WO2019047366A1

WO2019047366A1 - Artificial intelligence-based image recognition system and method

Info

Publication number: WO2019047366A1
Application number: PCT/CN2017/110477
Authority: WO
Inventors: 姚育东; 钱唯; 郑斌; 马贺; 齐守良; 赵明芳
Original assignee: 深圳市前海安测信息技术有限公司
Priority date: 2017-09-11
Filing date: 2017-11-10
Publication date: 2019-03-14
Also published as: CN107506604A

Abstract

The present invention discloses an artificial intelligence-based image recognition system and method, which are applied to medical terminal devices, wherein the method comprises the following steps: acquiring from an image acquisition terminal an image signal of a person to be detected; processing the image signal of the person to be detected as a screening image; in an image database of a medical information platform, matching image data for which the degree of similarity with the screening image is within a preset range; receiving, from an input unit of the medical terminal device, an image recognition instruction; in an instruction recognition library of the medical information platform, matching text corresponding to the image recognition instruction as the image recognition result according to the image recognition instruction. By implementing the present invention, a doctor need only input a simple image recognition instruction to obtain a recognition result of a screening image, thereby simplifying the operation of screening image recognition and increasing the efficiency and accuracy of image recognition.

Description

Image recognition system and method based on artificial intelligence

[0001] The present invention relates to the field of medical image processing and recognition technologies, and in particular, to an image recognition system and method based on artificial intelligence.

Background technique

[0002] At present, the use of screening images for assisted diagnosis has become an important method for screening and diagnosing early local organs of the body. After viewing the screening image, the doctor cannot write the report simultaneously, but switch to the interface or program of the report writing; and, after writing the report, the content needs to be input to the computer word by word. For some of the more common test results (such as no abnormality in a certain part), it still needs to make cumbersome input, which requires considerable energy and time, which is prone to misuse and causes the efficiency of local organ screening to decrease. It is not suitable for the census of a large number of samples of body parts. In addition, due to the large number of images of local organs in the body, it is difficult for doctors to directly identify each part of the body and manually report the report. It is difficult to ensure efficiency and accuracy, which may lead to missed diagnosis and misdiagnosis.

technical problem

[0003] The main object of the present invention is to provide an image recognition system and method based on artificial intelligence, which can obtain a screening image recognition result by simply inputting a simple image recognition instruction, simplifying the operation of screening image recognition, and improving The efficiency and accuracy of image recognition.

Problem solution

Technical solution

[0004] In order to achieve the above object, the present invention provides an image recognition system based on artificial intelligence, which is applied to a medical terminal device, and the medical terminal device is connected to an image collection terminal and a medical information platform through a communication network, and the manual is based on artificial The intelligent image recognition system includes: an image acquisition module, configured to acquire an image signal of the object to be detected from the image capturing terminal, process the image signal of the object to be detected into a screening image, and receive an image recognition instruction from an input unit of the medical terminal device. ;

[0005] an image matching module for matching and screening image similarity in an image database of a medical information platform Image data within a preset range;

[0006] The image recognition module is configured to match, according to the image recognition instruction, a text corresponding to the image recognition instruction in the instruction recognition library of the medical information platform as the image recognition result.

[0007] Preferably, the image capturing terminal includes an infrared generator, an infrared receiver, an analog to digital converter, and a communication port, wherein: the infrared generator is configured to generate infrared light and inspect the infrared light to the body of the object to be detected. The infrared receiver is configured to collect an infrared light signal transmitted through a body part of the body to be detected into an analog electrical signal containing body tissue structure information of the body to be detected; the analog to digital converter is used to The analog electrical signal analog to the body tissue structure information of the body to be detected is converted into an image signal in the form of a digital signal; the communication port is configured to transmit an image signal of the person to be detected to the medical terminal device.

[0008] Preferably, the image recognition instruction is a first identification mark, a second identification mark or a third identification mark, wherein:

[0009] when the image recognition instruction is the first identification mark 吋, the image recognition module matches the corresponding text in the instruction recognition library according to the first identification mark as a normal image recognition result;

[0010] when the image recognition instruction is a second identification mark 吋, the image recognition module matches the corresponding character in the instruction recognition library as a reference image recognition result according to the second identification mark;

[0011] When the image recognition command is the third identification mark 吋, the image recognition module matches the corresponding character in the command recognition library according to the third identification mark as an abnormal image recognition result.

[0012] Preferably, the artificial intelligence-based image recognition system further includes a result indication module, configured to identify a normal image recognition result, an abnormal image recognition result, or a reference image recognition result in the image recognition report.

[0013] Preferably, when the image recognition result corresponding to the image recognition instruction is not matched in the instruction identification library, the image recognition module is further configured to receive the image recognition result input by the doctor and add the image recognition result to the image recognition report. And adding the input image recognition result to the instruction recognition library.

[0014] The present invention also provides an image recognition method based on artificial intelligence, which is applied to a medical terminal device, and the medical terminal device is connected to the image capturing terminal and the medical information platform through a communication network, and the method includes the following steps: from the image capturing terminal Obtaining an image signal of the person to be tested;

[0015] processing the image signal of the person to be detected as a screening image; [0016] receiving an image recognition instruction from an input unit of the medical terminal device;

[0017] matching, in the image database of the medical information platform, image data whose screening image similarity is within a preset range;

[0018] The text corresponding to the image recognition instruction is matched in the instruction recognition library of the medical information platform as the image recognition result according to the image recognition instruction.

[0019] Preferably, the step of acquiring an image signal of the object to be detected from the image capturing terminal comprises: generating infrared light by using an infrared generator of the image capturing terminal and seeing the infrared light to a body part of the body to be detected; The infrared receiver of the image capturing terminal collects the infrared light signal transmitted through the body part of the body to be detected into an analog electrical signal containing the tissue structure information of the body part of the body to be detected; the analog to digital converter using the image capturing terminal will contain The analog electrical signal analogy of the body part organ structure information of the subject to be detected is converted into an image signal in the form of a digital signal; the communication port of the image capturing terminal is used to transmit the image signal of the subject to be detected to the medical terminal device.

[0020] Preferably, the image recognition instruction is a first identification mark, a second identification mark or a third identification mark, and the matching the image corresponding to the image recognition instruction in the instruction recognition library of the medical information platform according to the image recognition instruction The steps as the result of image recognition include:

[0021] when the image recognition instruction is the first identification mark 吋, matching the corresponding character in the instruction recognition library according to the first identification mark as a normal image recognition result;

[0022] when the image recognition instruction is the second identification mark 吋, matching the corresponding character in the instruction recognition library as a reference image recognition result according to the second identification mark;

[0023] When the image recognition instruction is the third identification mark 吋, the corresponding character is matched in the instruction recognition library according to the third identification mark as an abnormal image recognition result.

[0024] Preferably, the artificial intelligence-based image recognition method further comprises the following steps: identifying a normal image recognition result, an abnormal image recognition result or a reference image recognition result in the image recognition report

[0025] Preferably, the artificial intelligence-based image recognition method further comprises the steps of: receiving an image recognition result input by the doctor when the image recognition result corresponding to the image recognition instruction is not matched in the instruction recognition library And added to the image recognition report, and the input image recognition result is added to the instruction recognition library. Advantageous effects of the invention

Beneficial effect

Compared with the prior art, the artificial intelligence-based image recognition system and method of the present invention can be used in a preset instruction recognition library according to a doctor's input image recognition instruction during a process of screening image recognition by a doctor. The image recognition result corresponding to the image recognition instruction is matched, and the image data with the similarity of the image similarity in the preset image is matched in the preset image database, and the image recognition report is generated according to the image recognition result and the image data. Since the doctor only needs to input a simple image recognition command, the recognition result of the screening image can be obtained without the doctor inputting the cumbersome examination conclusion, thereby simplifying the operation of the image recognition of the body part organ and improving the efficiency of image recognition of the body part organ. And accuracy, assist doctors to improve the efficiency and accuracy of body part disease detection and screening, and improve the social efficiency of body partal organ screening.

Brief description of the drawing

DRAWINGS

1 is a schematic diagram of an application environment of a preferred embodiment of an artificial intelligence-based image recognition system according to the present invention;

2 is a flow chart of a preferred embodiment of an artificial intelligence based image recognition method according to the present invention.

[0029] The objects, features, and advantages of the invention will be apparent from the accompanying drawings.

BEST MODE FOR CARRYING OUT THE INVENTION

The specific embodiments, structures, features and utilities of the present invention are described in detail below with reference to the drawings and the preferred embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0031] Referring to FIG. 1, FIG. 1 is a schematic diagram of an application environment of a preferred embodiment of an artificial intelligence-based image recognition system according to the present invention. In the embodiment, the artificial intelligence-based image recognition system 10 is installed and operated in the medical terminal device 1. The medical terminal device 1 establishes a communication connection with the medical information platform 2 and the image capturing terminal 4 via the communication network 3. The medical terminal device 1 is disposed at a body part organ examination center Or a doctor's workstation computer or server of a large hospital, such as a computing device with data processing and communication functions. The medical information platform 2 can be a server in a medical information system platform, and the medical information platform 2 includes an instruction identification library 21 and an image database 22. In this embodiment, the preset instruction recognition library 21 pre-collects medical terms, radiology terms, and common image recognition result templates commonly used for body part organ health screening, and in general, the template corresponds to Identify the results for normal images. The image database 22 stores various normal or abnormal image data that are commonly used for reference. After receiving the screening image, the image data in the screening image and the image database can be similarly matched. The communication network 3 may be an internet network including a local area network, a wide area network, or a wireless transmission network including GSM, GPRS, and CDMA.

[0032] The body part organ image collecting terminal 4 is disposed in a medical examination institution such as a community medical workstation, and establishes a network communication connection with the medical terminal device i. In the present embodiment, the body part organ image collecting terminal 4 includes an infrared generator 41, an infrared receiver 42, an analog to digital converter 43, and a communication port 44. The infrared generator 41 generates infrared light and sees the infrared light onto a body part organ of the body to be detected; the infrared receiver 42 collects infrared light signals transmitted through the body part of the body to be detected and processes the body part organ structure The analog electrical signal of the information; the infrared light generated by the infrared generator 41 is fluorinated to the body part of the body to be detected, and the infrared light signal received by the infrared receiver 42 carries the infrared transmitted light of the body tissue structure information of the body. The analog-to-digital converter 43 analog-to-digital converts the analog electrical signal containing the body tissue structure information of the body to be detected collected by the infrared receiver 42 into an image signal in the form of a digital signal; the communication port 44 is used for the person to be detected The information and the digital signal containing the body part organ image information of the subject to be detected are transmitted to the cloud server 1 through the communication network 3. The communication port 44 can be a wireless communication interface with remote wireless communication functions, such as a communication interface supporting GSM, GPRS, CDMA.

[0033] In this embodiment, the medical terminal device 1 includes, but is not limited to, an artificial intelligence based image recognition system 10, an input unit 11, a storage unit 12, a processing unit 13, a communication unit 14, and a display unit 15 . The input unit 11, the storage unit 12, the processing unit 13, the communication unit 14, and the display unit 15 are all connected to the processing unit 13 through a data bus, and can perform information interaction with the artificial intelligence-based image recognition system 10 through the processing unit 13. The input unit 11 can be a hardware device such as a keyboard, a touch screen or a mouse. The storage unit 12 is a read only memory unit ROM, an electrically erasable storage unit EEPROM or Flash memory unit FLASH and other memory. The processing unit 13 is a central processing unit (CPU), a microprocessor, a microcontroller (MCU), a data processing chip, or an information processing unit having a data processing function. The communication unit 14 can be a wireless communication interface with remote wireless communication functions, such as a communication interface supporting GSM, GPRS, CDMA. The display unit 15 is a display for displaying a body part organ inspection report of the person to be examined.

In the embodiment, the artificial intelligence-based image recognition system 10 includes, but is not limited to, an image acquisition module 101, an image matching module 102, an image recognition module 103, and a result indication module 104. The module referred to in the present invention refers to a series of computer program instruction segments that can be executed by the processing unit 13 of the medical terminal device 1 and that can perform a fixed function, which are stored in the storage unit 12 of the medical terminal device 1. .

The image acquisition module 101 is configured to acquire an image signal of the object to be detected from the image capturing terminal 4. In this embodiment, the infrared generator 41 generates infrared light and sees the infrared light on a body part of the body to be detected; the infrared receiver 42 collects an infrared light signal passing through a body part of the body to be detected. And processing the analog electrical signal as the body tissue structure information of the body; the analog-to-digital converter 43 converts the analog electrical signal collected by the infrared receiver 42 and contains the information of the body tissue structure of the body to be detected into a digital signal. The image signal of the form is sent to the image acquisition module 101 by the communication port 44.

[0036] The image acquisition module 101 is further configured to process the image signal of the object to be detected as a screening image. Specifically, the image acquisition module 101 records the image data of the image to be detected by the digital image processing software in the form of a digital file, and then generates a screening image of the object to be detected according to the image data. In the present embodiment, the principle of detecting the local body organ of the infrared body is: infrared light illuminates the local organ part of the human body, and the local organ tissue of the human body exhibits different absorption characteristics through the infrared spectrum passing through the body, so the infrared light transmitted through the lesion part The intensity of the infrared signal is different between the light signal and the tissue of the normal body. The acquired gray image of the infrared image, the tissue structure, the external dimensions, especially the optical properties of the body part and body tissues, can detect the body. The location and size of the lesion in the local organ site.

[0037] The image acquisition module 101 is further configured to receive an image recognition instruction from the input unit 11. In the present embodiment, after identifying the screening image, the doctor inputs from the input unit 11 to check the health of the body part of the body. Image recognition instructions.

[0038] The image recognition module 103 is configured to match the text corresponding to the image recognition instruction in the instruction recognition library 21 of the medical information platform 2 as the image recognition result according to the image recognition instruction. In this embodiment, the image recognition instruction may be a first identification mark, a second identification mark or a third identification mark. Specifically, the image recognition instruction includes a preset first identification mark, and the first identification mark is used as an indicator for determining whether the image recognition result is normal, that is, when the image recognition instruction includes the first identification mark 吋, The image recognition module 103 determines that the image recognition result of the screening image is normal, and the content of the first identification mark is generally selected as a word different from related medical terms and radiological terms. For example, when the image recognition instruction includes "click + body part organ", it indicates that the image recognition result of the screening image is normal, and therefore, the text corresponding to the normal health of the body part of the body can be matched in the instruction recognition library 21 as a normal The image recognition result, for example, matches "the body part of the organ is not abnormal" as a normal image recognition result. The image recognition instruction includes a preset second identification mark, and the second identification mark is used as an indicator for determining whether the image recognition result is a reference image recognition result, that is, when the image recognition instruction includes the second identification mark. The image recognition module 103 needs to match the image recognition result corresponding to the screening image, and the image recognition module 103 matches the image corresponding to the screening image in the instruction recognition library 21 as the reference image recognition result according to the second identification flag. The content of the second identification mark may be set as a "reference" or the like which can clearly determine the image recognition result whose reference image recognition result is a reference. For example, when the image recognition command is "reference", the character corresponding to the conclusion of the image recognition report is matched in the command recognition library, and the character is used as the reference image recognition result. If the image recognition instruction is the third identification mark, the image recognition module 103 may determine that the image recognition result is not a normal image recognition result, and match the corresponding text in the instruction recognition library according to the third identification mark. If the matching text corresponding to the third identification mark has nothing to do with the image recognition result of the screening image, it indicates that the received image recognition instruction is invalid; on the contrary, it indicates that the image recognition result corresponding to the image recognition instruction is an abnormal image recognition result. . For example, if the recognized image recognition command is "a number of soft tissue density lesions visible on both sides of the breast, and the morphology conforms to the body part organs", the corresponding text is matched in the command recognition library, and the text is used as an abnormal image recognition result.

In the embodiment, if the image recognition module 103 does not match the image recognition result corresponding to the image recognition instruction in the instruction recognition library 21, it indicates that the corresponding file is not stored in the instruction identification library 21, Thereafter, the doctor can input the corresponding image recognition result into the image recognition report through the input unit 11, and The image recognition module 103 adds the input image recognition result to the instruction recognition library 21, so as to match the corresponding image recognition result from the instruction recognition library 21 after receiving the same or similar image recognition instruction next time.

[0040] The image matching module 102 is configured to match the image data in the preset image database 22 with the image similarity within the preset range, and add the matched image data to the corresponding image recognition report. in. When the similarity is within the preset range, the similar image data in the image database 22 is acquired as the image data corresponding to the screening image, and the preset range may be customized according to the relevant parts of the breast, for example, set to be greater than 80%. After the image matching module 102 matches the image data corresponding to the screening image, the matched image data is added to the corresponding image recognition report. The image recognition report not only includes the image recognition result of the screening image, but also the image data of the screening image for reference by the doctor and the examinee.

[0041] The result indication module 104 is configured to identify a normal image recognition result, an abnormal image recognition result, or a reference image recognition result in the image recognition report, and display the image recognition report on the display unit 15 in a preset display manner. Normal image recognition results, abnormal image recognition results, or reference image recognition results. Specifically, the result indication module 104 performs different identification on the normal image recognition result and the abnormal image recognition result according to whether the image recognition result is normal or abnormal, and the reference result can be identified as being distinguishable from the normal or abnormal image recognition result. Other identifiers, such as marking different image recognition results as different fonts or different colors, so as to display in a preset display manner after display, for example, the text content of the abnormal image recognition result is in bold form Display or highlight. In this embodiment, if the image recognition result in the image recognition report is a normal image recognition result, the text content corresponding to the recognition template may be identified as blue after matching the recognition template indicating that the image recognition result is normal. If the image recognition result in the image recognition report is an abnormal image recognition result, the text content may be identified as a red font after being matched with the text corresponding to the abnormal image recognition result, and displayed. The text content of the red font is displayed in bold or highlighted to remind the examinee to pay attention. After the image recognition report is generated, the normal image recognition result or the abnormal image recognition result in the image recognition report is differently identified, and the normal image recognition result or the abnormal image recognition result is displayed in a preset display manner, so that The image recognition report is more clear and easy for doctors and subjects to read.

[0042] The present invention also provides an image recognition method based on artificial intelligence, which is applied to a medical terminal device. As shown in FIG. 2, FIG. 2 is a flow chart of a preferred embodiment of the image recognition method based on artificial intelligence according to the present invention. In the embodiment, the image recognition method based on artificial intelligence includes the following steps: [0043] Step S21, the image acquisition module 101 acquires an image signal of the object to be detected from the image acquisition terminal 4, and is to be detected. The image signal processing is to screen the image; in the embodiment, the infrared generator 41 generates infrared light and sees the infrared light on the body part of the body to be detected; the infrared receiver 42 collects the body through the person to be tested The infrared light signal of the local organ is processed as an analog electrical signal of body tissue structure information of the body; the analog-to-digital converter 43 converts the analog electrical signal of the local organ structure information of the body of the subject to be detected by the infrared receiver 42 Processing is an image signal in the form of a digital signal; the communication port 44 transmits the image signal of the person to be detected to the image acquisition module 101. The image acquisition module 101 records the image data of the image to be detected by the digital image processing software in the form of a digital file, and then generates a screening image of the object to be detected according to the image data.

[0044] Step S22: The image matching module 102 matches the image data in the preset image database with the image similarity within the preset range, and adds the matched image data to the corresponding image recognition report. In this embodiment, the image database 22 stores various commonly used normal or abnormal image data for reference. After receiving the screening image, the image matching module 102 can perform screening on the image and image database. The image data is similarly matched. When the similarity is within the preset range, the similar image data in the image database is acquired as the image data corresponding to the screening image, and the preset range may be customized according to the breast related portion, for example, set to be greater than 80%. After matching the image data corresponding to the screening image, the image matching module 102 adds the image data according to the image data to the corresponding image recognition report. The image recognition report includes not only the image recognition result of the screening image but also the image data of the screening image for reference by the doctor and the examinee.

[0045] Step S23, the image acquisition module 101 receives an image recognition instruction from the input unit 11 of the medical terminal device.

In the present embodiment, after the screening image is recognized, the doctor inputs an image recognition command for checking the health condition of the body part from the input unit 11.

[0046] Step S24, the image recognition module 103 matches the character corresponding to the image recognition instruction in the command recognition library 21 as the image recognition result according to the image recognition instruction. In this embodiment, the image recognition instruction may be a first identification mark, a second identification mark or a third identification mark; the image recognition result may be a normal image recognition result, a reference image recognition result, or an abnormal image. Identify the results. Specifically, such as The image recognition instruction includes a preset first identification mark, and the first identification mark is used as an indicator for determining whether the image recognition result is normal, that is, the first identification mark is included in the image recognition instruction.

The image recognition module 103 determines that the image recognition result of the screening image is normal, and the content of the first identification mark is generally selected as a word different from related medical terms and radiological terms. For example, when the image recognition instruction includes "click + body part organ", it indicates that the image recognition result of the screening image is normal, and thus, the instruction recognition library 21 can match the text corresponding to the health of the body part organ, for example, matching. "There is no abnormality in the local organs of the body" as a result of normal image recognition. The image recognition instruction includes a preset second identification mark, and the second identification mark is used as an indicator for determining whether the image recognition result is a reference image recognition result, that is, when the image recognition instruction includes the second identification mark. The image recognition module 103 needs to match the image recognition result corresponding to the screening image, and the image recognition module 103 matches the image corresponding to the screening image in the instruction recognition library 21 as the reference image recognition result according to the second identification flag. The content of the second identification mark can be set to a word such as "reference" which can clearly determine the image recognition result whose indication is the reference image recognition result. For example, when the image recognition command is "reference", the character corresponding to the conclusion of the image recognition report is matched in the command recognition library 21, and the character is used as a reference image recognition result. If the image recognition instruction is the third identification mark, the image recognition module 103 may determine that the image recognition result is not a normal image recognition result, and match the corresponding text in the instruction recognition library 21 according to the third identification mark. In this case, if the matching text corresponding to the third identification mark has nothing to do with the image recognition result of the screening image, it indicates that the received image recognition instruction is invalid; on the contrary, it indicates that the image recognition result corresponding to the image recognition instruction is abnormal. Image recognition results. For example, if the recognized image recognition command is "a number of soft tissue densities visible on both sides of the breast, and the morphology is consistent with the body part organs", the corresponding recognition text is matched in the instruction recognition library 21, and the text is used as an abnormal image recognition result.

In the embodiment, if the image recognition module 103 does not match the image recognition result corresponding to the image recognition instruction in the instruction recognition library 21, it indicates that the corresponding file is not stored in the instruction recognition library 21, and The doctor can input the corresponding image recognition result into the image recognition report through the input unit 11, and the image recognition module 103 adds the input image recognition result to the instruction recognition library 21, so as to receive the same or similar image recognition next time. The command 匹配, the corresponding image recognition result is matched from the command recognition library 21.

[0048] Step S25, the result indication module 104 identifies the normal image recognition result, the abnormal image recognition result or the reference image recognition result in the image recognition report, and displays the preset display mode on the display unit 15 A normal image recognition result, an abnormal image recognition result, or a reference image recognition result in the image recognition report is displayed. Specifically, the result indication module 104 performs different identification on the normal image recognition result and the abnormal image recognition result according to whether the image recognition result is normal or abnormal, and the reference result can be identified as being distinguishable from the normal or abnormal image recognition result. Other identifiers, such as marking different image recognition results as different fonts or different colors, so that after display, the display unit 15 displays the display in a preset manner, for example, the text content of the abnormal image recognition result is Displayed in bold or highlighted.

[0049] In this embodiment, if the image recognition result in the image recognition report is a normal image recognition result, the text content identifier corresponding to the recognition template may be matched after matching the recognition template representing that the image recognition result is normal. It is a blue font and is displayed normally; if the image recognition result in the image recognition report is an abnormal image recognition result, the text content may be identified as a red font after matching the text corresponding to the abnormal image recognition result, and In the display, the text content of the red font is displayed in bold or highlighted to remind the examinee to pay attention. After the image recognition report is generated, the normal image recognition result or the abnormal image recognition result in the image recognition report is differently identified, and the normal image recognition result or the abnormal image recognition result is displayed in a preset display manner, so that The image recognition report is more clear and easy for doctors and examiners to read.

[0050] The body part organ screening intelligent identification system and method of the present invention can match and image in a preset instruction recognition library according to a doctor inputting a simple image recognition instruction during a process of screening image recognition by a doctor. Identifying the image recognition result corresponding to the instruction, matching the image data in the preset image database with the image similarity within the preset range, and generating the image recognition report according to the image recognition result and the image data, since the doctor only needs to input the simple The image recognition command can obtain the recognition result of the screening image without the doctor inputting the cumbersome examination conclusion, thereby simplifying the operation of the image recognition of the body part organ, improving the efficiency and accuracy of the image recognition of the body part organ, thereby assisting Doctors improve the efficiency and accuracy of detection and screening of body parts and diseases, and improve the social efficiency of body partal organ screening.

The above are only the preferred embodiments of the present invention, and are not intended to limit the scope of the present invention, and the equivalent structure or equivalent function transformations made by the description of the present invention and the contents of the drawings may be directly or indirectly applied to other related aspects. The technical field is equally included in the scope of patent protection of the present invention. Industrial applicability

Compared with the prior art, the artificial intelligence-based image recognition system and method of the present invention can match and match the image recognition instruction input by the doctor in the preset instruction recognition library during the process of screening image recognition by the doctor. The image recognition result corresponding to the image recognition instruction matches the image data in the preset image database with the image similarity within the preset range, and generates the image recognition report according to the image recognition result and the image data. Since the doctor only needs to input a simple image recognition command, the recognition result of the screening image can be obtained without the doctor inputting the cumbersome examination conclusion, thereby simplifying the operation of the image recognition of the body part organ and improving the efficiency of image recognition of the body part organ. And accuracy, assist doctors to improve the efficiency and accuracy of body part disease detection and screening, and improve the social efficiency of body partal organ screening.

Claims

Claim

[Claim 1] An artificial intelligence-based image recognition system is applied to a medical terminal device, and the medical terminal device is connected to an image acquisition terminal and a medical information platform through a communication network, wherein the artificial intelligence-based image is The identification system includes: an image acquisition module, configured to acquire an image signal of the object to be detected from the image capturing terminal, process the image signal of the object to be detected as a screening image, and receive an image recognition instruction from an input unit of the medical terminal device; a module, configured to match and image image data whose image similarity is within a preset range in an image database of the medical information platform; and an image recognition module, configured to match and image in the instruction recognition library of the medical information platform according to the image recognition instruction The text corresponding to the recognition command is used as the image recognition result.

2. The artificial intelligence-based image recognition system according to claim 1, wherein the image collection terminal comprises an infrared generator, an infrared receiver, an analog to digital converter, and a communication port, wherein: The infrared generator is configured to generate infrared light and fluoresce the infrared light to a body part organ of the body to be detected; the infrared receiver is configured to collect infrared light signals transmitted through the body part of the body to be detected to be processed to include the body of the person to be detected An analog electrical signal of local organ tissue structure information; the analog-to-digital converter is configured to convert an analog electrical signal containing the body tissue structure information of the body to be detected into an image signal in the form of a digital signal; the communication port is used for Sending an image signal of the person to be detected to the medical terminal device.

The AI-based image recognition system according to claim 1, wherein the image recognition instruction is a first identification mark, a second identification mark or a third identification mark, wherein: The image recognition command is a first identification mark 吋, and the image recognition module matches the corresponding text in the instruction recognition library as a normal image recognition result according to the first identification mark; when the image recognition instruction is the second identification mark, The image recognition module matches the corresponding text in the instruction recognition library as a reference image recognition result according to the second identification mark; when the image recognition instruction is the third identification mark 吋, the image recognition module is instructed according to the third identification mark The matching text in the recognition library is used as an abnormal image recognition result.

[Claim 4] The artificial intelligence-based image recognition system according to claim 3, wherein the system further comprises a result indication module, configured to perform normal image recognition results and abnormal image recognition in the image recognition report. The result is identified by reference to the image recognition result.

[Claim 5] The artificial intelligence-based image recognition system according to claim 1, wherein the image recognition result corresponding to the image recognition instruction is not matched in the instruction recognition library, the image The identification module is further configured to receive the image recognition result input by the doctor and add the image recognition result to the image recognition report, and add the input image recognition result to the instruction recognition library.

[Claim 6] An image recognition method based on artificial intelligence, which is applied to a medical terminal device, wherein the medical terminal device is connected to an image capturing terminal and a medical information platform through a communication network, wherein the method comprises the steps of: The acquisition terminal acquires the image signal of the object to be detected; processes the image signal of the object to be detected into a screening image; and matches the image data in the image database of the medical information platform with the similarity of the image in the preset range; The input unit of the device receives the image recognition instruction; and matches the text corresponding to the image recognition instruction in the instruction recognition library of the medical information platform as the image recognition result according to the image recognition instruction.

The method for recognizing an image based on artificial intelligence according to claim 6, wherein the step of acquiring an image signal of the object to be detected from the image capturing terminal comprises: generating by using an infrared generator of the image capturing terminal Infrared light and infrared light is incident on the body part of the body to be detected; infrared light signal collected through the body part of the body of the subject to be detected by the infrared receiver of the image capturing terminal is processed into a body part organ tissue containing the body to be detected Analog electrical signal of structural information; using an analog-to-digital converter of the image capturing terminal to convert an analog electrical signal modulus containing the body tissue structure information of the body to be detected into an image signal in the form of a digital signal; a communication port through the image capturing terminal And configured to send an image signal of the person to be detected to the medical terminal device.

The method for recognizing an image based on artificial intelligence according to claim 6, wherein the image recognition command is a first identification mark, a second identification mark or a third identification mark, and the image recognition is performed according to the image. The step of the instruction matching the text corresponding to the image recognition instruction in the instruction recognition library of the medical information platform as the image recognition result includes: when the image recognition instruction a first identification mark 吋, matching the corresponding text in the instruction recognition library according to the first identification flag block as a normal image recognition result; when the image recognition instruction is the second identification mark 吋, identifying the instruction according to the second identification mark The image recognition result is matched with the corresponding text as the reference image recognition result; when the image recognition instruction is the third identification mark 吋, the corresponding recognition text is matched in the instruction recognition library according to the third identification mark as an abnormal image recognition result.

The method for recognizing an artificial intelligence based image according to claim 8, further comprising the steps of: normal image recognition result in the image recognition report, abnormal image recognition result or reference The image recognition result is identified.

The method of claim 6, wherein the method further comprises the following steps:: the image corresponding to the image recognition instruction is not matched in the instruction recognition library; After the recognition result, the image recognition result input by the doctor is received and added to the image recognition report, and the input image recognition result is added to the instruction recognition library.