WO2020118534A1 - 一种b超机器人自动检测方法及系统 - Google Patents

一种b超机器人自动检测方法及系统 Download PDF

Info

Publication number
WO2020118534A1
WO2020118534A1 PCT/CN2018/120395 CN2018120395W WO2020118534A1 WO 2020118534 A1 WO2020118534 A1 WO 2020118534A1 CN 2018120395 W CN2018120395 W CN 2018120395W WO 2020118534 A1 WO2020118534 A1 WO 2020118534A1
Authority
WO
WIPO (PCT)
Prior art keywords
detection
similarity
human body
image data
organ
Prior art date
Application number
PCT/CN2018/120395
Other languages
English (en)
French (fr)
Inventor
林江峰
Original Assignee
广东医动科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 广东医动科技有限公司 filed Critical 广东医动科技有限公司
Priority to PCT/CN2018/120395 priority Critical patent/WO2020118534A1/zh
Publication of WO2020118534A1 publication Critical patent/WO2020118534A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B8/00—Diagnosis using ultrasonic, sonic or infrasonic waves
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object

Definitions

  • the present invention relates to the field of medical equipment, and in particular, to a B-ultrasonic robot automatic detection method and system.
  • B-ultrasound detection is the most direct and accurate detection method for monitoring human organs and blood flow information.
  • Existing B-ultrasound equipment requires professional technicians with professional medical knowledge and detection skills to operate the human body. According to the detection, it is a comprehensive judgment of the placement position, angle, and displayed image of the detection head to determine the detection image to be acquired as the detection conclusion.
  • the purpose of the present invention is to provide a B-ultrasonic robot automatic detection method and system, which aims to solve the technical problems that the existing B-ultrasound can only be operated by professional technicians, thus most of the basic medical institutions cannot perform corresponding detection services.
  • the method for automatically detecting a B-ultrasonic robot disclosed by the present invention includes the following steps:
  • Step 1 Obtain the human body image, compare the human body image with the pre-set stored human body standard database, and determine the location and area range of the corresponding human body organs;
  • Step 2 According to the determined position and area of the corresponding organ of the human body, the corresponding coordinate data of the human organ is formed;
  • Step 3 Provide a detection drive device connected with a detection head, and the detection drive device makes the first movement according to the coordinate data of the human organ until it moves to the corresponding organ position;
  • Step 4 The detection head detects the corresponding organ to form a first detection data stream, compares the first detection data stream with a pre-stored medical imaging standard database, and feeds back the similarity result of the comparison;
  • Step 5 The detection driving device continues to move to make a second movement, the detection head detects the corresponding organ to obtain a second detection data stream, and compares the second detection data stream with a pre-stored medical imaging standard database If yes, and feedback the similarity result of the comparison, go to step 6;
  • Step 6 Judge the size of the similarity result of this detection comparison and the similarity result of the previous detection comparison
  • the detection driving device When the similarity result of the current detection comparison is less than the similarity result of the previous detection comparison, the detection driving device returns to the initial position of the current movement, and performs step 5 again;
  • step 7 is performed
  • Step 7 Determine whether the similarity result of the current test comparison meets the corresponding standard determined by the medical test; if it is satisfied, the test head obtains the human organ test data stream of the test as the final test result; if If it is not satisfied, jump to step 5 to continue.
  • the human body standard database includes a plurality of human body image models that mark the positions of corresponding organs according to different body shapes; each human body image model corresponds to a human body shape, and each human body image model is marked with human organs and The location of the blood vessel.
  • step 2 includes:
  • Step 21 Determine the origin of human body coordinates
  • Step 22 Determine the coordinate position and range of the human organ relative to the coordinate origin according to the position and area of the corresponding organ of the human body;
  • Step 23 Set the coordinate positions and ranges as the coordinate data of the human organs.
  • the medical imaging standard database includes a plurality of effective image data, each effective image data is segmented from the detection data stream, and the image pictures that best describe the human body parts, organs, and types and names of diseases are selected;
  • Each effective image data is also provided with a mark and an attribute, the mark includes at least one of a plurality of mark points, a plurality of mark areas, or a plurality of mark parameters, and the attribute is a corresponding body part, organ, and corresponding disease Type and name;
  • the medical image standard database also stores similarity calculation parameters and similarity calculation methods corresponding to each effective image data.
  • step 4 includes:
  • Step 41 Split the first detection data stream into multiple image frames
  • Step 42 Compare all the image frames with the effective image data in the medical imaging standard database
  • Step 43 Calculate the similarity between the marked points, marked areas or multiple marking parameters in each effective image data and the image frame according to the preset similarity calculation parameters and similarity calculation method, and determine the most similar to the effective image data The similarity between the close image frames and the effective image data, and the similarity is similarity result.
  • step 5 comprises:
  • Step 51 Divide the second detection data stream into multiple image frames
  • Step 52 Compare all the image frames with the effective image data in the medical imaging standard database
  • Step 53 Calculate the similarity between the marked points, marked areas or multiple marked parameters in each effective image data and the image frame according to the preset similarity calculation parameters and similarity calculation method, and determine the most similar to the effective image data The similarity between the close image frame and the effective image data, and use the similarity as the similarity result.
  • the second movement has multiple times, and each time the detection driving device moves, the detection head performs a detection on the corresponding organ to form a second detection data stream and obtain a similarity result.
  • the invention also provides a B-ultrasonic robot automatic detection system, including:
  • An image detection device the image detection device is used to detect the human body to form a human body image
  • a human body standard database which includes a plurality of human body image models that mark the positions of corresponding organs according to different body shapes; each human body image model corresponds to a human body shape, and each human body image model is marked with a human organ location;
  • An operation module the operation module is used to calculate the coordinates of the human organs to form the coordinates of the human organs;
  • a detection head which is used to detect human organs and form a detection data stream
  • a detection driving device for driving the detection head to move drives the detection head to the initial position of the corresponding organ for detection according to the coordinates of the human organ;
  • control judgment module is used to control the subsequent movement of the detection driving device, and control the detection head to perform detection after each movement to form a detection data stream;
  • the control judging module compares the detection data stream formed by the two previous tests with the pre-stored medical image standard database, and forms two similarity results to determine the size of the two similarity results;
  • the control judgment module sends a control signal to the detection driving device to control the detection driving device to return to the initial position of the movement;
  • the detection drive device is effective in detecting motion this time, and the control judgment module sends a control signal to the detection drive device to control the Check that the drive device continues to move;
  • the control judgment module controls the detection drive device to move multiple times, and after multiple detections by the detection head, until the similarity result of a certain detection comparison meets the corresponding standard that the medical judgment is the same;
  • the detection head obtains the human organ detection data stream of a certain detection whose similarity is determined to be the same as the final detection result.
  • the medical imaging standard database includes a plurality of effective image data, each effective image data is segmented from the detection data stream, and the image pictures that best describe the human body parts, organs, and types and names of diseases are selected;
  • Each effective image data is also provided with a mark and an attribute, the mark includes at least one of a plurality of mark points, a plurality of mark areas, or a plurality of mark parameters, and the attribute is a corresponding body part, organ, and corresponding disease Type and name;
  • the medical image standard database also stores similarity calculation parameters and similarity calculation methods corresponding to each effective image data.
  • control and judgment module divides the detection data stream into a plurality of image frames; compares all image frames with valid image data in a medical imaging standard database, and calculates parameters and similarity calculations according to preset similarity calculation parameters The method calculates the similarity between the marked points, marked areas or multiple marking parameters in each effective image data and the image frame, determines the image frame closest to the effective image data and the similarity of the effective image data, and compares the similarity Degree as the similarity result.
  • the B-robot automatic detection method disclosed in the present invention first obtains the human body image, compares the human body image with the human body standard database, and then determines the location and area of each human body organ according to the location of each human body organ, and converts it to include
  • the detection drive device drives the detection head to the corresponding human organ according to the coordinates of the human organ, and starts to detect the human organ, each detection gets a detection data stream, and at the same time
  • the detection driving device moves once and performs the next detection to obtain another detection data stream; the detection data streams detected twice before and after are divided into multiple image frames respectively.
  • the human body image by comparing the human body image with the human body standard database, it is possible to determine the position and area of each organ of the human body with different body characteristics, which provides conditions for the initial positioning of the organ position by the later detection drive device; The judgment of the similarity result of the two tests, so as to intelligently judge whether the movement position of the detection drive device is valid, that is, whether such movement is developing in the same direction as the completion of the calibration, that is, the detection data obtained after this detection.
  • the image frames in the stream are more and more similar to the effective image data, so that after multiple detections, the similarity requirement that meets the medical definition is the same, so as to complete the automatic detection; among them, the medical definition can be set to the same similarity by setting Threshold value, when the similar threshold value is reached, the detected test data stream is automatically obtained directly as the test result.
  • the similarity in the marked point, marked area, and marked parameter reaches a certain similarity threshold, it is deemed to be the same .
  • 90% is the set similarity threshold.
  • different organs and parts can be set with different similarity thresholds.
  • the detection method can enable the detection drive device to autonomously move and fine-tune, further completely autonomously and intelligently determine the location and organ to be detected, and autonomously adjust the detection angle and depth data, and can Identify and compare the detected test data streams to determine whether the test data streams conform to the preset medical imaging standard database, so as to automatically obtain the test data streams as the final test result.
  • the entire detection process of the detection method does not require human intervention, and only requires a preset human body standard database and medical imaging standard database to complete the corresponding B-ultrasound detection.
  • the human body standard database and the medical imaging standard database are obtained by manually labeling human body images and multiple effective image data respectively. According to the increase in the amount of data of the detected unit individual, it can be continuously marked and expanded, so that the The human body standard database and the medical imaging standard database include and refine more human body shapes, medical diseases, and B-ultrasound images.
  • the B ultrasonic robot automatic detection system of the present invention adopts the corresponding automatic detection method of the present invention.
  • the B ultrasonic robot automatic detection system can intelligently complete the B ultrasonic detection automatically without human intervention.
  • FIG. 1 and 2 are schematic flowcharts of the automatic detection method of the B-ultrasonic robot of the present invention
  • FIG. 6 is a schematic diagram of a module that does not invent the B-ultrasonic robot automatic detection system.
  • the present invention provides a B-ultrasonic robot automatic detection method, the detection method includes:
  • Step 1 Obtain the human body image, compare the human body image with the pre-set stored human body standard database, and determine the location and area range of the corresponding organs of the human body.
  • the human body image is obtained by an image detection device, and multiple human body images at different angles can be obtained by the image detection device. According to the human body image, the tall, short, thin and thin body shape can further confirm the correspondence of each organ in different human bodies position.
  • the human body image is compared with the human body standard database.
  • the human body standard database includes a plurality of human body image models that manually mark the positions of the corresponding organs according to different body types.
  • the human body image model is a kind of data
  • the format, for example, is also an image.
  • each human body image model corresponds to a human body type, and each human body image model is marked with the location of human organs and blood vessels.
  • the human body image model in the human body standard database is formed by manually marking the corresponding organs and parts on the human body image.
  • the human body image models in a human body standard database are all artificially marked manually.
  • the artificial intelligence learns a human body standard database that is expanded after learning a plurality of manually labeled human image models.
  • Step 2 According to the determined position and area range of the corresponding organ of the human body, corresponding coordinate data of the human organ are formed.
  • this step includes:
  • Step 21 Determine the origin of the coordinates of the human body; wherein, the more characteristic positions of the human body can be used as the origin of the coordinates, such as the navel eye, or the midpoint of the line connecting the two eyes.
  • Step 22 Determine the coordinate position and range of the human organ relative to the coordinate origin according to the position and area of the corresponding organ of the human body;
  • the position and area range of the corresponding organ of the human body are transferred to the coordinate system where the coordinate origin is located by coordinate conversion, that is, the coordinate position and range of the human organ relative to the coordinate origin are determined.
  • Step 23 Set the coordinate positions and ranges as the coordinate data of the human organs.
  • the coordinate range of multiple organs and the center point coordinates of the organs can be formed according to different organs, and these coordinates are used as a set to form the coordinate data of the human organs corresponding to the body shape.
  • Step 3 Provide a detection drive device connected with a detection head, and the detection drive device makes the first movement based on the coordinate data of the human organs until it moves to the corresponding organ position.
  • the detection driving device extracts the desired detection based on the determined coordinate data of the human organ The coordinate position of the human organs in the coordinate system, so as to quickly move to the corresponding position according to the coordinate position, ready for the detection of the detection head.
  • the movement of the detection drive device moves to the corresponding approximate position of the organ at a faster rate, focusing on the speed of movement, and the moved position is only the approximate position of the human organ to be detected, and cannot achieve the accuracy of medical detection According to the performance requirements, in the subsequent steps, the detection drive device will drive the detection head to move accurately, and then perform the detection.
  • Step 4 The detection head detects corresponding organs to form a first detection data stream, compares the first detection data stream with a pre-stored medical imaging standard database, and feeds back the similarity result of the comparison.
  • the data used as the comparison standard is pre-stored, which includes multiple effective image data, and each effective image data is segmented from the detection data stream After that, select the image that best describes the parts and organs of the human body and the type and name of the disease. Marks and attributes are also set on each effective image data.
  • the marks include multiple mark points, multiple mark areas, or multiple marks At least one of the parameters, the attribute is the corresponding body part, organ and corresponding disease type and name; the medical imaging standard database also stores the similarity calculation parameters and similarity corresponding to each effective image data Calculation.
  • the first detection data stream is divided into multiple image frames, and each image frame is an image picture.
  • Each image contains B-mode information such as the corresponding position and structure of the human structure or organ.
  • the marked points mark special points on each organ. There are multiple such marked points, and the combination of multiple marked points can uniquely represent a corresponding organ; where the marked area can be multiple marked The area enclosed by the points, the size of a marked area, the shape of the edge change, etc. also uniquely represent a corresponding organ; the marking parameters are the range of the density, flow rate and other parameters of the fluid, blood flow or medium in the organ. Through these marking parameters It can uniquely identify a corresponding organ represented.
  • the human body parts, organs and corresponding disease types and names set for each effective image data can be used to classify effective image data.
  • the type of detection can be directly judged by the attribute represented by the effective image data.
  • the effective image data can also be classified and saved, and multiple types of effective image data form a medical imaging standard database.
  • classification preservation and recognition the speed of subsequent comparison of the image frame with the effective image data can be effectively improved.
  • the image frame only needs to be compared with the effective image data marked with the corresponding organ, corresponding disease, or corresponding human body part. Instead of comparing all the effective image data in the entire medical imaging standard database, the efficiency of the comparison is improved.
  • the similarity calculation parameters and the similarity calculation method are different for different organs and different parts.
  • Such calculation methods and calculation parameters can be pre-set manually.
  • the similarity calculation method and similarity calculation parameters of the kidney are artificially set.
  • it is calculated by the similarity calculation method and the similarity calculation parameter.
  • Different organs, body parts, and diseases may have different similarity calculation methods and similarity calculation parameters.
  • the effective image data in the medical imaging standard database is also formed by artificially marking the detection images intercepted in the detection data stream, filling in attributes, setting similarity calculation parameters and similarity calculation methods.
  • the medical imaging standard database can also be obtained after the artificial intelligence learns a lot of marked effective graphic data after learning.
  • the marked points, marked areas, marked parameters, similarity calculation parameters, and similarity calculation methods of an organ can be learned through artificial intelligence.
  • the human body image model and medical standard database can cover all the human body data and the B-ultrasound data corresponding to the human body, which can fully realize the automatic detection.
  • this step includes:
  • Step 41 Split the first detection data stream into multiple image frames
  • Step 42 Compare all the image frames with the effective image data in the medical imaging standard database
  • Step 43 Calculate the similarity between the marked points, marked areas or multiple marking parameters in each effective image data and the image frame according to the preset similarity calculation parameters and similarity calculation method, and determine the most similar to the effective image data The similarity between the close image frame and the effective image data, and use the similarity as the similarity result.
  • step 5 includes: the detection drive device continues to move to make a second movement, the detection head detects the corresponding organ to obtain a second detection data stream, and the second detection data stream is The stored medical image standard database is compared, and the similarity result of the comparison is fed back, and step 6 is executed.
  • Step 51 Divide the second detection data stream into multiple image frames
  • Step 52 Compare all the image frames with the effective image data in the medical imaging standard database
  • Step 53 Calculate the similarity between the marked points, marked areas or multiple marked parameters in each effective image data and the image frame according to the preset similarity calculation parameters and similarity calculation method, and determine the most similar to the effective image data The similarity between the close image frame and the effective image data, and use the similarity as the similarity result.
  • the detection drive device when the detection drive device moves to the corresponding organ position through the first movement, it only means that the detection head has found the corresponding organ, and cannot meet the accuracy of other parameters such as the angle, depth, position, etc. of the B-ultrasound detection. Claim.
  • the detection driving device continues to move to make a second movement.
  • the second movement is a precision movement, and the angle and depth can be adjusted according to higher accuracy to ensure the detection accuracy of the detection head.
  • Step 6 Judge the size of the similarity result of this detection comparison and the similarity result of the previous detection comparison
  • the detection driving device When the similarity result of the current detection comparison is less than the similarity result of the previous detection comparison, the detection driving device returns to the initial position of the current movement, and performs step 5 again;
  • step 7 is executed.
  • B-ultrasound detection requires multiple adjustments to the position, angle, and depth of the detection head. After a certain adjustment, the similarity result of the detection comparison is less than the similarity result of the previous detection comparison. : The moving position of the detection head this time is in the opposite direction to the completion of the detection, that is to say, the image frames in the detection data stream obtained after this detection are more and more dissimilar to the effective image data. If, according to this The results have continued to develop, and it will inevitably never be able to meet the medical requirements defined as the same similarity requirements, and automatic detection cannot be completed. At this time, the detection head needs to be restored to the original position, and the angle, depth, and position of the movement must be readjusted.
  • the similarity result of the detection comparison is not less than the similarity result of the previous detection comparison, which means: the movement position of the detection head is developed in the same direction as the completion of the calibration, that is to say, The image frames in the detection data stream obtained after this detection are more and more similar to the effective image data. If they continue to develop according to such results, they will always obtain the same similarity requirements that meet the medical definition, so as to complete the automatic Detection, at this time the detection head adjusts the angle, depth and position of the movement is effective.
  • Step 7 Determine whether the similarity result of the current test comparison meets the corresponding criteria determined by the medical test to be the same; if it is satisfied, the test head obtains the human organ test data stream of the current test as the final test result;
  • step 5 If it is not satisfied, jump to step 5 to continue.
  • the second movement has multiple times, and each time the detection driving device moves, the detection head performs a detection on the corresponding organ to form a second detection data stream and obtain a similarity result .
  • the similarity results of the detection comparison will continue to increase further.
  • the further improvement of the similarity result until the similarity result of this test comparison meets the corresponding criteria determined by the medical judgment as the same, it is considered that the test is completed, and the test head of the test is obtained at this time
  • the human organ detection data stream is used as the final detection result.
  • the detection head detects the human organs and compares with a valid image data in the medical standard database, according to the medical standard, the comparison is successful, then the detection data stream of the detection is automatically used as the detection result .
  • the same comparison according to the medical standard means that in the definition of medicine, when the image frame in the detected image data stream is compared with the effective image data, the similarity in the marked point, marked area, and marked parameter reaches a certain Similar thresholds are considered to be the same. For example, when comparing an image frame with valid image data, when the number of overlapping marker points accounts for 90% of the total number of marker points, it is considered to be the same, and 90% is the set similarity threshold. In this embodiment, different organs and parts can be set with different similarity thresholds.
  • the detection driving device will drive the detection head to move to the corresponding organ and part to be detected
  • the test will start, and the test drive will fine-tune the detection angle, depth, and position of the test head in real time until the B-ultrasound data obtained after a certain adjustment, that is, the test data stream and the pre-stored or learned medical imaging standard database After a certain effective image data in the comparison matches the same standard of medical judgment, this test will end, and the detection head will automatically obtain the adjusted test data stream as the result of this test.
  • the entire testing process can be completed without intervention and operation.
  • a human body standard database and a medical image standard database can be connected with multiple detection drive devices and detection heads, which can perform sufficient detection services and obtain sufficient detection results.
  • multiple detection heads and detection drive devices connected simultaneously through decentralization can provide enough detection results as learning samples for artificial intelligence learning, which can greatly Improve the efficiency of artificial intelligence learning and the reliability of the human body standard database and medical imaging standard database.
  • the present invention also provides a B-ultrasonic robot automatic detection system.
  • the B-ultrasonic robot automatic detection system includes:
  • An image detection device 100 which is used to detect the human body and form a human body image
  • a human body standard database 200 which includes a plurality of human body image models that mark the positions of corresponding organs according to different body shapes; each human body image model corresponds to a human body shape, and each human body image model is marked with The location of human organs;
  • An operation module 300 the operation module 300 is used to calculate the coordinates of the human organs to form the coordinates of the human organs;
  • a detection head 400 which is used to detect human organs and form a detection data stream
  • a detection drive device 500 which is used to drive the detection head 400 to move; the detection drive device 500 drives the detection head 400 to the initial position of the corresponding organ for detection according to the coordinates of the human organ;
  • a control judgment module 600 which is used to control the subsequent movement of the detection driving device 500, and controls the detection head 400 to perform detection after each movement to form a detection data stream;
  • the control and judgment module 600 compares the detection data stream formed by the two previous and subsequent inspections with the pre-stored medical imaging standard database 700, and forms two similarity results to determine the size of the two similarity results;
  • the control judgment module 600 sends a control signal to the detection driving device 500 to control the detection driving device 500 to return to the initial position of this movement Position; when the similarity result of this detection comparison is not less than the similarity result of the previous detection comparison, the detection drive device 500 detects motion is valid this time, and the control judgment module 600 sends a control signal to the detection drive Device 500, controlling the detection driving device 500 to continue to move;
  • the control judgment module 600 controls the detection driving device 500 to move multiple times, and the detection head 400 performs multiple detections until the similarity result of a certain detection comparison meets the corresponding standard determined by the medical judgment as the same;
  • the detection head 400 obtains the detection data stream of a certain detected human organ whose similarity is determined to be the same as the final detection result.
  • the medical imaging standard database 700 includes a plurality of effective image data, each effective image data is segmented from the detection data stream, and the image pictures that best describe the human body parts, organs, and types and names of diseases are selected. ;
  • Each effective image data is also provided with a mark and an attribute, the mark includes at least one of a plurality of marking points, a plurality of marking areas, or a plurality of marking parameters, and the attributes are corresponding body parts, organs, and corresponding Disease type and name;
  • the medical imaging standard database 700 also stores similarity calculation parameters and similarity calculation methods corresponding to each effective image data.
  • control and judgment module 600 divides the detection data stream into a plurality of image frames; compares all the image frames with the effective image data in the medical imaging standard database 700, and calculates the parameters and similarities according to the preset similarity
  • the degree calculation method calculates the similarity between the marked points, marked areas or multiple marking parameters in each effective image data and the image frame to determine the image frame closest to the effective image data and the similarity of the effective image data, and This similarity is used as a similarity result.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Biophysics (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Radiology & Medical Imaging (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Ultra Sonic Daignosis Equipment (AREA)

Abstract

一种B超机器人自动检测方法及相应的系统，该方法包括：将获取人体图像与预存的人体标准数据库（200）比对，确定人体器官坐标数据；根据人体器官坐标数据驱动检测驱动装置（500）运动至相应的器官位置；之后开启检测头（400），对人体进行多次检测，检测一次获取一个检测数据流，每次检测数据流与预存的医学标准影像数据库（700）比对后得到一个相似度结果，之后再驱动检测驱动装置精细调整位置后进行下一次检测；将前后两次检测的相似度结果进行比对，若相似度结果变小，则恢复至本次移动的初始位置，否则本次运动有效；重复多次后，直至判断出某次相似度结果符合医学判定为相同的标准为止，以判定为相同时刻的检测数据流为检测结果。

Description

一种B超机器人自动检测方法及系统 技术领域
本发明涉及医疗设备领域，具体地说，涉及一种B超机器人自动检测方法及系统。
背景技术
B超检测是现有针对人体器官以及血流信息监测最为直接和准确地检测手段，现有的B超设备在对人体进行检测时，需要具有专业医学知识以及检测技能的专业技师操作，技师要根据检测是检测头的放置位置，角度，显示的图像等综合判断，从而确定所要获取的检测影像作为检测结论。
因为以上原因，目前检测技师人员严重不足，在基层医院以及在较偏远的卫生所，卫生室等基层医疗机构不可能也配备检测技师，所以就会导致在基层医疗机构无法做B超检测，对有相应检测需求的病患存在检测不及时，延误诊治的可能。
发明内容
本发明的目的在于提供一种B超机器人自动检测方法及系统，旨在解决的现有B超只有借助专业技师才能操作，从而多数导致基层医疗机构无法做相应检测服务的技术问题。
本发明公开的一种B超机器人自动检测方法，包括以下步骤：
步骤1：获取人体图像，并将人体图像与预设置存储的人体标准数据库比对，确定人体相应器官所在的位置以及区域范围；
步骤2：根据确定的人体相应器官所在的位置以及区域范围，形成相应的人体器官坐标数据；
步骤3：提供连接有检测头的检测驱动装置，所述检测驱动装置根据人体器官坐标数据，做出第一运动，直至运动至相应的器官位置；
步骤4：所述检测头对相应器官进行检测形成第一检测数据流，将所述第一检测数据流与预存储的医学影像标准数据库比对，并反馈比对的相似度结 果；
步骤5：所述检测驱动装置继续运动，做出第二运动，所述检测头对相应器官进行检测得到第二检测数据流，将所述第二检测数据流与预存储的医学影像标准数据库比对，并反馈比对的相似度结果，执行步骤6；
步骤6：判断本次检测比对的相似度结果与上一次检测比对的相似度结果的大小；
当本次检测比对的相似度结果小于上一次检测比对的相似度结果时，所述检测驱动装置返回本次运动的初始位置，并重新执行步骤5；
当本次检测比对的相似度结果不小于上一次检测比对的相似度结果时，所述检测驱动装置本次运动有效，并执行步骤7；
步骤7：判断本次检测比对的所述相似度结果是否满足医学判定为相同的相应标准；若满足，则所述检测头获取本次检测的人体器官检测数据流做为最终检测结果；若不满足，则跳转至所述步骤5继续执行。
进一步地，所述人体标准数据库包括根据人体体型的不同标注出相应器官所在位置的多个人体图像模型；每个人体图像模型对应于一种人体体型，每个人体图像模型上标注有人体器官以及血管所在位置。
进一步地，所述步骤2包括：
步骤21：确定人体坐标原点；
步骤22：根据人体相应器官所在的位置以及区域范围，确定人体器官相对于所述坐标原点的坐标位置以及范围；
步骤23：将所述坐标位置以及范围形成集合，作为所述人体器官坐标数据。
进一步地，所述医学影像标准数据库包括多幅有效图像数据，每幅有效图像数据均是从检测数据流中分割后，挑选出的最能说明人体部位、器官以及疾病种类和名称的图像图片；每个有效图像数据上还设置标记以及属性，所述标记包括多个标记点、多个标记区域或多个标记参数中的至少一个，所述属性为所对应的人体部位、器官和相对应疾病种类和名称；所述医学影像标准数据库中还储存有每个有效图像数据所对应的相似度计算参数以及相似度计算方式。
进一步地，所述步骤4包括：
步骤41：将第一检测数据流分割为多个图像帧；
步骤42：将所有图像帧与医学影像标准数据库中的有效图像数据进行比对；
步骤43：根据预设的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧的相似度进行计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似度相似度结果。
如权利要求5所述的B超机器人自动检测方法，其特征在于，所述步骤5包括：
步骤51：将第二检测数据流分割为多个图像帧；
步骤52：将所有图像帧与医学影像标准数据库中的有效图像数据进行比对；
步骤53：根据预设的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧的相似度进行计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似度作为相似度结果。
进一步地，所述第二运动具有多次，所述检测驱动装置每运动一次，所述检测头均对相应器官进行一次检测，形成一个第二检测数据流，并得到一个相似度结果。
本发明还提供了一种B超机器人自动检测系统，包括：
图像检测装置，所述图像检测装置用于检测所述人体，形成人体图像；
人体标准数据库，所述人体标准数据库包括根据人体体型的不同标注出相应器官所在位置的多个人体图像模型；每个人体图像模型对应于一种人体体型，每个人体图像模型上标注有人体器官所在位置；
运算模块，所述运算模块用于计算所述人体器官所在的坐标，形成人体器官坐标；
检测头，所述检测头用于对人体器官检测，形成检测数据流；
检测驱动装置，所述检测驱动装置用于驱动所述检测头运动；所述检测驱 动装置根据所述人体器官坐标驱动所述检测头运动至相应器官进行检测的初始位置；
控制判断模块，所述控制判断模块用于控制所述检测驱动装置的后续运动，并控制所述检测头在每次运动后进行检测，形成检测数据流；
所述控制判断模块将前后两次检测形成的检测数据流与预存储的医学影像标准数据库比对，并形成两个相似度结果，判断两个相似度结果的大小；
当所述本次检测的相似度结果小于上一次检测的相似度结果时，所述控制判断模块发送控制信号至所述检测驱动装置，控制所述检测驱动装置返回本次运动的初始位置；当本次检测比对的相似度结果不小于上一次检测比对的相似度结果时，检测驱动装置本次检测运动有效，且所述控制判断模块发送控制信号至所述检测驱动装置，控制所述检测驱动装置继续运动；
所述控制判断模块控制所述检测驱动装置多次运动，以及检测头多次检测后，直至某次检测比对的所述相似度结果满足医学判定为相同的相应标准为止；
所述检测头获取相似度判定为相同的某次检测的人体器官检测数据流做为最终检测结果。
进一步地，所述医学影像标准数据库包括多幅有效图像数据，每幅有效图像数据均是从检测数据流中分割后，挑选出的最能说明人体部位、器官以及疾病种类和名称的图像图片；每个有效图像数据上还设置标记以及属性，所述标记包括多个标记点、多个标记区域或多个标记参数中的至少一个，所述属性为所对应的人体部位、器官和相对应疾病种类和名称；所述医学影像标准数据库中还储存有每个有效图像数据所对应的相似度计算参数以及相似度计算方式。
进一步地，所述控制判断模块将检测数据流分割为多个图像帧；将所有图像帧与医学影像标准数据库中的有效图像数据进行比对，并根据预设的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧中的相似度计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似度作为相似度结果。
本发明公开的B超机器人自动检测方法首先获取人体图像，并将人体图像 与人体标准数据库比对，之后根据人体各个器官的位置，确定人体各个器官所在的位置和区域，并将其转换为包括检测驱动装置在内的机器人所能识别的人体器官坐标数据；检测驱动装置根据人体器官坐标驱动检测头移动到相应的人体器官，并开始对人体器官检测，每检测一次得到一个检测数据流，同时检测驱动装置移动一次并进行下一次的检测，得到另一个检测数据流；将前后两次检测的检测数据流分别分割成多个图像帧。将每个图像帧与预存储的医学影像标准数据库中的有效图像数据比对，分别得到前后两次与有效图像数据最接近的图像帧以及有效图像数据相似度，并将该相似度作为相似度结果；比对前后两次相似度结果的变化；根据前后两次相似度结果的大小变化确定，后一次所述检测驱动装置的移动是否有效；在检测驱动装置移动无效的情况下，检测驱动装置恢复至后一次移动的初始位置，并重新移动以及检测；在检测驱动装置移动有效的情况下，检测驱动装置移动下一次，检测头获得下一次的检测数据流，并重复上述的比较；本方法在进行上述的多次重复的移动和检测后，直至某次移动和检测后的图像帧与有效图像数据的相似度结果满足医学判定为相同的标准为止，并将某次检测的检测数据流作为最终的检测结果。
在本实施方式中，通过人体图像与人体标准数据库的比对，能够判定不同体型特征的人体各个器官所处的位置以及区域，为后期检测驱动装置对器官位置的初步定位提供条件；通过对前后两次检测的相似度结果的判断，从而智能判断检测驱动装置的运动位置是否有效，即这样的运动是否是朝着校测完成相同的方向发展，也就是说，本次检测后得到的检测数据流中的图像帧与有效图像数据越来越相似，从而在经过多次检测后，获得满足医学定义为相同的相似度要求，从而完成自动检测；其中，可以通过设置满足医学定义为相同的相似阈值，在达到相似阈值时，直接自动获得检测的检测数据流作为检测结果。是指在医学的定义上，当其中的检测图像数据流中的图像帧与有效图像数据的比对，在标记点、标记区域、标记参数中的相似度达到某一个相似阈值，则认定为相同。例如，对图像帧与有效图像数据对比时，当重合的标记点数量占标记点总数量的90％时，则认定为相同，此时90％就是设定的相似阈值。
在本实施方式中，不同的器官、部位可以设定不同的相似阈值。
根据以上说明，可以确定的是利用本检测方法能够使检测驱动装置自主的移动及微调，进一步的完全自主智能的确定所要检测的位置以及器官，自主的调整检测的角度、深度等数据，且能够对检测的检测数据流进行识别比对，判定检测数据流是否符合预设的医学影像标准数据库，从而自动获取检测数据流作为最终的检测结果。本检测方法的整个检测过程不需要人为介入，仅需要有预设的人体标准数据库以及医学影像标准数据库就能完成相应的B超检测。另外，其中的人体标准数据库以及医学影像标准数据库，通过人为分别对人体图像以及多幅有效图像数据标记后获得，根据检测的单位个体的数据数量的增加，能够不断的标记和扩充，从而使该人体标准数据库以及医学影像标准数据库囊括并细化更多的人体体型、医学疾病以及B超影像等。
再有，该人体标准数据库以及医学影像标准数据库在达到基本数量后，通过人工智能的学习，能够极大地丰富和扩展数据类型和数量，进一步地提升自动检测的准确度，以及识别效率，能够更好地服务于被检测对象。
本发明的B超机器人自动检测系统采用了本发明相应的自动检测方法，该B超机器人自动检测系统能够智能化的自动完成B超的检测，不需要人为介入。
附图说明
图1及图2是本发明B超机器人自动检测方法的流程示意图；
图3是本发明人体标准数据库说明图；
图4是本发明医学影像标准数据库说明图；
图5是本发明有效图像数据的说明图；
图6是不发明B超机器人自动检测系统的模块示意图。
具体实施方式
下面结合具体实施例和说明书附图对本发明做进一步阐述和说明:
请参考图1以及图2，本发明提供一种B超机器人自动检测方法，该检测方法包括：
步骤1：获取人体图像，并将人体图像与预设置存储的人体标准数据库比对，确定人体相应器官所在的位置以及区域范围。
人体图像是通过图像检测装置获得，可以通过该图像检测装置获得不同角度的多幅人体图像，根据人体图像表现出的人体的高矮胖瘦的体型可以进一步的确认不同的人体中各个器官所对应的位置。
参阅图3，具体的，将人体图像与人体标准数据库比对，人体标准数据库包括根据人体体型的不同，人为手动标注出相应器官所在位置的多个人体图像模型，该人体图像模型为一种数据格式，例如也是一种图像。其中，每个人体图像模型对应于一种人体体型，每个人体图像模型上标注有人体器官以及血管所在的位置。将人体图像与人体标准数据库比对后，从所述人体标准数据库中找出最相接近待检测的人体体型的人体图像模型，将最接近的人体图像模型中标记的器官以及血管作为该人体的相应器官所在的位置以及区域范围。
另外，还需要说明的是，在本发明中人体标准数据库中的人体图像模型是人为对人体图像上标记相应器官、部位后形成，一个人体标准数据库中的人体图像模型均是人为手动标记，也可是是人工智能学习对多个已经手动标记的人体图像模型学习后扩展得到的人体标准数据库。
步骤2：根据确定的人体相应器官所在的位置以及区域范围，形成相应的人体器官坐标数据。
具体的，该步骤包括：
步骤21：确定人体坐标原点；其中，可以以人体较为有特征的位置作为坐标原点，例如肚脐眼，或者两眼连线的中点。
步骤22：根据人体相应器官所在的位置以及区域范围，确定人体器官相对于所述坐标原点的坐标位置以及范围；
在本步骤中，将人体相应器官所在的位置以及区域范围通过坐标换算转移至坐标原点所在的坐标系中，即确定人体器官相对于坐标原点的坐标位置以及范围。
步骤23：将所述坐标位置以及范围形成集合，作为所述人体器官坐标数据。
其中，根据不同的器官可以形成多个器官的坐标范围和器官的中心点坐标，将这些坐标作为集合，就形成了与人体体型相对应的人体器官坐标数据。
步骤3：提供连接有检测头的检测驱动装置，所述检测驱动装置根据人体 器官坐标数据，做出第一运动，直至运动至相应的器官位置。
在本实施方式中，在检测开始前，可以通过录入等方式确定所要检测的人体器官，或者检测器官区域等信息，之后，所述检测驱动装置根据已经确定的人体器官坐标数据，提取出所要检测的人体器官在坐标系中的坐标位置，从而按照坐标位置快速的运动至相应位置，为检测头的检测做好准备。
在本步骤中，检测驱动装置的移动以较快速度移动至相应的器官大致位置，注重的是移动的速度，而移动的位置仅是待检测人体器官的大致位置，并不能达到医学检测的准确性要求，在后续步骤中，检测驱动装置将会驱动检测头精确移动，再进行检测。
步骤4：所述检测头对相应器官进行检测形成第一检测数据流，将所述第一检测数据流与预存储的医学影像标准数据库比对，并反馈比对的相似度结果。
参阅图4及图5，所述医学影像标准数据库由人为手动标记后，预先储存作为比对标准使用的数据，其包括多幅有效图像数据，每幅有效图像数据均是从检测数据流中分割后，挑选出的最能说明人体部位、器官以及疾病种类和名称的图像图片；每个有效图像数据上还设置标记以及属性，所述标记包括多个标记点、多个标记区域或多个标记参数中的至少一个，所述属性为所对应的人体部位、器官和相对应疾病种类和名称；所述医学影像标准数据库中还储存有每个有效图像数据所对应的相似度计算参数以及相似度计算方式。
在本实施方式中，将第一检测数据流分割成多个图像帧，每个图像帧就是一幅图像图片。每个图像图片中都包含人体结构或者器官的相应位置、构造等B超信息。
在本实施方式中，标记点标记的为每个器官上的特殊点，这样的标记点具有多个，多个标记点的组合就能唯一代表一个相应的器官；其中标记区域可以是多个标记点所合围的区域，一个标记区域的尺寸、边缘变化形状等信息也唯一代表一个相应的器官；标记参数是器官中的液体、血流或者介质的密度、流量等参数的范围，通过这些标记参数能够唯一识别所代表的一个相应器官。
在本实施方式中，每个有效图像数据设置的人体部位、器官以及相应疾病 种类和名称等，利用这些数据可以对有效图像数据进行分类，当某个图像帧与某一个有效图像数据判定为相同时，可以直接以该有效图像数据所代表的属性来判断该检测的类型。同时，也可以将有效图像数据分类保存，有多类的有效图像数据形成一个医学影像标准数据库。通过分类保存、识别的方式，能够有效提高后续将图像帧与有效图像数据比对的速度，此时图像帧只需要与标记有相应器官、相应疾病或者相应人体部位的有效图像数据进行比对，而不是整个医学影像标准数据库中的所有有效图像数据进行比对，提高比对的效率。
在本实施方式中，相似度计算参数以及相似度计算方式针对不同的器官、不同的部位并不相同。这样的计算方式以及计算参数，可以通过人为的预设定。例如通过人为设定肾脏的相似度计算方式和相似度计算参数。在对比图像帧与有效图像数据相似度时，通过该相似度计算方式和相似度计算参数来计算。不同的器官、人体部位以及疾病可能具有不同的该相似度计算方式和相似度计算参数。
其中，医学影像标准数据库中的有效图像数据也是通过人为对检测数据流中截取的检测图像进行标记，属性填充，设置相似度计算参数、相似度计算方法后形成。当然，该医学影像标准数据库也可是人工智能对多个已经标记的有效图形数据的学习后扩展得到地。
同样的，一个器官上标记的标记点、标记区域、标记参数、相似度计算参数的数值、相似度计算方法等都可以通过人工智能学习得到。
在本实施方式中，只要人工手动初始标记的相应的人体器官、部位的位置准确；只要人工手动标记的有效图像数据的初始标记点、标记区域、标记参数以及相似度计算参数、相似度计算方法准确，在经过足够多的录入的数据的人工智能学习后，该人体图像模型、医学标准数据库就能足够涵盖全部的人体数据、以及人体所对应的B超数据，能够完全实现自动化检测。
具体的，该步骤包括:
步骤41：将第一检测数据流分割为多个图像帧；
步骤42：将所有图像帧与医学影像标准数据库中的有效图像数据进行比对；
步骤43：根据预设的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧的相似度进行计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似度作为相似度结果。
与步骤4相似，其步骤5包括：所述检测驱动装置继续运动，做出第二运动，所述检测头对相应器官进行检测得到第二检测数据流，将所述第二检测数据流与预存储的医学影像标准数据库比对，并反馈比对的相似度结果，并执行步骤6。
具体的，包括：
步骤51：将第二检测数据流分割为多个图像帧；
步骤52：将所有图像帧与医学影像标准数据库中的有效图像数据进行比对；
步骤53：根据预设的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧的相似度进行计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似度作为相似度结果。
在本实施方式中，当检测驱动装置经第一运动运动至相应的器官位置后，仅表示，检测头找到了相应的器官，并不能满足B超检测的角度，深度，位置等其他参数的精度要求。此时检测驱动装置继续运动，做出第二运动，该第二运动是精度运动，可以按照较高精度进行角度、深度的调整，保证检测头的检测精度。
步骤6：判断本次检测比对的相似度结果与上一次检测比对的相似度结果的大小；
当本次检测比对的相似度结果小于上一次检测比对的相似度结果时，所述检测驱动装置返回本次运动的初始位置，并重新执行步骤5；
当本次检测比对的相似度结果不小于上一次检测比对的相似度结果时，所述检测驱动装置本次运动有效，并执行步骤7。
在本实施方式中，B超检测需要多次对检测头的位置、角度、深度进行调 整，在某次调整后，检测比对的相似度结果小于上一次检测比对的相似度结果即意味着：本次检测头的运动位置为朝着检测完成相反的方向发展，也就是说，本次检测后得到的检测数据流中的图像帧与有效图像数据越来越不相似了，若，按照这样的结果一直发展下去，势必永远不能获得满足医学定义为相同的相似度要求，无法完成自动检测，此时需要检测头恢复至初始位置，重新调整运动的角度、深度以及位置。
在某次调整后，检测比对的相似度结果不小于上一次检测比对的相似度结果即意味着：本次检测头的运动位置为朝着校测完成相同的方向发展，也就是说，本次检测后得到的检测数据流中的图像帧与有效图像数据越来越相似了，若，按照这样的结果一直发展下去，势必总会获得满足医学定义为相同的相似度要求，从而完成自动检测，这时检测头调整运动的角度、深度以及位置是有效的。
步骤7：判断本次检测比对的所述相似度结果是否满足医学判定为相同的相应标准；若满足，则所述检测头获取本次检测的人体器官检测数据流做为最终检测结果；
若不满足，则跳转至所述步骤5继续执行。
在本实施方式中，所述第二运动具有多次，所述检测驱动装置每运动一次，所述检测头均对相应器官进行一次检测，形成一个第二检测数据流，并得到一个相似度结果。随着检测的次数的增加，检测对比的相似度结果会持续的进一步提高。在随着相似度结果的进一步提高，直至本次检测比对的所述相似度结果满足医学判定为相同的相应标准，则认为一次检测的完成，此时将该次检测的所述检测头获取的人体器官检测数据流做为最终检测结果。
也就是说这时，检测头对人体器官的检测内容与医学标准数据库中某一个有效图像数据比对后，按照医学标准相同，比对成功，则自动将该次检测的检测数据流作为检测结果。
按照医学标准比对相同，是指在医学的定义上，当其中的检测图像数据流中的图像帧与有效图像数据的比对，在标记点、标记区域、标记参数中的相似度达到某一个相似阈值，则认定为相同。例如，对图像帧与有效图像数据对比 时，当重合的标记点数量占标记点总数量的90％时，则认定为相同，此时90％就是设定的相似阈值。在本实施方式中，不同的器官、部位可以设定不同的相似阈值。
利用本发明的方法，在检测时，只需要检测者躺在相应的位置后，通过拍照、图像识别判定人体的体型后，检测驱动装置就会驱动检测头运动至所要检测的相应的器官和部位开始进行检测，同时检测驱动装置会实时微调检测头的检测角度、深度、位置，直至经过某次调整后获取的B超检测数据，即检测数据流与预先储存或者学习后得到的医学影像标准数据库中的某个有效图像数据比对后，符合医学判定相同的标准，则本次检测就会结束，检测头自动获取该次调整后的检测数据流，作为本次检测的结果。整个检测的过程不需要认为介入和操作即可完成，现有B超只有借助专业技师才能操作，从而多数导致基层医疗机构无法做相应检测服务的技术问题。在本发明的一种实施例中，一个人体标准数据库、医学影像标准数据库上可以连接多个检测驱动装置以及检测头，能够进行足够多的检测服务，同时获得足够度的检测结果。当采用人工智能学习获得人体标准数据库、医学影像标准数据库时，通过分散式同时连接的多个检测头和检测驱动装置，又能够提供足够多的检测结果作为人工智能学习的学习样本，能够极大的提高人工智能学习的效率以及人体标准数据库、医学影像标准数据库的可靠性。
请参阅图6，本发明同时还提供了一种B超机器人自动检测系统，该B超机器人自动检测系统包括：
图像检测装置100，所述图像检测装置100用于检测所述人体，形成人体图像；
人体标准数据库200，所述人体标准数据库200包括根据人体体型的不同标注出相应器官所在位置的多个人体图像模型；每个人体图像模型对应于一种人体体型，每个人体图像模型上标注有人体器官所在位置；
运算模块300，所述运算模块300用于计算所述人体器官所在的坐标，形成人体器官坐标；
检测头400，所述检测头400用于对人体器官检测，形成检测数据流；
检测驱动装置500，所述检测驱动装置500用于驱动所述检测头400运动；所述检测驱动装置500根据所述人体器官坐标驱动所述检测头400运动至相应器官进行检测的初始位置；
控制判断模块600，所述控制判断模块600用于控制所述检测驱动装置500的后续运动，并控制所述检测头400在每次运动后进行检测，形成检测数据流；
所述控制判断模块600将前后两次检测形成的检测数据流与预存储的医学影像标准数据库700比对，并形成两个相似度结果，判断两个相似度结果的大小；
当所述本次检测的相似度结果小于上一次检测的相似度结果时，所述控制判断模块600发送控制信号至所述检测驱动装置500，控制所述检测驱动装置500返回本次运动的初始位置；当本次检测比对的相似度结果不小于上一次检测比对的相似度结果时，检测驱动装置500本次检测运动有效，且所述控制判断模块600发送控制信号至所述检测驱动装置500，控制所述检测驱动装置500继续运动；
所述控制判断模块600控制所述检测驱动装置500多次运动，以及检测头400多次检测后，直至某次检测比对的所述相似度结果满足医学判定为相同的相应标准为止；
所述检测头400获取相似度判定为相同的某次检测的人体器官的检测数据流做为最终检测结果。
进一步地，所述医学影像标准数据库700包括多幅有效图像数据，每幅有效图像数据均是从检测数据流中分割后，挑选出的最能说明人体部位、器官以及疾病种类和名称的图像图片；每个有效图像数据上还设置标记以及属性，所述标记包括多个标记点、多个标记区域或多个标记参数中的至少一个，所述属性为所对应的人体部位、器官和相对应疾病种类和名称；所述医学影像标准数据库700中还储存有每个有效图像数据所对应的相似度计算参数以及相似度计算方式。
进一步地，所述控制判断模块600将检测数据流分割为多个图像帧；将所有图像帧与医学影像标准数据库700中的有效图像数据进行比对，并根据预设 的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧中的相似度计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似度作为相似度结果。
最后应当说明的是，以上实施例仅用以说明本发明的技术方案，而非对本发明保护范围的限制，尽管参照较佳实施例对本发明作了详细地说明，本领域的普通技术人员应当理解，可以对本发明的技术方案进行修改或者等同替换，而不脱离本发明技术方案的实质和范围。

Claims (10)

  1. 一种B超机器人自动检测方法，其特征在于，包括以下步骤：
    步骤1：获取人体图像，并将人体图像与预设置存储的人体标准数据库比对，确定人体相应器官所在的位置以及区域范围；
    步骤2：根据确定的人体相应器官所在的位置以及区域范围，形成相应的人体器官坐标数据；
    步骤3：提供连接有检测头的检测驱动装置，所述检测驱动装置根据人体器官坐标数据，做出第一运动，直至运动至相应的器官位置；
    步骤4：所述检测头对相应器官进行检测形成第一检测数据流，将所述第一检测数据流与预存储的医学影像标准数据库比对，并反馈比对的相似度结果；
    步骤5：所述检测驱动装置继续运动，做出第二运动，所述检测头对相应器官进行检测得到第二检测数据流，将所述第二检测数据流与预存储的医学影像标准数据库比对，并反馈比对的相似度结果，执行步骤6；
    步骤6：判断本次检测比对的相似度结果与上一次检测比对的相似度结果的大小；
    当本次检测比对的相似度结果小于上一次检测比对的相似度结果时，所述检测驱动装置返回本次运动的初始位置，并重新执行步骤5；
    当本次检测比对的相似度结果不小于上一次检测比对的相似度结果时，所述检测驱动装置本次运动有效，并执行步骤7；
    步骤7：判断本次检测比对的所述相似度结果是否满足医学判定为相同的相应标准；若满足，则所述检测头获取本次检测的人体器官检测数据流做为最终检测结果；若不满足，则跳转至所述步骤5继续执行。
  2. 如权利要求1所述的B超机器人自动检测方法，其特征在于，所述人体标准数据库包括根据人体体型的不同标注出相应器官所在位置的多个人体图像模型；每个人体图像模型对应于一种人体体型，每个人体图像模型上标注有人体器官以及血管所在位置。
  3. 如权利要求1所述的B超机器人自动检测方法，其特征在于，所述步骤2包括：
    步骤21：确定人体坐标原点；
    步骤22：根据人体相应器官所在的位置以及区域范围，确定人体器官相对于所述坐标原点的坐标位置以及范围；
    步骤23：将所述坐标位置以及范围形成集合，作为所述人体器官坐标数据。
  4. 如权利要求1至3任意一项所述的B超机器人自动检测方法，其特征在于，所述医学影像标准数据库包括多幅有效图像数据，每幅有效图像数据均是从检测数据流中分割后，挑选出的最能说明人体部位、器官以及疾病种类和名称的图像图片；每个有效图像数据上还设置标记以及属性，所述标记包括多个标记点、多个标记区域或多个标记参数中的至少一个，所述属性为所对应的人体部位、器官和相对应疾病种类和名称；所述医学影像标准数据库中还储存有每个有效图像数据所对应的相似度计算参数以及相似度计算方式。
  5. 如权利要求4所述的B超机器人自动检测方法，其特征在于，所述步骤4包括：
    步骤41：将第一检测数据流分割为多个图像帧；
    步骤42：将所有图像帧与医学影像标准数据库中的有效图像数据进行比对；
    步骤43：根据预设的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧的相似度进行计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似度作为相似度结果。
  6. 如权利要求5所述的B超机器人自动检测方法，其特征在于，所述步骤5包括：
    步骤51：将第二检测数据流分割为多个图像帧；
    步骤52：将所有图像帧与医学影像标准数据库中的有效图像数据进行比对；
    步骤53：根据预设的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧的相似度进行计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似 度作为相似度结果。
  7. 如权利要求6所述的B超机器人自动检测方法，其特征在于，所述第二运动具有多次，所述检测驱动装置每运动一次，所述检测头均对相应器官进行一次检测，形成一个第二检测数据流，并得到一个相似度结果。
  8. 一种B超机器人自动检测系统，其特征在于，包括：
    图像检测装置，所述图像检测装置用于检测所述人体，形成人体图像；
    人体标准数据库，所述人体标准数据库包括根据人体体型的不同标注出相应器官所在位置的多个人体图像模型；每个人体图像模型对应于一种人体体型，每个人体图像模型上标注有人体器官所在位置；
    运算模块，所述运算模块用于计算所述人体器官所在的坐标，形成人体器官坐标；
    检测头，所述检测头用于对人体器官检测，形成检测数据流；
    检测驱动装置，所述检测驱动装置用于驱动所述检测头运动；所述检测驱动装置根据所述人体器官坐标驱动所述检测头运动至相应器官进行检测的初始位置；
    控制判断模块，所述控制判断模块用于控制所述检测驱动装置的后续运动，并控制所述检测头在每次运动后进行检测，形成检测数据流；
    所述控制判断模块将前后两次检测形成的检测数据流与预存储的医学影像标准数据库比对，并形成两个相似度结果，判断两个相似度结果的大小；
    当所述本次检测的相似度结果小于上一次检测的相似度结果时，所述控制判断模块发送控制信号至所述检测驱动装置，控制所述检测驱动装置返回本次检测运动的初始位置；当本次检测比对的相似度结果不小于上一次检测比对的相似度结果时，检测驱动装置本次检测运动有效，且所述控制判断模块发送控制信号至所述检测驱动装置，控制所述检测驱动装置继续运动；
    所述控制判断模块控制所述检测驱动装置多次运动，以及检测头多次检测后，直至某次检测比对的所述相似度结果满足医学判定为相同的相应标准为止；
    所述检测头获取相似度判定为相同的某次检测的人体器官检测数据流做 为最终检测结果。
  9. 如权利要求8所述的B超机器人自动检测系统，其特征在于，所述医学影像标准数据库包括多幅有效图像数据，每幅有效图像数据均是从检测数据流中分割后，挑选出的最能说明人体部位、器官以及疾病种类和名称的图像图片；每个有效图像数据上还设置标记以及属性，所述标记包括多个标记点、多个标记区域或多个标记参数中的至少一个，所述属性为所对应的人体部位、器官和相对应疾病种类和名称；所述医学影像标准数据库中还储存有每个有效图像数据所对应的相似度计算参数以及相似度计算方式。
  10. 如权利要求8所述的B超机器人自动检测系统，其特征在于，所述控制判断模块将检测数据流分割为多个图像帧；将所有图像帧与医学影像标准数据库中的有效图像数据进行比对，并根据预设的相似度计算参数以及相似度计算方法对每个有效图像数据中标记的标记点、标记区域或多个标记参数与图像帧中的相似度计算，确定与有效图像数据最为接近的图像帧以及有效图像数据相似度，并将该相似度作为相似度结果。
PCT/CN2018/120395 2018-12-11 2018-12-11 一种b超机器人自动检测方法及系统 WO2020118534A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/120395 WO2020118534A1 (zh) 2018-12-11 2018-12-11 一种b超机器人自动检测方法及系统

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/120395 WO2020118534A1 (zh) 2018-12-11 2018-12-11 一种b超机器人自动检测方法及系统

Publications (1)

Publication Number Publication Date
WO2020118534A1 true WO2020118534A1 (zh) 2020-06-18

Family

ID=71076154

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/120395 WO2020118534A1 (zh) 2018-12-11 2018-12-11 一种b超机器人自动检测方法及系统

Country Status (1)

Country Link
WO (1) WO2020118534A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112132805A (zh) * 2020-09-22 2020-12-25 深圳市人工智能与机器人研究院 一种基于人体特征的超声机器人状态归一化方法及系统

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010279408A (ja) * 2009-06-02 2010-12-16 Toshiba Medical Systems Corp 超音波診断装置
CN105877780A (zh) * 2015-08-25 2016-08-24 上海深博医疗器械有限公司 全自动超声扫描仪及扫描检测方法
CN106419957A (zh) * 2016-10-09 2017-02-22 深圳市罗伯医疗科技有限公司 一种超声扫描装置辅助系统
CN106580368A (zh) * 2016-11-26 2017-04-26 汕头市超声仪器研究所有限公司 一种全自动超声波诊断方法
CN107427270A (zh) * 2015-02-23 2017-12-01 西门子保健有限责任公司 用于医疗诊断装置的自动定位的方法和系统
CN107582097A (zh) * 2017-07-18 2018-01-16 中山大学附属第医院 一种基于多模态超声组学的智能辅助决策系统
CN107679574A (zh) * 2017-09-29 2018-02-09 深圳开立生物医疗科技股份有限公司 超声图像处理方法及系统
CN108577885A (zh) * 2017-12-06 2018-09-28 浙江大学 一种超声检测装置

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010279408A (ja) * 2009-06-02 2010-12-16 Toshiba Medical Systems Corp 超音波診断装置
CN107427270A (zh) * 2015-02-23 2017-12-01 西门子保健有限责任公司 用于医疗诊断装置的自动定位的方法和系统
CN105877780A (zh) * 2015-08-25 2016-08-24 上海深博医疗器械有限公司 全自动超声扫描仪及扫描检测方法
CN106419957A (zh) * 2016-10-09 2017-02-22 深圳市罗伯医疗科技有限公司 一种超声扫描装置辅助系统
CN106580368A (zh) * 2016-11-26 2017-04-26 汕头市超声仪器研究所有限公司 一种全自动超声波诊断方法
CN107582097A (zh) * 2017-07-18 2018-01-16 中山大学附属第医院 一种基于多模态超声组学的智能辅助决策系统
CN107679574A (zh) * 2017-09-29 2018-02-09 深圳开立生物医疗科技股份有限公司 超声图像处理方法及系统
CN108577885A (zh) * 2017-12-06 2018-09-28 浙江大学 一种超声检测装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112132805A (zh) * 2020-09-22 2020-12-25 深圳市人工智能与机器人研究院 一种基于人体特征的超声机器人状态归一化方法及系统
CN112132805B (zh) * 2020-09-22 2024-02-09 深圳市人工智能与机器人研究院 一种基于人体特征的超声机器人状态归一化方法及系统

Similar Documents

Publication Publication Date Title
CN112215843B (zh) 超声智能成像导航方法、装置、超声设备及存储介质
CN112155729B (zh) 手术穿刺路径智能自动化规划方法及系统和医疗系统
US8009887B2 (en) Method and system for automatic quantification of aortic valve function from 4D computed tomography data using a physiological model
JP2021520939A (ja) 適応的超音波スキャニング
CN111134727B (zh) 基于神经网络的静脉与动脉识别的穿刺引导系统
WO2020125499A9 (zh) 一种操作提示方法及眼镜
US20210077068A1 (en) Systems and methods for automated ultrasound image labeling and quality grading
JP7442548B2 (ja) 誘導式超音波撮像
WO2020118534A1 (zh) 一种b超机器人自动检测方法及系统
Son et al. Crosswalk guidance system for the blind
Charriere et al. Automated surgical step recognition in normalized cataract surgery videos
CN114898882A (zh) 基于超声对右心功能进行评估的方法和系统
CN111145137A (zh) 基于神经网络的静脉与动脉识别方法
Toporek et al. Autonomous image-based ultrasound probe positioning via deep learning
WO2020087732A1 (zh) 基于神经网络的静脉与动脉识别方法与系统
US20210015449A1 (en) Methods and systems for processing and displaying fetal images from ultrasound imaging data
CN111144163A (zh) 基于神经网络的静脉与动脉识别系统
US11850090B2 (en) Guided lung coverage and automated detection using ultrasound devices
US11446002B2 (en) Methods and systems for a medical imaging device
CN115023188A (zh) 超声引导方法和系统
WO2019126611A1 (en) Simplified, interactive, real-time ultrasound biofeedback system for speech remediation
WO2020118535A1 (zh) 一种b超智能辅助采集方法及系统
Tian et al. A Vision-Based Target Localization Method for Robot-Assisted Sonography
EP4094695A1 (en) Ultrasound imaging system
Zin et al. Feature detection and classification of cow motion for predicting calving time

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18942746

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18942746

Country of ref document: EP

Kind code of ref document: A1