CN113658198A - Interactive emphysema focus segmentation method, device, storage medium and equipment - Google Patents
Interactive emphysema focus segmentation method, device, storage medium and equipment Download PDFInfo
- Publication number
- CN113658198A CN113658198A CN202111017521.4A CN202111017521A CN113658198A CN 113658198 A CN113658198 A CN 113658198A CN 202111017521 A CN202111017521 A CN 202111017521A CN 113658198 A CN113658198 A CN 113658198A
- Authority
- CN
- China
- Prior art keywords
- emphysema
- image
- lung
- mask
- segmentation
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention relates to the technical field of digital medical treatment, and provides an interactive emphysema focus segmentation method, device, storage medium and equipment. The method comprises the following steps: acquiring a lung image to be processed, and performing lung parenchyma segmentation processing and filtering processing on the lung image to obtain a lung parenchyma image; according to a preset heat capacity threshold value, pre-dividing the lung parenchyma image to obtain an emphysema initial division mask; performing connected domain analysis and binarization processing on the emphysema initial segmentation mask to obtain an emphysema candidate segmentation mask; performing iterative connected domain analysis and binarization processing on the emphysema candidate segmentation mask by a human-in-loop interaction mode to obtain an emphysema focus segmentation mask; and performing mask operation on the emphysema focus segmentation mask to obtain an emphysema focus region image. The method effectively shortens the time for marking the emphysema nidus, improves the efficiency for segmenting and marking the emphysema nidus, and simultaneously effectively reduces the error rate of label missing of the emphysema positive nidus.
Description
Technical Field
The invention relates to the technical field of digital medical treatment, in particular to an interactive emphysema focus segmentation method, device, storage medium and computer equipment.
Background
In recent years, the incidence of pulmonary diseases has gradually increased due to various causes such as deterioration of air quality, increase in harm of second-hand smoke, and influence of occupational factors, and among these, pulmonary emphysema is a highly harmful pulmonary disease, and the focus segmentation method thereof has been a more important subject of much attention in the industry.
In the prior art, a method for automatically segmenting a focus region of emphysema is often implemented by adopting a deep learning algorithm. However, the emphysema automatic segmentation algorithm implemented based on the deep learning algorithm usually needs to use a large amount of manual labeling data, the acquisition time period of the large amount of manual labeling data is long, and the cost is very high, in addition, due to the fact that the shapes of regions of interest of emphysema are different, the workload and the working difficulty of physician labeling are greatly increased, the problem of inaccurate labeling is easily caused, and finally, the emphysema focus segmentation cost is too high, the efficiency is low, and the accuracy is poor.
Disclosure of Invention
In view of this, the present application provides an interactive emphysema focus segmentation method, apparatus, storage medium and computer device, and mainly aims to solve the technical problems of high cost, low efficiency and poor accuracy of emphysema focus segmentation.
According to a first aspect of the present invention, there is provided an interactive emphysema focus segmentation method, including:
acquiring a lung image to be processed, and performing lung parenchyma segmentation processing and filtering processing on the lung image to obtain a lung parenchyma image;
according to a preset heat capacity threshold value, pre-dividing the lung parenchyma image to obtain an emphysema initial division mask;
performing connected domain analysis and binarization processing on the emphysema initial segmentation mask to obtain an emphysema candidate segmentation mask;
performing iterative connected domain analysis and binarization processing on the emphysema candidate segmentation mask by a human-in-loop interaction mode to obtain an emphysema focus segmentation mask;
and performing mask operation on the emphysema focus segmentation mask to obtain an emphysema focus region image.
According to a second aspect of the present invention, there is provided an interactive emphysema lesion segmentation apparatus, comprising:
the image preprocessing module is used for acquiring a lung image to be processed, and performing lung parenchyma segmentation processing and filtering processing on the lung image to obtain a lung parenchyma image;
the image pre-segmentation module is used for pre-segmenting the lung parenchyma image according to a preset heat capacity threshold value to obtain an emphysema initial segmentation mask;
the image segmentation module is used for carrying out connected domain analysis and binarization processing on the emphysema initial segmentation mask to obtain an emphysema candidate segmentation mask;
the image iteration segmentation module is used for carrying out iterative connected domain analysis and binarization processing on the emphysema candidate segmentation mask through a human-in-loop interaction mode to obtain an emphysema focus segmentation mask;
and the image conversion module is used for carrying out mask operation on the emphysema focus segmentation mask to obtain an emphysema focus region image.
According to a third aspect of the present invention, there is provided a storage medium having stored thereon a computer program which, when executed by a processor, implements the above-described interactive emphysema lesion segmentation method.
According to a fourth aspect of the present invention, there is provided a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the above interactive emphysema lesion segmentation method when executing the program.
The invention provides an interactive emphysema focus segmentation method, a device, a storage medium and computer equipment. According to the method, automatic segmentation of the emphysema focus area is achieved without data models based on a deep learning algorithm and the like, and cost of segmenting the pulmonary tuberculosis focus is effectively reduced. In addition, by means of an interaction mode of a human in a loop, an automatic segmentation algorithm of connected domain analysis, binarization processing and the like, a labeling doctor does not need to label the emphysema nidus from zero, the workload of the labeling doctor is greatly reduced, the time for labeling the emphysema nidus is shortened, the efficiency of segmenting and labeling the emphysema nidus is improved, and the missing label error rate of the emphysema positive nidus is effectively reduced.
The foregoing description is only an overview of the technical solutions of the present application, and the present application can be implemented according to the content of the description in order to make the technical means of the present application more clearly understood, and the following detailed description of the present application is given in order to make the above and other objects, features, and advantages of the present application more clearly understandable.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:
fig. 1 is a flowchart illustrating an interactive emphysema focus segmentation method according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram illustrating an interactive emphysema focus segmentation apparatus according to an embodiment of the present invention;
fig. 3 shows an internal structure diagram of a computer device according to an embodiment of the present invention.
Detailed Description
The invention will be described in detail hereinafter with reference to the accompanying drawings in conjunction with embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Emphysema is a lung disease with great harm, a focus area of the emphysema is often required to be segmented from a chest CT image so as to be used for doctors to follow up the condition development, automatic segmentation of emphysema focuses by using a deep learning algorithm is a standard method commonly used in the industry, but the segmentation of the emphysema focuses is difficult to label: the emphysema focus usually presents a dispersed cavity shadow in a chest CT image, and usually fills the whole lung lobe, or disperses in a star point shape in the whole lung, if a marking doctor manually marks each emphysema region pixel by pixel, the efficiency is extremely low, and emphysema positive regions are easily missed, so that the marking result is not accurate enough, and the inaccuracy of sample marking further causes the model precision to be low, so that the emphysema focus segmentation result output by a deep learning model is not accurate enough.
In an embodiment, as shown in fig. 1, an interactive emphysema lesion segmentation method is provided, which is described by taking an example that the method is applied to a computer device such as a server, where the server may be an independent server, or a cloud server that provides basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a Content Delivery Network (CDN), and a big data and artificial intelligence platform. The method comprises the following steps:
101. and acquiring a lung image to be processed, and performing lung parenchyma segmentation processing and filtering processing on the lung image to obtain a lung parenchyma image.
Specifically, the computer device may acquire a lung image to be labeled (e.g., a chest CT image) through a data interface, and may process the acquired lung image sequentially through a series of algorithms such as a binarization algorithm, a connected domain algorithm, a morphological closing operation, a morphological expanding operation, and a mask algorithm to obtain a rough initial lung parenchyma image, and may further perform a filtering process on the initial lung parenchyma image to remove various types of noise (e.g., salt and pepper noise) in the lung parenchyma image, and finally obtain the filtered lung parenchyma image.
The lung image refers to an image obtained by performing computed tomography on a lung, and generally refers to a chest CT image; the binarization algorithm is to perform binarization operation on pixels in an image, that is, the gray value of a pixel point on the image is set to be 0 or 255, so that the whole image has an obvious black and white effect, and common binarization algorithms mainly comprise a gray average value method, a double peak method, an OTSU method, a Niblack method and the like; the connected domain algorithm is a common binarization image processing method, and can find out adjacent pixels with the same pixel value in each connected domain and mark the adjacent pixels so as to distinguish different connected domains; the morphological closed operation is an algorithm for expanding the image and then corroding the result; the morphological dilation algorithm is an algorithm that integrates all background points in contact with an object into the object and expands the boundary to the outside; the mask algorithm is an algorithm for multiplying the processed image and the pixels of the original image correspondingly to obtain the target image. In this embodiment, the names of the algorithms used in the binarization algorithm, the connected domain algorithm, the morphological closing operation, the morphological dilation operation, and the mask algorithm may be selected according to actual situations, and this embodiment is not specifically limited herein.
Here, the human lung tissue is composed of lung parenchyma and lung interstitium, wherein the lung interstitium includes connective tissue, lymphatic vessels, blood vessels, and the like, and the lung parenchyma includes the extremely terminal alveolar structure of bronchi at all levels in the lung. Some lung diseases occur in the lung interstitium, such as pulmonary fibrosis and interstitial pneumonia, while most serious diseases are caused by pathological changes of the lung parenchyma, and emphysema is a disease occurring in the lung parenchyma, so accurate segmentation of the lung parenchyma is very important for segmentation and labeling of emphysema.
102. And pre-dividing the lung parenchyma image according to a preset heat capacity threshold value to obtain an emphysema initial division mask.
The heat capacity value is an HU value, which is a unit for expressing a CT value in a computed tomography, and different organs and tissues have different HU values according to the degree of absorption of X-rays. Furthermore, the heat capacity threshold refers to a manually set heat capacity value, and by setting the heat capacity threshold, each voxel value of the lung parenchyma image can be compared with the set heat capacity threshold, so that the lung parenchyma image is pre-segmented according to the comparison result to obtain an emphysema initial segmentation mask.
Specifically, emphysema generally appears to be less than-850 HU on the chest CT image, and therefore, this embodiment may use-850 HU as a heat capacity threshold to pre-segment the filtered lung parenchyma image to obtain an emphysema initial segmentation mask. Specifically, the computer device may set-850 HU as a heat capacity threshold, and then mark a region having a HU value less than-850 HU and a region having a HU value exceeding-850 HU in the filtered lung parenchyma image as different values, respectively, thereby obtaining an emphysema candidate region having an emphysema initial segmentation mask.
103. And carrying out connected domain analysis and binarization processing on the emphysema initial segmentation mask to obtain an emphysema candidate segmentation mask.
Specifically, after the emphysema initial segmentation mask is obtained, a connected domain algorithm can be adopted to perform connected domain analysis on the emphysema initial segmentation mask so as to find a connected domain in the emphysema segmentation mask, and then, a plurality of connected domains with different labels can be obtained by marking voxels of the same connected domain. Further, for each connected domain in the connected domain set, the HU value of each connected domain may be obtained as a dynamic threshold of the connected domain through a binarization algorithm of a maximum inter-class variance method. Further, for each connected domain, performing binarization processing on the connected domain by using a dynamic threshold corresponding to the connected domain, wherein a region with an HU value smaller than the dynamic threshold is an emphysema candidate region in the connected domain and can be marked as a numerical value; the region with the HU value exceeding the dynamic threshold is a non-emphysema region, namely a normal region, and can be marked as another numerical value, and the emphysema segmentation mask of the ith connected domain can be obtained in this way. And finally, merging the segmentation masks of all the connected domains to obtain the emphysema candidate segmentation mask.
104. And carrying out iterative connected domain analysis and binarization processing on the emphysema candidate segmentation mask by a human-in-loop interaction mode to obtain an emphysema focus segmentation mask.
The human-in-the-loop interaction mode refers to a mode of interaction between manual examination serving as a link of a closed loop and an emphysema focus segmentation method.
Specifically, the computer device may display the emphysema candidate segmentation mask, so that the user may audit the segmented emphysema candidate segmentation mask to determine whether the mask meets the requirement. Then, the computer device can receive various mask operation requests sent by the user through the input device, and perform corresponding operation processing on the emphysema candidate segmentation mask according to the requests, further, the computer device can perform automatic connected domain analysis and binarization processing on the operated emphysema candidate segmentation mask to obtain an emphysema iterative segmentation mask after automatic segmentation, and finally, the computer device can display the emphysema iterative segmentation mask, so that the user can check and modify the segmented emphysema candidate segmentation mask again. The iteration process can be carried out for a plurality of times until a mask confirmation instruction sent by a user is received, and the emphysema iteration segmentation mask obtained by the last segmentation can be set as the emphysema focus segmentation mask. By the method, the emphysema candidate segmentation mask which does not meet the emphysema focus standard can be continuously optimized in an iterative mode until the qualified emphysema segmentation mask is obtained.
105. And performing mask operation on the emphysema focus segmentation mask to obtain an emphysema focus region image.
The mask algorithm is an algorithm for multiplying the processed image and the pixels of the original image correspondingly to obtain the target image. Specifically, after obtaining the emphysema lesion segmentation mask, the computer device may perform a mask operation on the emphysema lesion segmentation mask, for example, the emphysema lesion segmentation mask may be multiplied (and operated) by the lung image to be labeled, so as to obtain a segmented emphysema lesion region image.
The interactive emphysema focus segmentation method provided by the embodiment includes the steps of firstly carrying out lung parenchyma segmentation processing and filtering processing on a lung image to obtain a lung parenchyma image, then carrying out pre-segmentation on the lung parenchyma image to narrow a focus segmentation range, further generating an emphysema candidate segmentation mask through automatic segmentation algorithms such as connected domain analysis and binarization processing, and finally continuously and iteratively repairing the emphysema candidate segmentation mask through a human-in-loop interaction mode by means of the automatic segmentation algorithm until a qualified emphysema focus segmentation mask is obtained and converted into an emphysema focus region image. According to the method, automatic segmentation of the emphysema focus area is achieved without data models based on a deep learning algorithm and the like, and cost of segmenting the pulmonary tuberculosis focus is effectively reduced. In addition, by means of an interaction mode of a human in a loop, an automatic segmentation algorithm of connected domain analysis, binarization processing and the like, a labeling doctor does not need to label the emphysema nidus from zero, the workload of the labeling doctor is greatly reduced, the time for labeling the emphysema nidus is shortened, the efficiency of segmenting and labeling the emphysema nidus is improved, and the missing label error rate of the emphysema positive nidus is effectively reduced.
In an embodiment, optionally, the step 101 may be implemented by: firstly, carrying out binarization processing on a lung image through a binarization algorithm to obtain a binarized lung image, then carrying out connected processing on the binarized lung image through a connected domain algorithm to obtain a binarized lung connected image, further carrying out morphological closing operation and morphological expanding operation on the binarized lung connected image to obtain a lung parenchyma mask, finally converting the lung parenchyma mask into an initial lung parenchyma image through a mask algorithm, and carrying out median filtering processing on the initial lung parenchyma image to obtain the lung parenchyma image.
In the above embodiment, after the lung image to be processed is acquired, the lung image may be binarized according to a thermal capacity threshold of the lung region to obtain a binarized lung image. Generally speaking, the HU value of the lung region is less than-600 HU, so this embodiment can use-600 HU as the thermal capacity threshold value to perform binarization processing on the lung image, so as to mark the region value with HU value less than-600 HU in the image as one numerical value, and mark the region with HU value over-600 HU in the image as another numerical value, thereby obtaining a binarized lung image. Further, after the binary lung image is obtained, different connected domains in the binary lung image can be marked through a connected domain algorithm, voxels in the same connected domain are marked as the same label, so that different connected domains with different labels are obtained, and finally the volume of each connected domain is calculated, wherein the two connected domains with the largest volume are the connected domains of the left lung region and the right lung region, and the binary lung connected image can be obtained by reserving the connected domains of the left lung region and the right lung region. Furthermore, the cavities of structures such as blood vessels, trachea and the like in the lung in the binary lung communicated image can be filled by performing morphological closing operation and morphological expanding operation on the binary lung communicated image, so that the integrity of the image is improved, and the problem of incomplete lung parenchyma segmentation is avoided. Furthermore, the lung parenchyma mask is multiplied by the chest CT image to be marked, a well segmented initial lung parenchyma image can be obtained, and finally, the initial lung parenchyma image is subjected to median filtering processing, so that white point noise or black point noise (namely salt-pepper noise) which randomly appears in the image can be effectively removed, and a complete and clear lung parenchyma image is obtained.
Compared with the single lung parenchyma segmentation algorithm in the prior art, the method can effectively remove the thorax contour in the lung image and reserve the focus area in the lung image, thereby improving the integrity of the lung parenchyma segmentation.
In one embodiment, the method for performing the morphological closing operation and the morphological dilation operation on the lung image may comprise the following steps: firstly, performing morphological closing operation on left and right lung regions in the binarized lung connected image through a circular structure with a preset radius (for example, a circular structure with a radius of 10), then detecting the edge of the binarized lung connected image after the closing operation through a Roberts operator, and filling a small closed region inside a maximum closed region in the binarized lung connected image after the closing operation into a preset numerical value (for example, filling the small closed region into 1). In this embodiment, the largest closed region is the lung edge, and the small closed region is the lung internal structure such as blood vessels and trachea in the lung. By the method, cavities of structures such as blood vessels, trachea and the like in the lung in the binary lung communicated image can be filled, so that the integrity of the image is improved, and the problem of incomplete lung parenchyma segmentation is avoided. Further, in order to ensure that the lung parenchymal region is completely segmented, the filled binarized lung connected image may be subjected to a morphological dilation operation through a circular structure with a preset radius to obtain a lung parenchymal mask. In this embodiment, the radius of the circular structure may be selected according to actual conditions, and the radius of the circular structure used in the morphological closing operation and the morphological dilation operation is the same. It should be noted that the binarized pulmonary connectivity image is usually obtained by binarizing a heat capacity value (HU value) smaller than-600, but the HU value of structures such as blood vessels and trachea in the lung in the image is usually larger than-600, so that some fine cavities may exist in the binarized pulmonary connectivity image, and the method can effectively fill the fine cavities.
In an embodiment, optionally, the step 102 may be implemented by: and marking the area with the heat capacity value smaller than the heat capacity threshold value in the lung parenchyma image as a first numerical value, and marking the area with the heat capacity value not smaller than the heat capacity threshold value in the lung parenchyma image as a second numerical value to obtain an emphysema initial segmentation mask. In this embodiment, emphysema is generally less than-850 HU on the CT image of the chest, so this embodiment may pre-segment the filtered lung parenchyma image using-850 HU as the heat capacity threshold to obtain an initial segmentation mask for emphysema. In particular, the computer device may set-850 HU to a thermal capacity threshold, and then mark the filtered lung parenchyma image with a first value (e.g., marked as 1) for areas where HU values are less than-850 HU, and with a second value (e.g., marked as 0) for areas where HU values exceed-850 HU, therebyObtaining an emphysema candidate region with an emphysema initial segmentation mask, wherein the emphysema initial segmentation mask can be marked as mkWherein k is 0.
In an embodiment, optionally, the step 103 may be implemented by: firstly, performing connected domain analysis on an emphysema initial segmentation mask by using a connected domain analysis algorithm to obtain a connected domain set comprising a plurality of connected domains, then obtaining a dynamic threshold of each connected domain by using a maximum inter-class variance algorithm for each connected domain in the connected domain set, further performing binarization processing on a lung parenchyma image corresponding to each connected domain by using a binarization algorithm according to the dynamic threshold of each connected domain to obtain an emphysema segmentation mask of each connected domain, and finally merging the emphysema segmentation masks of all the connected domains to obtain an emphysema candidate segmentation mask. The dynamic threshold is a dynamic HU value of which the HU value is continuously adjusted along with continuous iterative segmentation of an initial segmentation mask for emphysema.
In the above embodiment, after obtaining the initial segmentation mask for emphysema, a connected domain algorithm may be used to initially segment the mask for emphysema mkAnd performing connected domain analysis to find connected domains in the emphysema segmentation mask, and then marking voxels of the same connected domain to obtain a plurality of connected domains with different labels. For convenience of illustration, a connected domain set formed by a plurality of connected domains may be referred to herein as { CiAnd c, wherein Ci is a point set of voxels of the ith connected domain. Further, for the connected domain set { CiAnd obtaining the HU value of each connected domain as the dynamic threshold value of the connected domain through a binarization algorithm (OTSU algorithm) of a maximum inter-class variance method. The OTSU algorithm can divide an image into a background part and a target part, the inter-class variance between the background and the target can be used for distinguishing the background from the target, and compared with a common binarization algorithm, the binarization algorithm based on the maximum inter-class variance method has higher accuracy. Further, for the ith connected domain, a dynamic threshold t corresponding to the ith connected domain may be utilizediPerforming binarization processing on the connected domain, wherein HU valueLess than tiThe area of (a) is an emphysema candidate area in the connected domain, and can be marked as a numerical value; HU value exceeding tiThe area of (a) is a non-emphysema area, i.e. a normal area, and can be marked as another numerical value, and in this way, an emphysema segmentation mask of the ith connected domain can be obtained. Furthermore, i is made to pass through all the connected domains, and emphysema segmentation masks of all the connected domains can be obtained. Finally, combining the segmentation masks of all the connected domains to obtain an emphysema candidate segmentation mask, wherein the emphysema candidate segmentation mask can be recorded as mkWherein k is 1.
In an embodiment, optionally, the step 104 may be implemented by: the method comprises the steps of firstly displaying an emphysema candidate segmentation mask, receiving a mask operation request sent by a user at the same time, wherein the mask operation request can comprise a mask erasing request, a mask modifying request, a mask adding request and the like, then carrying out corresponding operation processing on the emphysema candidate segmentation mask according to the mask operation request to obtain an operated emphysema candidate segmentation mask, further carrying out connected domain analysis and binarization processing on the operated emphysema candidate segmentation mask to obtain an emphysema iterative segmentation mask, displaying the emphysema iterative segmentation mask, finally receiving a mask confirming instruction sent by the user, and setting the emphysema iterative segmentation mask as an emphysema focus segmentation mask.
In the above embodiment, the computer device may display the emphysema candidate segmentation mask, so that the user may review the segmented emphysema candidate segmentation mask to determine whether the mask meets the requirement. Then, the computer device can receive various mask operation requests sent by the user through the input device, and perform corresponding operation processing on the emphysema candidate segmentation mask according to the requests, further, the computer device can perform automatic connected domain analysis and binarization processing on the operated emphysema candidate segmentation mask to obtain an emphysema iterative segmentation mask after automatic segmentation, and finally, the computer device can display the emphysema iterative segmentation mask, so that the user can check and modify the segmented emphysema candidate segmentation mask again. The iteration process can be carried out for a plurality of times until a mask confirmation instruction sent by a user is received, and the emphysema iteration segmentation mask obtained by the last segmentation can be set as the emphysema focus segmentation mask. By the method, the emphysema candidate segmentation mask which does not meet the emphysema focus standard can be continuously optimized in an iterative mode until the qualified emphysema segmentation mask is obtained.
It should be noted that, before performing the iterative segmentation, the emphysema candidate segmentation mask may still have a small amount of inaccuracy, for example, the small amount of emphysema candidate segmentation mask may not be a focus of emphysema, or the position of the small amount of emphysema candidate segmentation mask is not accurate enough, and so on. The annotating physician may manually review the emphysema candidate segmentation masks to screen out unsatisfactory emphysema segmentation masks, i.e. masks that are not accurate enough, and optimize for the masks that are not accurate enough, for example, the operations of erasing masks, modifying masks or adding new masks may be performed. Then, automatic segmentation operations such as connected domain analysis and binarization processing can be carried out on the emphysema candidate segmentation mask after manual examination and optimization, the dynamic threshold of each connected domain is recalculated, and finally, the emphysema candidate segmentation mask is re-segmented according to the updated dynamic threshold, so that the optimized lung qi iterative selection segmentation mask can be obtained, and then a labeling doctor can carry out examination and optimization on the optimized emphysema candidate segmentation mask, and the operation is continuously circulated until the qualified emphysema segmentation mask is obtained. The implementation takes the manual review process as a link in a closed loop and interacts with an automatic segmentation algorithm, so that the segmentation accuracy of the emphysema focus can be effectively improved.
In one embodiment, the method for generating the emphysema iterative segmentation mask may include the following steps: firstly, performing connected domain analysis on the operated emphysema candidate segmentation mask by using a connected domain analysis algorithm to obtain a connected domain set comprising a plurality of connected domains, and then obtaining each connected domain in the connected domain set by using a maximum inter-class variance algorithmAnd performing binarization processing on the lung parenchyma image corresponding to each connected domain through a binarization algorithm according to the dynamic threshold of each connected domain to obtain an emphysema segmentation mask of each connected domain, and finally merging the emphysema segmentation masks of all the connected domains to obtain an emphysema iteration segmentation mask. Specifically, the processing method of the emphysema candidate segmentation mask proposed in this embodiment is similar to the processing method of the emphysema initial segmentation mask in the above embodiment, and the difference is mainly that the processing objects are different, the processing objects are the emphysema candidate segmentation mask after the operation of the user, and the generated target is the emphysema iteration segmentation mask. Wherein, the iterative segmentation mask for emphysema can be denoted as mkAnd k is k + 1. In this implementation, the computer device may perform continuous iteration operation according to the emphysema candidate segmentation mask operated by the user until an emphysema iteration segmentation mask meeting the requirement is obtained, and may set the emphysema candidate segmentation mask as the emphysema focus segmentation mask.
In an embodiment, optionally, after step 104, the interactive emphysema lesion segmentation method may further include the following steps: firstly, obtaining a plurality of lung image samples, then performing focus segmentation on each lung image sample by the interactive emphysema focus segmentation method provided in the steps from 101 to 104 to obtain an emphysema focus region image corresponding to each lung image sample, further constructing a deep learning neural network model, taking the plurality of lung image samples as input, taking the emphysema focus region images corresponding to the plurality of lung image samples as output, and training the deep learning neural network model to obtain an emphysema focus segmentation model. Furthermore, after the trained emphysema focus segmentation model is obtained, the lung image to be processed can be processed through the emphysema focus segmentation model to obtain an emphysema focus region image. According to the method, the focus segmentation is carried out on the sample required by model training through the interactive emphysema focus segmentation method, so that the segmentation efficiency and the segmentation accuracy of the lung image sample can be effectively improved, the model training cost is effectively reduced, and the model accuracy is improved.
Further, as a specific implementation of the method shown in fig. 1, this embodiment provides an interactive emphysema lesion segmentation apparatus, as shown in fig. 2, the apparatus includes: the image preprocessing module 21, the image pre-segmentation module 22, the image segmentation module 23, the image iterative segmentation module 24 and the image conversion module 25.
The image preprocessing module 21 is configured to obtain a lung image to be processed, and perform lung parenchymal segmentation processing and filtering processing on the lung image to obtain a lung parenchymal image;
the image pre-segmentation module 22 is configured to pre-segment the lung parenchyma image according to a preset thermal capacity threshold to obtain an emphysema initial segmentation mask;
the image segmentation module 23 is configured to perform connected domain analysis and binarization on the initial segmentation mask for emphysema to obtain an emphysema candidate segmentation mask;
the image iterative segmentation module 24 is configured to perform iterative connected domain analysis and binarization processing on the emphysema candidate segmentation mask in a human-in-circuit interaction manner to obtain an emphysema focus segmentation mask;
the image conversion module 25 may be configured to perform mask operation on the emphysema lesion segmentation mask to obtain an emphysema lesion region image.
In a specific application scenario, the image preprocessing module 21 is specifically configured to perform binarization processing on the lung image through a binarization algorithm to obtain a binarized lung image; performing connected processing on the binary lung image through a connected domain algorithm to obtain a binary lung connected image; performing morphological closing operation and morphological expanding operation on the binary lung connected image to obtain a lung parenchymal mask; and converting the lung parenchymal mask into an initial lung parenchymal image through a mask algorithm, and performing median filtering processing on the initial lung parenchymal image to obtain the lung parenchymal image.
In a specific application scenario, the image preprocessing module 21 is specifically configured to perform morphological closing operation on the binarized pulmonary connected image through a circular structure with a preset radius to obtain a closed binarized pulmonary connected image; detecting the edge of the binaryzation lung communicated image after the closing operation through a Robert operator, and filling a small closed region inside a maximum closed region in the binaryzation lung communicated image after the closing operation into a preset numerical value to obtain a filled binaryzation lung communicated image; and performing morphological expansion operation on the filled binary lung connected image through a circular structure with a preset radius to obtain the lung parenchymal mask.
In a specific application scenario, the image pre-segmentation module 22 may be specifically configured to mark a region of the lung parenchyma image with a heat capacity value smaller than a heat capacity threshold as a first numerical value, and mark a region of the lung parenchyma image with a heat capacity value not smaller than the heat capacity threshold as a second numerical value, so as to obtain an emphysema initial segmentation mask.
In a specific application scenario, the image segmentation module 23 may be specifically configured to perform connected domain analysis on the emphysema initial segmentation mask through a connected domain analysis algorithm to obtain a connected domain set including a plurality of connected domains; aiming at each connected domain in the connected domain set, obtaining a dynamic threshold value of each connected domain through a maximum between-class variance algorithm; according to the dynamic threshold value of each connected domain, carrying out binarization processing on the lung parenchyma image corresponding to each connected domain through a binarization algorithm to obtain an emphysema segmentation mask of each connected domain; and merging the emphysema segmentation masks of all the connected domains to obtain the emphysema candidate segmentation mask.
In a specific application scenario, the image iterative segmentation module 24 may be specifically configured to display an emphysema candidate segmentation mask, and receive a mask operation request sent by a user, where the mask operation request includes a mask erasing request, a mask modifying request, and a mask adding request; according to the mask operation request, performing corresponding operation processing on the emphysema candidate segmentation mask to obtain an operated emphysema candidate segmentation mask; performing connected domain analysis and binarization processing on the operated emphysema candidate segmentation mask to obtain an emphysema iterative segmentation mask, and displaying the emphysema iterative segmentation mask; and receiving a mask confirmation instruction sent by a user, and setting the emphysema iteration segmentation mask as an emphysema focus segmentation mask.
In a specific application scenario, the apparatus further includes a model training module 25, where the model training module 25 is specifically configured to obtain a plurality of lung image samples and emphysema focus region images corresponding to the plurality of lung image samples; and constructing a deep learning neural network model, taking a plurality of lung image samples as input, taking emphysema focus region images corresponding to the lung image samples as output, and training the deep learning neural network model to obtain an emphysema focus segmentation model.
It should be noted that, other corresponding descriptions of the functional units related to the interactive emphysema focus segmentation apparatus provided in this embodiment may refer to the corresponding description in fig. 1, and are not repeated herein.
Based on the method shown in fig. 1, correspondingly, the present embodiment further provides a storage medium, on which a computer program is stored, which when executed by a processor, implements the interactive emphysema lesion segmentation method shown in fig. 1.
Based on such understanding, the technical solution of the present application may be embodied in the form of a software product, and the software product to be identified may be stored in a non-volatile storage medium (which may be a CD-ROM, a usb disk, a removable hard disk, or the like), and include several instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute the method according to the implementation scenarios of the present application.
Based on the method shown in fig. 1 and the interactive emphysema lesion segmentation apparatus embodiment shown in fig. 2, in order to achieve the above object, as shown in fig. 3, this embodiment further provides a computer device capable of implementing the interactive emphysema lesion segmentation method, which may be specifically a personal computer, a server, a smart phone, a tablet computer, a smart watch, or other network devices, and the computer device includes a storage medium and a processor, where the processor is connected to the storage medium through a system bus; a storage medium operable to store a computer program; a processor operable to execute a computer program to implement the method as described above and illustrated in fig. 1.
Optionally, the computer device may further include an internal memory, a network interface, a display screen, an input device, and the like. The input device may include a Keyboard (Keyboard), a mouse, and the like, and the optional user interface may further include a USB interface, a card reader interface, and the like. The network interface may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface), etc.
It will be understood by those skilled in the art that the present embodiment provides a computer apparatus structure for interactive emphysema lesion segmentation, which does not constitute a limitation of the computer apparatus, and may include more or less components, or combine some components, or arrange different components.
The storage medium may further include an operating system and a network communication module. The operating system is a program that manages the hardware of the above-described computer device and the software resources to be identified, and supports the execution of the information processing program and other software and/or programs to be identified. The network communication module is used for realizing communication among components in the storage medium and communication with other hardware and software in the information processing computer equipment.
Through the above description of the embodiments, those skilled in the art will clearly understand that the present application can be implemented by software plus a necessary general hardware platform, and can also be implemented by hardware. By applying the technical scheme of the application, the lung parenchymal segmentation processing and the filtering processing are firstly carried out on the lung image to obtain the lung parenchymal image, then the lung parenchymal image is pre-segmented to reduce the range of focus segmentation, then the emphysema candidate segmentation mask is generated through automatic segmentation algorithms such as connected domain analysis and binarization processing, finally the emphysema candidate segmentation mask is continuously and iteratively repaired through an automatic segmentation algorithm in a human-in-loop interaction mode until the qualified emphysema focus segmentation mask is obtained, and the emphysema focus segmentation mask is converted into the emphysema focus region image. Compared with the prior art, the method does not need to realize automatic segmentation of the emphysema focus region by means of a data model based on a deep learning algorithm and the like, and effectively reduces the cost of segmenting the pulmonary tuberculosis focus. In addition, by means of an interaction mode of a human in a loop, an automatic segmentation algorithm of connected domain analysis, binarization processing and the like, a labeling doctor does not need to label the emphysema nidus from zero, the workload of the labeling doctor is greatly reduced, the time for labeling the emphysema nidus is shortened, the efficiency of segmenting and labeling the emphysema nidus is improved, and the missing label error rate of the emphysema positive nidus is effectively reduced.
Those skilled in the art will appreciate that the figures are merely schematic representations of one preferred implementation scenario and that the blocks or flow diagrams in the figures are not necessarily required to practice the present application. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario may be distributed in the devices in the implementation scenario according to the description of the implementation scenario, or may be located in one or more devices different from the present implementation scenario with corresponding changes. The modules of the implementation scenario may be combined into one module, or may be further split into a plurality of sub-modules.
The above application serial numbers are for description purposes only and do not represent the superiority or inferiority of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of the present application, but the present application is not limited thereto, and any variations that can be made by those skilled in the art are intended to fall within the scope of the present application.
Claims (10)
1. An interactive emphysema focus segmentation method, characterized in that the method comprises:
acquiring a lung image to be processed, and performing lung parenchyma segmentation processing and filtering processing on the lung image to obtain a lung parenchyma image;
according to a preset heat capacity threshold value, pre-dividing the lung parenchyma image to obtain an emphysema initial division mask;
performing connected domain analysis and binarization processing on the emphysema initial segmentation mask to obtain an emphysema candidate segmentation mask;
performing iterative connected domain analysis and binarization processing on the emphysema candidate segmentation mask by a human-in-loop interaction mode to obtain an emphysema focus segmentation mask;
and performing mask operation on the emphysema focus segmentation mask to obtain an emphysema focus region image.
2. The method according to claim 1, wherein the performing lung parenchymal segmentation processing and filtering processing on the lung image to obtain a lung parenchymal image comprises:
carrying out binarization processing on the lung image through a binarization algorithm to obtain a binarized lung image;
performing connected processing on the binarized lung image through a connected domain algorithm to obtain a binarized lung connected image;
performing morphological closing operation and morphological expanding operation on the binary lung connected image to obtain a lung parenchymal mask;
and converting the lung parenchymal mask into an initial lung parenchymal image through a mask algorithm, and performing median filtering processing on the initial lung parenchymal image to obtain the lung parenchymal image.
3. The method of claim 2, wherein said performing a morphological closing operation and a morphological dilation operation on said binarized pulmonary connectivity image to obtain a pulmonary parenchymal mask comprises:
performing morphological closing operation on the binary lung communicated image through a circular structure with a preset radius to obtain a closed-operation binary lung communicated image;
detecting the edge of the binaryzation lung communicated image after the closing operation through a Robert operator, and filling a small closed region inside a maximum closed region in the binaryzation lung communicated image after the closing operation into a preset numerical value to obtain a filled binaryzation lung communicated image;
and performing morphological expansion operation on the filled binary lung connected image through the circular structure with the preset radius to obtain the lung parenchyma mask.
4. The method according to claim 1, wherein the pre-segmenting the lung parenchymal image according to a preset heat capacity threshold to obtain an emphysema initial segmentation mask comprises:
and marking the area with the heat capacity value smaller than the heat capacity threshold value in the lung parenchyma image as a first numerical value, and marking the area with the heat capacity value not smaller than the heat capacity threshold value in the lung parenchyma image as a second numerical value to obtain an emphysema initial segmentation mask.
5. The method of claim 1, wherein the performing connected component analysis and binarization on the emphysema initial segmentation mask to obtain an emphysema candidate segmentation mask comprises:
performing connected domain analysis on the emphysema initial segmentation mask through a connected domain analysis algorithm to obtain a connected domain set comprising a plurality of connected domains;
aiming at each connected domain in the connected domain set, obtaining a dynamic threshold value of each connected domain through a maximum inter-class variance algorithm;
according to the dynamic threshold value of each connected domain, carrying out binarization processing on the lung parenchyma image corresponding to each connected domain through a binarization algorithm to obtain an emphysema segmentation mask of each connected domain;
and merging the emphysema segmentation masks of all the connected domains to obtain the emphysema candidate segmentation mask.
6. The method of claim 1, wherein the iteratively analyzing a connected domain and binarizing the candidate segmentation mask for emphysema to obtain an emphysema lesion segmentation mask through human-in-circuit interaction comprises:
displaying the emphysema candidate segmentation mask, and receiving a mask operation request sent by a user, wherein the mask operation request comprises an erasing mask request, a modifying mask request and an adding mask request;
according to the mask operation request, performing corresponding operation processing on the emphysema candidate segmentation mask to obtain an operated emphysema candidate segmentation mask;
performing connected domain analysis and binarization processing on the operated emphysema candidate segmentation mask to obtain an emphysema iterative segmentation mask, and displaying the emphysema iterative segmentation mask;
and receiving a mask confirmation instruction sent by a user, and setting the emphysema iteration segmentation mask as an emphysema focus segmentation mask.
7. The method of claim 1, further comprising:
acquiring a plurality of lung image samples and emphysema focus area images corresponding to the lung image samples;
and constructing a deep learning neural network model, taking the lung image samples as input, taking emphysema focus region images corresponding to the lung image samples as output, and training the deep learning neural network model to obtain an emphysema focus segmentation model.
8. An interactive emphysema lesion segmentation apparatus, the apparatus comprising:
the image preprocessing module is used for acquiring a lung image to be processed, and performing lung parenchyma segmentation processing and filtering processing on the lung image to obtain a lung parenchyma image;
the image pre-segmentation module is used for pre-segmenting the lung parenchyma image according to a preset heat capacity threshold value to obtain an emphysema initial segmentation mask;
the image segmentation module is used for carrying out connected domain analysis and binarization processing on the emphysema initial segmentation mask to obtain an emphysema candidate segmentation mask;
the image iteration segmentation module is used for carrying out iterative connected domain analysis and binarization processing on the emphysema candidate segmentation mask to obtain an emphysema focus segmentation mask in a human-in-loop interaction mode;
and the image conversion module is used for carrying out mask operation on the emphysema focus segmentation mask to obtain an emphysema focus region image.
9. A storage medium having a computer program stored thereon, the computer program, when being executed by a processor, realizing the steps of the method of any one of claims 1 to 7.
10. A computer arrangement comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 7 when executed by the processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111017521.4A CN113658198A (en) | 2021-08-31 | 2021-08-31 | Interactive emphysema focus segmentation method, device, storage medium and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111017521.4A CN113658198A (en) | 2021-08-31 | 2021-08-31 | Interactive emphysema focus segmentation method, device, storage medium and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113658198A true CN113658198A (en) | 2021-11-16 |
Family
ID=78482656
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111017521.4A Pending CN113658198A (en) | 2021-08-31 | 2021-08-31 | Interactive emphysema focus segmentation method, device, storage medium and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113658198A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115100179A (en) * | 2022-07-15 | 2022-09-23 | 北京医准智能科技有限公司 | Image processing method, device, equipment and storage medium |
-
2021
- 2021-08-31 CN CN202111017521.4A patent/CN113658198A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115100179A (en) * | 2022-07-15 | 2022-09-23 | 北京医准智能科技有限公司 | Image processing method, device, equipment and storage medium |
| CN115100179B (en) * | 2022-07-15 | 2023-02-21 | 北京医准智能科技有限公司 | Image processing method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7383010B2 (en) | Medical image recognition method and system, model training method, computer device, and program | |
| CN107798682B (en) | Image segmentation system, method, apparatus and computer-readable storage medium | |
| CN107622492B (en) | Lung fissure segmentation method and system | |
| JP2008520345A (en) | Method and system for detecting and classifying lesions in ultrasound images | |
| CN111145160B (en) | Method, device, server and medium for determining coronary artery branches where calcified regions are located | |
| CN111242956A (en) | U-Net-based ultrasonic fetal heart and fetal lung deep learning joint segmentation method | |
| CN112861961B (en) | Pulmonary blood vessel classification method and device, storage medium and electronic equipment | |
| CN111415335B (en) | Blood vessel labeling method and device | |
| CN114092450A (en) | Real-time image segmentation method, system and device based on gastroscopy video | |
| CN111724389B (en) | Method, device, storage medium and computer equipment for segmenting CT image of hip joint | |
| CN113160153A (en) | Lung nodule screening method and system based on deep learning technology | |
| CN102419864B (en) | Method and device for extracting skeletons of brain CT (computerized tomography) image | |
| TW201726064A (en) | Medical image processing apparatus and breast image processing method thereof | |
| CN113888566B (en) | Target contour curve determination method and device, electronic equipment and storage medium | |
| CN113658198A (en) | Interactive emphysema focus segmentation method, device, storage medium and equipment | |
| CN113160248B (en) | Image processing method, device and equipment and readable storage medium | |
| CN116309647B (en) | Method for constructing craniocerebral lesion image segmentation model, image segmentation method and device | |
| CN111627023B (en) | Method and device for generating coronary artery projection image and computer readable medium | |
| CN111918611B (en) | Method for controlling abnormal display of chest X-ray image, recording medium and apparatus | |
| CN112085695A (en) | Image processing method, device and storage medium | |
| CN115294401B (en) | CT image-based classification method, device, equipment and storage medium | |
| CN111325758A (en) | Lung image segmentation method and device and training method of image segmentation model | |
| CN111429457A (en) | Intelligent evaluation method, device, equipment and medium for brightness of local area of image | |
| CN110910409A (en) | Gray scale image processing method and device and computer readable storage medium | |
| CN113889238B (en) | Image identification method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20220524 Address after: 518000 China Aviation Center 2901, No. 1018, Huafu Road, Huahang community, Huaqiang North Street, Futian District, Shenzhen, Guangdong Province Applicant after: Shenzhen Ping An medical and Health Technology Service Co.,Ltd. Address before: Room 12G, Area H, 666 Beijing East Road, Huangpu District, Shanghai 200001 Applicant before: PING AN MEDICAL AND HEALTHCARE MANAGEMENT Co.,Ltd. |