain multiple cell images to be detected.
S104: Input multiple cell images to be detected into the traîned parasite détection model to obtain the classification results corresponding to multiple cell images to be detected.
In an embodiment of the present invention, a plurality of cell images to be detected are input into the traîned parasite détection model, and the classification results corresponding to each cell image to be detected can be obtained, and the classification resuit is the presence of parasites or the absence of parasites.
In one embodiment of the present invention, prior to S104 described above, the parasite détection method may further include the following steps;
Get a training sample set, which includes multiple training images;
The de-interference operation is performed on each training image to obtain the de-jamming training image.
The training images after de-interference are segmented separateiy, and multiple training cell images corresponding to each training image are obtained. Wherein, each training cell image has been labeled as parasites present or absent;
Based on multiple training cell images corresponding to each training image, the pre-built parasite détection model is trained, and the trained parasite détection model is obtained.
Among them, the training images in the training sample set can also be collected by professionals through mobile phones and macro lenses. The de-interference operation of the training image is the same as the de-interference operation ofthe detected image described above, and the process of segmenting the training image after de-interference is the same as the above de-interference of the image to be detected. The process of segmentation is the same and will not be repeated here. Multiple segmented training cell images corresponding to each training image are marked by experts as parasite presence or parasite absence to facilitate parasite détection model training.
In one embodiment of the present invention, the model of the present invention consists of two parts: an encoder and a décoder. The model is improved based on VGG to reduce the training parameters of the model by changing the convolutional layer in the encoder to a deep space convolutional layer, thereby reducing the running time; by connecting the encoder and the décoder, the characteristîcs of different scales can be fused, increase the features learned by the model, and increase the accuracy of the System. The spécifie structure of the model is as follows; in the coding process, the parasite détection model includes thirteen deep spatial convolutional layers and five maximum pooîing layers; in the decoding process, the parasite détection mode! converts the features of each dimension into features of the same size through the fully connected layer, and adds the features of the same size in turn to obtain multiple feature sums, and extracts the features of each feature sum through the flattening layer and the fully connected layer. During the decoding process, the parasite détection model classifies the image of the cells to be detected by means of the Softmax activation function.
Specifically, Figure 3 is a schematic dîagram of the structure of the parasite détection model, as shown in Figure 3, 31 is the input layer, 32 is the deep spatial convolutional layer, 33 is the maximum pooling layer, and 34 is the flattening layer, 35 is the fully connected layer, 36 is the addition operation, 37 is the discarded layer, and 38 is the output layer.
The parasite détection model, which includes thirteen deep spatial convolutional layers and five maximum pooling layers during coding. Assuming that the input is a feature vector with dimension n, if the convoiution kernel of (3,3) is used, the step size is 1, and the output dimension is m, the required traîning parameters are (3*3*n+l * 1 *m); If the convoiution kernel of (3,3) is used, the convolutional layer with a step size of 1 and an output dimension of m is required, and the required training parameters are (3*3*n*m). Therefore, replacing the convolutional layer with a deep spatial convolutional layer can greatly reduce the traîning parameters and shorten the traîning time.
In the parasite détection model, in the decoding process, the fuHy connected layer is used to convert the features ofthe four dimensions into features ofthe same size, and the four features of the same size are added to obtain the sum of three features, that is, the first feature is added to the second feature sum by adding the second feature sum, the first feature and the third feature are added to get the second feature sum, and the second feature and the fourth feature are added to get the third feature sum. Then, the flattened layer and the fully connected layer are used to extract the features and features of each feature. By connecting the jumping layers, features at multiple scales can be learned instead of judging the image by only the last feature, which can make more accurate judgments and împrove the accuracy of parasite détection. Finally, the softmax activation function is used to classify the cell images to be detected and détermine whether there are parasites in the cell images to be detected.
In one embodiment of the present invention, the parasite détection model is based on the VGG architecture, modîfied to optimize artificial intelligence (AI) diagnosis (Diagnosis) with skip connections model, called the AID model.
S105: If the classification results corresponding to multiple cell images to be tested are ail free of parasites, it is determined that there are no parasites in the images to be tested; if any one of the multiple cell images to be tested corresponds to a classification resuit of the presence of parasites, the presence of parasites in the image to be tested is determined.
The classification results of multiple cell images to be tested can détermine whether parasites are present in the images to be tested, and thus whether the person corresponding to the image to be tested has malaria (whether he has been infected by parasites). Specifically, if there are no parasites in multiple cell images to be tested, it is determined that there are no parasites in the images to be tested, that is, the person corresponding to the image to be tested is not sick Malaria; If a parasite is present in any of the multiple cell images to be tested, it is determined that the parasite îs present in the image to be tested, that is, the person corresponding to the image to be tested has malaria.
It can be seen from the above description that the embodiment of the present invention fîrst obtains the image to be detected, performs a de-interférence operation on the detected image, obtains the image to be detected after de-înterference, can reduce the size of the image to be detected, thereby reducing the subséquent amount of calculation; then, the image to be detected after de-interference is segmented to obtain multiple cell images to be detected, and multiple cell images to be detected are separated. it is input into the trained parasite détection model, and the classification results corresponding to multiple cell images to be detected are obtained. According to the above classification results, detennine whether there are parasites in the image to be detected, and the parasite détection model can automatically classîfy the detected image, which can reduce human labor; the present invention can save costs and reduce the complexity of parasite détection in low-income places such as malaria-prone Africa as well where there is a shortage of experienced and skilied malaria testing experts, it can be widely cited, which in turn can save more lives.
It should be understood that the size of the serial number of each step in the above embodiment does not imply the order of execution, and the order of execution of each process should be determined by its function and internai logic, and should not constîtute any limitation of the embodiment of the present invention.
Corresponding to the parasite détection method described above, embodiments of the present invention also provide a parasite détection system. FIG 4 îs a schematic diagram of the structure of a parasite détection system provided by an embodiment of the present invention, for illustrative purposes, only a portion related to an embodiment of the present invention is shown.
Referring to FIG. 4, the corresponding parasite détection System 400 may include: image acquisition module 401, de-interference module 402, image segmentation module 403, image classification module 404 and classification results détermination module 405.
wherein, the image acquisition module 401 is used to acquîre the image to be inspected;
The de-interference module 402 is used to de-interfere with the image to be detected, and the image to be detected after de-interférence is obtained;
Image segmentation module 403, configured to segment the image to be detected after de-interference, to obtain a plurality of cell images to be detected;
Image classification module 404, for entering the plurality of cell images to be detected into the trained parasite détection model separately. The classification results corresponding to the multiple cell images to be detected are obtained;
Classification results détermination module 405, for determining the absence of parasites in the image to be detected if the classification results corresponding to a plurality of cell images to be detected are ail absent; if any one of the multiple cell images to be detected corresponds to the classification resuit of the presence of parasites, it is determined that there are parasites in the images to be tested.
Optionally, the image to be inspected is a circular image;
1
The de-interference module 402 is specifically used for:
Obtain a circular image of the image to be înscribed with the image to be inspected as the image to be detected after de-interference.
Optionally, the image segmentation module 403 is specifically for:
The image to be detected after de-interference is grayscale to obtain a grayscale image;
According to the grayscale image, the grayscale value statistical chart is obtained; wherein, the abscissa of the gray value statistical chart is the gray value, and the ordînate is the number of occurrences of the corresponding gray value in the grayscale image;
in the gray value chart, the target gray value is obtained; wherein, between the first and the second gray value, the ordînate value corresponding to the target gray value is the smallest, and the first gray value corresponds the ordînate and the ordînate corresponding to the second gray value are the two peaks in the gray value chart;
According to the target gray value, the image to be detected after de-interference is segmented to obtain a plurality of cell images to be detected.
Optionally, the parasite détection system 200 may also include: a training module.
a training module for obtaining a set of training samples comprising a plurality of training images; the de-interference operation is performed on each training image to obtain the de-jamming training image. The training images after de-interference are segmented separately, and multiple training cell images corresponding to each training image are obtained. Wherein, each training cell image has been labeled as parasites present or absent; based on a plurality of training cell images corresponding to each training image, the pre-constructed parasite détection mode! îs trained, and the traîned parasite détection model is obtained.
Optionally, during coding, the parasite détection model comprises thirteen deep spatial convolutional layers and five maximum pooling layers; in the decoding process, the parasite détection model couverts the features of each dimension into features of the same size through a fully connected layer, and adds the features of the same size in turn to obtain a plurality of features sum. And by flattening layer and fully connected layer, the features of sums of each feature are extracted; during the decoding process, the parasite détection model classifies the image of the cells to be detected by means of a Softmax activation function.
The parasite détection model is based on the VGG architecture, and the artificial intelligence (AI) diagnostic (Diagnosis) model with Skip connections is optimized and optimized, called Al DM AN model. Wherein the detected parasite is a training model of Plasmodium malaria, Babesiella, amoeba parasite, Richmondia donovali, Toxoplasma gondii, Trypanosoma irhelmeria respectively trained to obtain the corresponding parasite, and more specifically may be a single training model, such as a training model for malaria parasites, it can also be a training model of multiple parasites to fonn a training model set, which automatically detects the détection object or provides a choice of which parasite to detect when it is detected.
Optionally, the image acquisition module 401 is specifically used for:
The mobile terminal caméra and macro lens are used to acquire the image to be inspected.
Those skilled in the art can clearly understand, in order to describe the convenicnce and conciseness, only to the above functional units, modules of the division for example, in practical applications, the above functions can be assigned by different functional units, modules according to the needs to complété, that is, the internai structure of the parasite détection system is divided into different functional units or modules, to complété ail or part of the functions described above. Each functional unit and module in an embodiment may be integrated in a processing unit, or each unît may exist physically alone, or two or more units may be integrated in one unit, and the integrated unit may be implemented in the fonn of hardware or software functional unit. In addition, the spécifie names of each functional unit and module are only for the convenience of distînguishing each other, and are not used to limit the scope of protection of the present application. The spécifie working process of the unit and module in the above device mav refer to the corresponding process in the embodiment of the aforementioned method, and will not be repeated herein.
FIG 5 is a schematîc block diagram of a terminal détection device provided by an embodiment of the present invention. As shown in FIG. 5, the terminal device 500 of this embodiment comprises: one or more processors 501, memory 502 and a computer program 503 stored in the memory 502 and may run on the processor 501. The processor 501 implements the steps in each embodiment of the parasite détection method described above when executing the computer program 503, such as steps S10I to S105 shown in FIG. 1. Alternatively, the processor 501 implements the functions of each module / unit în the embodiment of the parasite détection system when executing the computer program 503, such as modules 401 to 405 shown in FIG. 4 of the feature.
In an exemplary way, the computer program 503 may be divided into one or more modules / units, the one or more modules / units are stored in the memory 502, and executed by the processor 50] to complété the present application. The module / unit may be a sériés of computer program instruction segments capable of completing a particular function, the instruction segment is used to describe the computer program 503 în the terminal device 500 execution process. For exampie, the computer program 503 may be segmented into an image acquisition module, a de-interference module, an image segmentation module, an image classification module and a classification resuit détermination module, and the spécifie functions of each module are as follows:
Image acquisition module for acquîring images to be inspected;
The de-interference module is used to de-interfere with the image to be detected, and the image to be detected after de-interference is obtained;
The image segmentation module, for segmenting the image to be detected after de-interference, to obtain a plurality of cell images to be detected;
The image classification module îs used to input the plurality of cell images to be detected into the trained parasite détection model respectively. The classification results corresponding to the plurality of cell images to be detected;
The classification resuit détermination module is used to détermine the absence of parasites in the image to be detected if the classification results corresponding to the plurality of cell images to be detected are ail free of parasites; if any one of the multiple cell images to be detected corresponds to the classification resuit of the presence of parasites, it is determined that there are parasites in the images to be tested.
Other modules or units may refer to the description in the embodiment shown in FIG. 4, and will not be repeated herein.
The terminal device 500 may be a desktop computer, notebook, handheld computer, mobile phone, embedded device and cloud server and other computing devices. The terminal device 500 includes but is not limited to processor 501, memory 502. Those skilled in the art may understand that FIG. 5 is only an example of the terminal device 500, does not constitute a limit to the terminal device 500, may include more or fewer components than shown, or a combination of certain components, or different components, such as the terminal device 500 may also include input devices, output devices, network access devices, buses, etc.
The processor 501 may be a central processing unit (CPU), may also be other general-purpose processors, digital signal processors (DSP), application spécifie integrated circuits (ASICs), field programmable gâte arrays (FPGA) or other programmable logic devices, discrète gates or transistor logic devices, discrète hardware components, etc. A general-purpose processor can be a microprocessor, or the processor can be any conventional processor, etc.
The memory 502 may be the internai memory unit of the terminal device 500, such as a hard disk or memory of the terminal device 500. The memory 502 may also be an external storage device of the terminal device 500, such as a pluggable hard disk equipped on the terminal device 500, a smart memory card (Smart Media Card, SMC), a Secure Digital (SD) card, a flash memory card (Flash Card) and the like. Further, the memory 502 may further include both an internai memory unit of the terminal device 500 and an external storage device. The memory 502 îs used to store the computer program 503 and other programs and data required by the terminal device 500. The memory 502 may also be used to temporarily store input related data, such as image information and sample-related information obtained by remote încoming or direct input image acquisition module. The memory 502 may also be used to temporarily store the data that has been output or will be output, and finally the results are displayed by the classification resuit display module via a remote wireless upload network, or by a display device if the screen directiy displays the results.
Application embodiment
Application embodiment 1
In a practîcal example test samples, the samples were taken from the blood smear of 120 local patients at the China-Sierra Leone Friendship Hospital in Sîetra Leone, Africa, and images of the samples to be tested were obtained through a mobile phone (Huaweî Meta20X). The images acquired by S101 are ail circular images, and the resolution of the acquired images is 3648*2736. As shown in Figure 2, the leftmost image of Figure 2 îs the acquired image to be inspected. and the second image from the left of Figure 2 is the image to be detected after de-interference. The image of the inscribed square is captured in the middle of the image to be detected as the image to be detected after de-interference, and the rest îs discarded.
When the image is collected with a mobile phone, there will be some interférence factors, and the part in the middle of the image is needed, so the image of the inscribed square of the intercepted image îs used as the target for subséquent détection, and its size is 1733*1733. This can avoid the interférence of the cells in the part where the circie tangentially meets the black border, and ai the same time reduce the size of the image, which can reduce the amount of subséquent calculations and reduce the occupation of computer resources by subséquent steps. In this example, the image to be detected after de-interference îs first grayscale processed to obtain a grayscale image. Then, count the number of occurrences of each grayscale value in the grayscale image to obtain the grayscale value statistical chart. Among them, in grayscale images, the range of grayscale value (pixel value) is 0-255. As shown in Figure 2, the third image front the left îs a grayscale value statistical chart, whose abscissa îs a continuons integer gray value from 0 to 255, and the ordinate is the number of times the grayscale value corresponding to the abscissa appears in the grayscale image.
Since the color différence between the cell région and the extracellular région is large, and the cell région and the extracellular région account for most of the image, and the cell edge part accounts for a small part, the third image in Figure 2 is selected, between the two peaks, the gray value of the abscissa corresponding to the ordinate minimum is used as the basis for image segmentation, and the image îs segmented, and the cell région îs divided from the extracellular région, leaving only the cell région. Among them, the images of multiple cells to be detected after segmentation are shown in the fourth image from the left in Figure 2. The image of the cell pîcture after the segmentation module was passed into the détection system, and it was found that there were no cells infected by the malaria parasite in the segmented picture, so the sample picture was not infected by the malaria parasite.
The indicators of this system in the ten-fold vérification are shown in the table, the model predicts the table indicators obtained by predicting 1 709 cell images that hâve never appeared în the training set, and the average number of prédiction errors in these 1709 pictures is 1592 and the average number of prédiction errors is 137. The model used in this system has good accuracy (99.20 ± 0.49%), sensitivity (99.21 ± 0.49%), specificity (99.04 ± 1.02%), précision (99.21 ± 0.49%) and AUC = 99.94 ± 0.09. Among them, AUC (Area Under Curve) is the area under the ROC curve, which can indicate that the probabîlity of predicting positive samples correctly is greater than the probabîlity of predicting négative sample errors, that is, the larger AUC, the more accurate the model predicts. The average value of these indicators is greater than 99%, and the larger the area under the ROC curve, the higher the accuracy of the model, from Figure 6 it can be seen that the area under the ROC curve is basically equal to the rectangle enclosed by this coordinate system, so it can be proved that the prédiction accuracy of the system is extremely high, and the error is low. These indicators are calculated by the confusion matrix, in which TP is the number of samples with positive labels, and the model prédiction results are also positive samples; FN is the number of samples labeled positive and the model predicts négative samples; FP is the number of samples whose labels are négative and the model predicts that the resuit is positive; TF is the number of samples with négative labels, and the model predicts that the results are also négative. Accuracy = (TP+TN)/(TP+TN+FP+FN), which means the proportion of ail correct judgments of the classification model to the total observations; Sensitivity = TP/(TP+FP) which means the proportion of correct prédiction in the results predicted by the classification model as a positive sample; Précision = Sensitivity = TP/ (TP+FN) which means that the spécifie gravity predicted correctly by the model in ail outcomes where the true value is positive; Specificity = TN/(TN+FP) which means that in ail outcomes where the true value is négative, the spécifie weight of the model predicting correctly, Fl score = 2*Sensitivîty*Precision/(Sensitivity + Précision) is the harmonie average of sensitivity and précision, which is used to measure the performance of the system. ROC (Receiver Operating Characteristic) curve predicts the threshold from 0 to 1 according to the prédiction results obtained by the model, that is, each sample predicted as a positive sample is înitially treated, and as the threshold increases, the number of positive samples predicted by the model is less. Until the final prédiction results are ail less than or equal to the maximum threshold of 1, two important values are calculated each time in the process, which are used as the abscissor and ordinale ofthe ROC curve. Abscissa FPR (False Positive Rate) = FP/(TN+FP), and ordinate TPR (True Positive Rate) = TP/(TP+FN). The results are shown in the table below and Figure 6.
model | Accuracy | Sensitivity | Précision | Specificity | Fl score | AUC |
0 | 98.595 | 0.986 | 0.986 | 0.97 | 0.986 | 0.978 |
1 | 99.297 | 0.993 | 0.993 | 0.998 | 0.993 | 0.995 |
2 | 99.297 | 0.993 | 0.993 | 0.998 | 0.993 | 0.995 |
3 | 99.766 | 0.998 | 0.998 | 0.996 | 0.998 | 0.997 |
4 | 99.745 | 0.994 | 0.994 | 0.998 | 0.994 | 0.996 |
5 | 98.829 | 0.988 | 0.988 | 0.986 | 0.988 | 0.987 |
6 | 99.532 | 0.995 | 0.995 | 0.995 | 0.995 | 0.995 |
7 | 99.649 | 0.997 | 0.996 | 0.999 | 0.997 | 0.998 |
8 | 98.244 | 0.983 | 0.982 | 0.976 | 0.983 | 0.979 |
9 | 99.415 | 0.994 | 0.994 | 0.988 | 0.994 | 0.991 |
average | 99.2369 | 0.9921 | 0.9919 | 0.9904 | 0.9921 | 0.9911 |
Application embodiment 2
The present embodiment is an application of the test results of the present invention with RDT détection and expert test results analysis. The results were analyzed based on the mobile phone pictures of 38 patients of the China-Sierra Leone Friendship Hospital (Huawei Meta20X, the résolution of the acquired images was 3648*2736. Smear of these patients had never been used to develop a model before. After staining, if infected by malaria parasites, they will appear purple ring-shaped, or purple-banana-lîke, which hâve obvious parasite-related characteristics; if the cells are not infected by the malaria parasite, the cells are not stained or hâve irregular purple patterns due to imperfect staining processes. Experts tested 38 patients, 20 of whom were infected and 18 were uninfected. In the AI model, a patient is considered infected when at least one single red blood cell containing the parasite is seen in the picture. The prédictions of the AIDMAN MODEL detected 19 out of 20 infected people as infected, and only one piciure was misjudged. In addition, the 18-bit uninfected images were judged corrcctly by the AI-modeL Importantly, the AIDMAN model of the present invention detects that a négative resuit reported by RDT contains two positive smears, and a positive resuit reported by RDT includes two négative pictures (see Fig. 7a and Fig. 7b). The results show that the algorithm of the present invention is exactly the same as the results of expert détection, and the accuracy of RDT détection results relative to the expert diagnosis results is 89.47%. Therefore, the performance of the AIDMAN model of the present invention is comparable to that of experts, and has a very practical value.
In the above embodiments, the description of each embodiment has its own emphasis, and the part not detailcd or documented in a certain embodiment may refer to the relevant description of other embodiments.
Those of ordinary skill in the art may be aware that the units and algorithm steps described in combination with the embodiments disclosed herein may be implemented in combination with electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software dépends on the spécifie application and design constraints of the technical solution. Professional technical personnel may use different methods for each particular application to achieve the described function, but such implémentation should not be consîdered beyond the scope of the present application.
In the embodiments provided in the present application, it should be understood that the disclosed system and method may be implemented by other means. For example, the system embodiments described above are only schematic, for example, the division of the module or unit, only for a logical function division, the actual implémentation may hâve another division method, such as a plurality of units or components may be combined or integrated into another system, or some features may be ignored, or not performed. Another point, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, which may be electrical, mechanical or other.
The unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical unit, i.e., may be located in one place, or may also be distributed on a plurality of network units. Some or ail of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
Further, each functional unit în each embodiment of the present application may be integrated in a processing unit, or each unit may exist physically alone, or two or more units may be integrated in a unit. The above integrated unit can be implemented in the form of hardware or software function unit.
The integrated module / unit may be stored in a computer-readable storage medium if implemented in the form of a software functional unît and sold or used as a separate product. Based on this understanding, the present application implements ail or part of the process in the method of the above embodiment, and may also be completed by a computer program to instruct the related hardware, the computer program may be stored in a computer-readable storage medium, the computer program can realize the steps of each embodiment of the above method when executed by the processor. Wherein the computer program includes computer program code, the computer program code may be in source code form, object code form, exécutable file or some intermediate form, etc. The computer-readable medium may comprise; any entity or device capable of carrying the computer program code, recording medium, U disk, portable hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), 5 random access memory (RAM, Random Access Memory), electrical carrier signais, télécommunications signais, and software distribution media. It should be noted that the contents of the computer-readable medium may be appropriately increased or decreased according to the requirements of the législation and patent practice in the jurisdiction, for example, in some jurisdictions, according to the législation and patent practice, the computer-readable medium 10 does not include electrical carrier signais and télécommunications signais.
The embodiments described above are only used to illustrate the technical solution of the present application, and not to limit it; although the present application is described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand: they may still modify the technical solutions described in each of the foregoing embodiments, or 15 equivalently replace some of the technical features; and these modifications or replacements do not départ the essence of the corresponding technical solution from the spirit and scope of the technical solution of each embodiment of the present application, and shall be included in the scope of protection of the present application.
Claims (9)
1. A parasite détection System, the parasite refers to the formation of a parasite with a ring and/or trophozoite morphology in or around the blood cells in the host body after infection, which is characterized in that, comprising:
Image acquisition module for acquiring images to be inspected;
The de-interference module is used to de-interfere with the image to be inspected, and the image to be detected after de-interference is obtained, including obtaining an image of a circle of the image to be inscribed as the image to be detected after de-interference;
Image segmentation module, for segmentîng the image to be detected after de-interference, obtaining a plurality of cell images to be detected, comprising: grayscale the image to be detected after de-interference, obtaining grayscale images; according to the grayscale image, the grayscale value statistical chart îs obtained; wherein, the abscissa of the gray value statistical chart is the gray value, and the ordinate is the number of occurrences of the corresponding gray value in the grayscale image; in the gray value chart, the target gray value is obtained; wherein, between the first and the second gray value, the ordînate value corresponding to the target gray value is the smallest, and the first gray value corresponds the ordinale and the ordînate corresponding to the second gray value are the two peaks in the gray value chart; according to the target gray value, the image to be detected after de-interference is segmented, and a plurality of cell images to be detected is obtained; image classification module for entering the plurality of cell images to be detected into the trained parasite détection model separately, the classification results corresponding to the multiple cell images to be detected are obtained;
The classification resuit détermination module is used to détermine the absence of parasites in the image to be detected if the classification results corresponding to the plurality of cell images to be detected are ail free of parasites; if the classification resuit corresponding to any one of the cell images to be detected is the presence of parasites, it is determined that there are parasites in the images to be tested;
Classification resuit display module;
The parasite détection mode! is composed of two parts, encoder and décoder, the model is based on VGG, the convolutional layer in the encoder is changed to a deep spatial convolutional layer, and the characteristics of different scales are fused by connecting the encoder and the décoder by leaping, the parasite détection model includes thirteen deep spatial convolutional layers and five maximum pooling layers; In the decoding process, the parasite détection model converts the features of each dimension into features of the same size through a fully connected layer, and adds the features of the same size in turn to obtaîn a plurality of features sum. and by flattening layer and fully connected layer, the features of sum of each feature are extracted; during the decodîng process, the parasite détection mode! classifies the image of the cells to be detected by means of a softmax activation function.
2. The parasite détection system according to claim 1, wherein the parasite détection model is obtaîned by the foîlowing method:
Obtain a set of training samples comprising a plurality of training images;
The de-interférence operation îs performed on each training image to obtain the de-jamming traîning image.
The training images after de-interference are segmented separately, and multiple training cell images corresponding to each training image are obtaîned. Wherein, each training cell image has been labeled as parasites present or absent;
Based on the plurality of training cell images corresponding to each training image, the pre-constructed parasite détection mode! is trained, and the traîned parasite détection model is obtaîned;
The parasite is one or more of Plasmodium malaria, Babesiella, amoeba parasite, Richmondia donali, Toxoplasma gondii. and Trypanosoma irhelmeria training model for the corresponding parasite.
3. The parasite détection system according to claim l, wherein the image acquisition module comprises a direct reading or wireless method to obtain images.
4. The parasite détection system according to claim 3, wherein the image is acquired by a mobile terminal.
5. The parasite détection system according to claim 4, wherein the image is acquired by a mobile phone.
6. The parasite détection System according to claim 1, wherein the classification resuit dispiay module display s the results via a remote wireless upload network, or directiy display s the results through a display device.
7. A détection device comprising a parasite détection system according to claims 1 to 6, comprising a memory, a processor, and a computer program stored in the memory and may be run on the processor, the processor executing the computer program, the computer program coding implémentation of the parasite détection system according to any one of claims 1 to 6.
8. The détection device of the parasite détection system according to claim 7, wherein it further comprises a supporting image caméra device, and a display device.
9. The détection device of the parasite détection system according to claim 8, wherein the photographie device is a mobile phone, and the display device is a remote resuit display.
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010693922.0 | 2020-07-17 |
Publications (1)
Publication Number | Publication Date |
---|---|
OA21122A true OA21122A (en) | 2023-11-13 |
Family
ID=
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP6192747B2 (en) | Machine learning system based on tissue objects for automatic scoring of digital hall slides | |
WO2022012459A1 (en) | Parasite detection method and system based on artificial intelligence, and terminal device | |
JP2024045407A (en) | Methods and apparatus for detecting entity in bodily sample | |
Quinn et al. | Automated blood smear analysis for mobile malaria diagnosis | |
CN112380900A (en) | Deep learning-based cervical fluid-based cell digital image classification method and system | |
CN110796661B (en) | Fungal microscopic image segmentation detection method and system based on convolutional neural network | |
WO2012041333A1 (en) | Automated imaging, detection and grading of objects in cytological samples | |
Davidson et al. | Automated detection and staging of malaria parasites from cytological smears using convolutional neural networks | |
US20230393380A1 (en) | Method And System For Identifying Objects In A Blood Sample | |
Lovell et al. | International contest on pattern recognition techniques for indirect immunofluorescence images analysis | |
CN111797786B (en) | Detection method for in vitro biological sample, four-classification, computer device and computer-readable storage medium | |
Wang et al. | Application of deep learning in clinical settings for detecting and classifying malaria parasites in thin blood smears | |
CN117173460A (en) | Multitasking joint optimization method and device for detecting prostate sporadic cancer | |
OA21122A (en) | Parasite detection method and system based on artificial intelligence, and terminal device. | |
CN113178228B (en) | Cell analysis method based on nuclear DNA analysis, computer device, and storage medium | |
CN113723441B (en) | Intelligent analysis system and method for lip gland pathology | |
CN113313685B (en) | Renal tubular atrophy region identification method and system based on deep learning | |
Yu et al. | Automatic Detection and Identification of Trichomonas Vaginalis from Fluorescence Microscopy Images. | |
Krishnadas et al. | S., SK; Prabhu, S. Classification of Malaria Using Object Detection Models. Informatics 2022, 9, 76 | |
CN113295692B (en) | Cell analysis method based on cell nuclear DNA and TBS double analysis method, computer equipment and storage medium | |
CN116824579B (en) | Method and device for detecting yarrowia pneumocystis based on direct immunofluorescence staining | |
Hussain et al. | Automated detection of Kaposi sarcoma-associated herpesvirus infected cells in immunohistochemical images of skin biopsies | |
Bae et al. | Embedded-deep-learning-based sample-to-answer device for on-site malaria diagnosis | |
Kumar et al. | Human Malaria Detection using Random Forest Tree | |
Aulia et al. | A Novel Digitized Microscopic Images of ZN-Stained Sputum Smear and Its Classification Based on IUATLD Grades |