CN111080594A - Human body part recognition method, computer device and readable storage medium - Google Patents
Human body part recognition method, computer device and readable storage medium Download PDFInfo
- Publication number
- CN111080594A CN111080594A CN201911249073.3A CN201911249073A CN111080594A CN 111080594 A CN111080594 A CN 111080594A CN 201911249073 A CN201911249073 A CN 201911249073A CN 111080594 A CN111080594 A CN 111080594A
- Authority
- CN
- China
- Prior art keywords
- medical image
- human body
- network model
- body part
- regression network
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 230000006870 function Effects 0.000 claims description 85
- 238000012549 training Methods 0.000 claims description 31
- 238000004590 computer program Methods 0.000 claims description 15
- 230000000694 effects Effects 0.000 description 8
- 210000004072 lung Anatomy 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 7
- 210000004061 pubic symphysis Anatomy 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 238000004195 computer-aided diagnosis Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000002059 diagnostic imaging Methods 0.000 description 3
- 239000003550 marker Substances 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 210000003689 pubic bone Anatomy 0.000 description 3
- 210000000683 abdominal cavity Anatomy 0.000 description 2
- 230000003187 abdominal effect Effects 0.000 description 2
- 210000000746 body region Anatomy 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 210000003625 skull Anatomy 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 210000001015 abdomen Anatomy 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 210000004197 pelvis Anatomy 0.000 description 1
- 230000002685 pulmonary effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
Abstract
The invention relates to a human body part identification method, computer equipment and a readable storage medium, wherein the method comprises the following steps: acquiring a medical image to be analyzed; inputting the medical image into a classification network model, and determining a human body part to which the medical image belongs; and inputting the medical image into the regression network model corresponding to the human body part to obtain the mark corresponding to the human body part in the medical image. In the method, the computer equipment can determine the human body part to which the medical image belongs through the classification network model, then the medical image is input into the regression network model corresponding to the determined human body part to which the medical image belongs, the human body part in the medical image can be accurately marked aiming at the human body part to which the medical image belongs, and the accuracy of the marking corresponding to the human body part in the obtained medical image is improved.
Description
Technical Field
The present invention relates to the field of medical images, and in particular, to a human body part recognition method, a computer device, and a readable storage medium.
Background
With the development of computer-aided diagnosis technology, the computer-aided diagnosis technology can assist in finding out the focus by combining with the analysis and calculation of a computer through the imaging, medical image processing technology and other possible physiological and biochemical means. Before the existing computer-aided diagnosis technology is used, the human body part covered by the input medical image is judged, and then the corresponding computer-aided diagnosis technology is called according to the human body part covered by the determined medical image to diagnose the medical image, so that a more accurate diagnosis result can be obtained, and therefore, the judgment of the human body part covered by the medical image is particularly important.
In the traditional technology, labels are respectively distributed on the head and neck, the chest, the lung and the abdominal cavity of a human body in a piecewise linear mode according to a preselected key point, the labels and medical images are input into a neural network to be trained, a pre-trained regression network is obtained, the regression network is used for determining the human body part covered by the medical images, and the human body part covered in the medical images is identified.
However, the conventional method for recognizing human body parts has a problem of low recognition accuracy.
Disclosure of Invention
In view of the above, it is necessary to provide a human body part recognition method, a computer device, and a readable storage medium, in order to solve the problem of low recognition accuracy of the conventional human body part recognition method.
In a first aspect, an embodiment of the present invention provides a human body part identification method, where the method includes:
acquiring a medical image to be analyzed;
inputting the medical image into a classification network model, and determining a human body part to which the medical image belongs;
and inputting the medical image into the regression network model corresponding to the human body part to obtain a mark corresponding to the human body part in the medical image.
In one embodiment, the regression network model is labeled with a label of a corresponding human body part in advance, and before the medical image is input into the regression network model corresponding to the human body part and the label corresponding to the human body part in the medical image is obtained, the method further includes:
and determining a regression network model corresponding to the human body part according to the corresponding relation between the human body part to which the medical image belongs and the label of the regression network model.
In one embodiment, the training process of the regression network model includes:
acquiring a first sample medical image;
inputting the first sample medical image into a corresponding preset initial regression network model to obtain a sample mark of a human body part in the first sample medical image;
obtaining a value of a loss function of the initial regression network model according to the sample mark and the mark of the human body part in the first sample medical image in advance;
and training the initial regression network model by using the value of the loss function of the initial regression network model to obtain the regression network model.
In one embodiment, the training the initial regression network model by using the value of the loss function of the initial regression network model to obtain the regression model includes:
obtaining a value of a Gaussian kernel weighted loss function of the initial regression network model according to the value of the loss function of the initial regression network model;
and training the initial regression network model by using the value of the Gaussian kernel weighted loss function to obtain the regression model.
In one embodiment, the gaussian kernel weighted loss functions of the regression network model are different for different human body parts.
In one embodiment, the loss function of the initial regression network model comprises any one of the following functions: a mean square error function; mean absolute error function.
In one embodiment, before obtaining the value of the loss function of the initial regression network model according to the sample label and the label of the human body part in the first sample medical image in advance, the method further includes:
and according to a preset division rule, dividing the human body part to obtain the mark of the human body part in the first sample medical image in advance.
In one embodiment, the training process of the classification network model includes:
acquiring a second sample medical image;
inputting the second sample medical image into a preset initial classification network model, and determining a human body part to which the second sample medical image belongs;
obtaining a loss function value of the initial classification network model according to the human body part to which the second sample medical image belongs and a marking result of the second sample medical image in advance;
and training the initial classification network model according to the value of the loss function of the initial classification network model to obtain the classification network model.
In a second aspect, an embodiment of the present invention provides a human body part identification apparatus, where the apparatus includes:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring a medical image to be analyzed;
the first determining module is used for inputting the medical image into a classification network model and determining a human body part to which the medical image belongs;
and the identification module is used for inputting the medical image into the regression network model corresponding to the human body part to obtain the mark corresponding to the human body part in the medical image.
In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and the processor implements the following steps when executing the computer program:
acquiring a medical image to be analyzed;
inputting the medical image into a classification network model, and determining a human body part to which the medical image belongs;
and inputting the medical image into the regression network model corresponding to the human body part to obtain a mark corresponding to the human body part in the medical image.
In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the following steps:
acquiring a medical image to be analyzed;
inputting the medical image into a classification network model, and determining a human body part to which the medical image belongs;
and inputting the medical image into the regression network model corresponding to the human body part to obtain a mark corresponding to the human body part in the medical image.
In the human body part method, the human body part device, the computer device, and the readable storage medium provided in the above embodiments, the computer device obtains the medical image, inputs the medical image into the classification network model, determines the human body part to which the medical image belongs, and inputs the medical image into the regression network model corresponding to the determined human body part, thereby obtaining the marker corresponding to the human body part in the medical image. In the method, the computer equipment can determine the human body part to which the medical image belongs through the classification network model, then the medical image is input into the regression network model corresponding to the determined human body part to which the medical image belongs, the human body part in the medical image can be accurately marked aiming at the human body part to which the medical image belongs, and the accuracy of the marking corresponding to the human body part in the obtained medical image is improved.
Drawings
FIG. 1 is a schematic diagram of an internal structure of a computer device according to an embodiment;
FIG. 2 is a schematic flow chart of a human body part recognition method according to an embodiment;
FIG. 2(a) is a schematic diagram of human body part division according to an embodiment;
fig. 3 is a schematic flow chart of a human body part identification method according to another embodiment;
fig. 4 is a schematic flowchart of a human body part identification method according to another embodiment;
fig. 5 is a schematic structural diagram of a human body part recognition apparatus according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The human body part identification method provided by the embodiment of the application can be applied to computer equipment shown in figure 1. The computer device comprises a processor and a memory connected by a system bus, wherein a computer program is stored in the memory, and the steps of the method embodiments described below can be executed when the processor executes the computer program. Optionally, the computer device may further comprise a network interface, a display screen and an input device. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a nonvolatile storage medium storing an operating system and a computer program, and an internal memory. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. Optionally, the computer device may be a server, a personal computer, a personal digital assistant, other terminal devices such as a tablet computer, a mobile phone, and the like, or a cloud or a remote server, and the specific form of the computer device is not limited in the embodiment of the present application.
In a traditional human body part automatic identification task, labels are respectively and linearly distributed on the head and neck, the chest, the lung and the abdominal cavity of a human body in a segmented mode according to a preselected key point, then the labels and images are sent to a neural network for training, the reliability of the labels corresponding to each image slice is inconsistent in the process, the reliability is lower when the labels are farther away from the key point, otherwise, the reliability is higher when the labels are closer to the key point, and therefore the labels corresponding to the human body parts in a medical image cannot be accurately marked through the trained neural network. Therefore, embodiments of the present invention provide a human body part identification method, a computer device, and a storage medium, which aim to solve the above technical problems of the conventional technology.
The following describes the technical solution of the present invention and how to solve the above technical problems with specific examples. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments.
Fig. 2 is a schematic flow chart of a human body part identification method according to an embodiment. Fig. 2(a) is a schematic diagram of human body part division according to an embodiment. The embodiment relates to a specific implementation process of inputting a medical image into a classification network model by computer equipment, determining a human body part to which the medical image belongs, inputting the medical image into a regression network model corresponding to the determined human body part, and obtaining a mark corresponding to the human body part in the medical image. As shown in fig. 2, the method may include:
s201, acquiring a medical image to be analyzed.
Wherein the medical image to be analyzed is a two-dimensional slice image or a 2.5-dimensional slice image. Alternatively, the computer device may acquire the medical image to be analyzed from a PACS (Picture Archiving and Communication Systems) server, or may acquire the medical image to be analyzed from a medical imaging device in real time. Alternatively, the medical image may be a cross-sectional image of the human body.
S202, inputting the medical image into the classification network model, and determining the human body part to which the medical image belongs.
Specifically, the computer device inputs the medical image into a classification network model, and determines the human body part to which the medical image belongs. Alternatively, as shown in fig. 2(a), the human body part to which the medical image belongs may belong to at least one of the following parts: above the head and neck, below the head and neck, pectoral and pulmonary, abdominopelvic and pubic symphysis. Optionally, before the computer device inputs the medical image into the classification network model, normalization processing, clipping processing, and the like may be performed on the medical image, and the processed medical image is input into the classification network model.
S203, inputting the medical image into the regression network model corresponding to the human body part to obtain the mark corresponding to the human body part in the medical image.
Specifically, the computer device determines a regression network model corresponding to the human body part according to the determined human body part to which the medical image belongs, inputs the medical image into the regression network model corresponding to the human body part to which the medical image belongs, and obtains a mark corresponding to the human body part in the medical image. It should be noted that, the classification network model inputs slice images, when it is necessary to determine the marks corresponding to the human body parts to which the complete 3D volume data belongs, each layer of medical images of the human body may be input into the classification network model, the human body part to which each layer of medical images belongs is determined, then the regression network model corresponding to the human body part to which each layer of medical images belongs is determined, each layer of medical images is input into the corresponding regression network model, the marks corresponding to the human body parts in each layer of medical images are obtained, and the marks corresponding to the human body parts in each layer of medical images are post-processed, so that the marks corresponding to the human body parts to which the complete 3D volume data belongs are obtained.
In this embodiment, the computer device can determine the human body part to which the medical image belongs through the classification network model, and then input the medical image into the regression network model corresponding to the determined human body part to which the medical image belongs, so that the human body part in the medical image can be accurately marked according to the human body part to which the medical image belongs, and the accuracy of the marking corresponding to the human body part in the obtained medical image is improved.
The regression model is labeled with a label of a corresponding human body part in advance, and as an optional implementation manner on the basis of the foregoing embodiment, before S203, the method further includes: and determining the regression network model corresponding to the human body part according to the corresponding relation between the human body part to which the medical image belongs and the label of the regression network model.
The regression model is labeled with labels of corresponding human body parts in advance, illustratively, the labels of the regression network model labels corresponding to the head and neck parts are the head and neck parts, and the labels of the regression network model labels corresponding to other parts of the human body are analogized in sequence. Specifically, after the computer device determines the human body part to which the medical image belongs, the computer device determines a regression network model corresponding to the human body part to which the medical image belongs according to the correspondence between the human body part to which the medical image belongs and the label of the regression model, inputs the medical image into the regression network model corresponding to the human body part, and obtains the mark corresponding to the human body part in the medical image.
In this embodiment, the computer device can quickly and accurately determine the regression network model corresponding to the human body part according to the correspondence between the human body part to which the medical image belongs and the label of the regression model, and then can input the medical image into the regression network model corresponding to the human body part to which the medical image belongs, so as to quickly and accurately obtain the mark corresponding to the human body part in the medical image, thereby improving the accuracy and efficiency of obtaining the mark corresponding to the human body part in the medical image.
Fig. 3 is a schematic flow chart of a human body part identification method according to another embodiment. The embodiment relates to a specific implementation process of training a regression network model by computer equipment. As shown in fig. 3, on the basis of the foregoing embodiment, as an alternative implementation, the training process of the regression network model may include:
s301, a first sample medical image is acquired.
Alternatively, the computer device may acquire the first sample medical image from a PACS (Picture Archiving and communications systems) server, or may acquire the first sample medical image from a medical imaging device in real time. It should be noted that the acquired first sample medical image already marks the body part to which the first sample medical image belongs.
S302, inputting the first sample medical image into a corresponding preset initial regression network model to obtain a sample mark of the human body part in the first sample medical image.
Specifically, the computer device inputs the first sample medical image into a preset initial regression network model corresponding to the human body part to which the first sample medical image belongs, and obtains a sample mark of the human body part in the first sample medical image. For example, when the first sample medical image is a head and neck image, the preset regression network model corresponding to the first sample medical image is a head and neck regression network model, and the computer device inputs the first sample medical image into the corresponding initial head and neck regression network model to obtain a sample label of the human body part in the first sample medical image as a sample head and neck label.
S303, obtaining the value of the loss function of the initial regression network model according to the sample mark and the mark of the human body part in the first sample medical image in advance.
Specifically, the computer device obtains the value of the loss function of the initial regression network model according to the obtained sample mark and the mark of the human body part in the first sample medical image in advance. Optionally, the loss function of the initial regression network model includes any one of the following functions: a mean square error function; mean absolute error function. Optionally, the loss function of the initial regression network model may further include a deformation function of a mean square error function, or a deformation function of a mean absolute error function, for example, a smoothed mean absolute error function. It should be noted that the loss function of the initial regression network model may be a loss function suitable for regression, and is not limited to the above description of the loss function of the initial regression network model, and for example, a Huber loss function may be used as the loss function of the initial regression network model.
S304, training the initial regression network model by using the value of the loss function of the initial regression network model to obtain the regression network model.
Specifically, the computer device trains the initial regression network model by using the value of the loss function of the initial regression network model, and the corresponding initial regression network model when the value of the loss function of the initial regression network model reaches a stable valueThe network model is determined as the regression network model described above. Optionally, the computer device may obtain a value of a gaussian kernel weighted loss function of the initial regression network model according to the value of the loss function of the initial regression network model, and train the initial regression network model by using the value of the gaussian kernel weighted loss function to obtain the regression model, where the gaussian kernel weighted loss functions of the regression network models corresponding to different human body parts are different. The following description will be made by taking the divided human body parts above the head and neck, pectoral-pulmonary, abdominopelvic and pubic symphysis as shown in fig. 2(a) as an example: the label at the top of the skull is L1And the label of the center of the C7 vertebral body is L2And the label of the center of the T12 vertebral body is L3The label of the pubic symphysis is L4If the human body part to which the first sample medical image belongs is above the head and neck, the value of the gaussian kernel weighted loss function of the initial regression network model corresponding to the first sample medical image is
Wherein t ∈ Range0，Range0Representing the head and neck, where σ is a constant for determining the width of the Gaussian kernel weighted loss function, t is a pre-labeled human body region in the first sample medical image, x is a sample label of the obtained human body region in the first sample medical image, floss(x, t) is the value of the loss function of the initial regression network model corresponding to the human body part belonging to the first sample medical image above the head and neck, and is used for estimating the degree of inconsistency between the sample mark x of the human body part in the first sample medical image and the mark t of the human body part in the first sample medical image in advance, and so on, when the human body part belonging to the first sample medical image is any one of the head and neck, the chest and the lung and the abdominal and pelvic cavity, the value of the Gaussian kernel weighted loss function of the initial regression network model corresponding to the first sample medical image is
Wherein t ∈ Range1，Range1Indicates head, neck, chest and lung(ii) an arbitrary part in the pelvic region or abdomen region, wherein σ is a constant for determining the width of the Gaussian kernel weighted loss function, t is a mark of the human body part in the first sample medical image in advance, x is a sample mark of the human body part in the obtained first sample medical image, and floss(x, t) is the value of the loss function of the corresponding initial regression network model when the part of the human body to which the first sample medical image belongs is the head and neck part, the chest lung part and the abdominal and pelvic cavity part; when the human body part to which the first sample medical image belongs is below the pubic symphysis, the value of the Gaussian kernel weighted loss function of the initial regression network model corresponding to the first sample medical image is
Wherein t ∈ Range2，Range2Representing the pubic symphysis, where σ is a constant for determining the width of the Gaussian kernel weighted loss function, t is a pre-determined marker of the region of the body in the first medical image, x is a sample marker of the region of the body in the first medical image obtained, f is a measure of the distance between the first and second points, f is a measure of thelossAnd (x, t) is the value of the loss function of the corresponding initial regression network model when the human body part to which the first sample medical image belongs is below the pubic symphysis.
In this embodiment, the computer device inputs the first medical image into the corresponding pre-set initial regression network model to obtain the sample markers of the human body parts in the first medical image, the method comprises the steps of obtaining a value of a loss function of an initial regression network model according to a sample mark and a mark of a human body part in a first sample medical image in advance, accurately training the initial regression network model by using the value of the loss function of the initial regression network model, improving accuracy of the obtained regression network model, and being better.
On the basis of the foregoing embodiment, as an optional implementation manner, before the foregoing S303, the method further includes: and according to a preset division rule, dividing the human body part to obtain a mark of the human body part in the first sample medical image in advance.
Specifically, the computer device divides the human body part according to a preset division rule to obtain a mark of the human body part in the first sample medical image in advance. Alternatively, the computer device may follow pre-selected key points as shown in fig. 2 (a): the method comprises the steps of combining the skull top, the C7 centrum, the T12 centrum center and the pubis, dividing a first sample medical image into a part above the head and neck, a part below the head and neck, a chest and lung part, an abdominopelvic cavity and a pubis, obtaining marks of human body parts in the first sample medical image in advance, and obtaining marks above the head and neck, marks of the chest and lung part, marks of the abdominopelvic cavity and marks below the pubis.
In this embodiment, the computer device divides the human body part according to the preset division rule, and can quickly obtain the mark of the human body part in the first sample medical image in advance, so that the efficiency of obtaining the loss function of the initial regression network model according to the sample mark and the mark of the human body part in the first sample medical image in advance is improved.
Fig. 4 is a schematic flowchart of a human body part identification method according to another embodiment. The embodiment relates to a specific implementation process for training a classification network model by computer equipment. As shown in fig. 4, on the basis of the foregoing embodiment, as an alternative implementation, the training process of the classification network model may include:
s401, a second sample medical image is obtained.
Alternatively, the computer device may acquire the second sample medical image from a PACS (Picture Archiving and communications systems) server, or may acquire the second sample medical image from a medical imaging device in real time.
S402, inputting the second sample medical image into a preset initial classification network model, and determining the human body part to which the second sample medical image belongs.
Specifically, the computer device inputs the second sample medical image into a preset initial classification network model, and determines the human body part to which the second sample medical image belongs. Optionally, the part of the human body to which the second sample medical image belongs may be above the head and neck, or may be the head and neck, or the chest and lung, or the abdominopelvic cavity, or below the pubic symphysis.
And S403, obtaining a loss function value of the initial classification network model according to the human body part to which the second sample medical image belongs and the marking result of the second sample medical image in advance.
Specifically, the computer device obtains a value of a loss function of the initial classification network model according to the obtained human body part to which the second sample medical image belongs and a result of marking the second sample medical image in advance. And the value of the loss function of the initial classification network model is used for measuring the difference between the obtained human body part to which the second sample medical image belongs and the labeling result of the second sample medical image in advance.
S404, training the initial classification network model according to the loss function value of the initial classification network model to obtain a classification network model.
Specifically, the computer device trains the initial classification network model according to the value of the loss function of the initial classification network model, and determines the corresponding initial classification network model as the classification network model when the value of the loss function of the initial classification network model reaches a stable value.
In this embodiment, the computer device inputs the second sample medical image into a preset initial classification network model, determines a human body part to which the second sample medical image belongs, obtains a value of a loss function of the initial classification network model according to the human body part to which the second sample medical image belongs and a result of labeling the second sample medical image in advance, and can perform relatively accurate training on the initial classification network model according to the value of the loss function of the initial classification network model, thereby improving the accuracy of the obtained classification network model.
It should be understood that although the various steps in the flow charts of fig. 2-4 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2-4 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternating with other steps or at least some of the sub-steps or stages of other steps.
Fig. 5 is a schematic structural diagram of a human body part recognition apparatus according to an embodiment. As shown in fig. 5, the apparatus may include: a first obtaining module 10, a first determining module 11 and an identifying module 12.
Specifically, the first obtaining module 10 is configured to obtain a medical image to be analyzed;
the first determining module 11 is configured to input the medical image into the classification network model, and determine a human body part to which the medical image belongs;
the recognition module 12 is configured to input the medical image into the regression network model corresponding to the human body part to obtain a mark corresponding to the human body part in the medical image.
The human body part recognition device provided in this embodiment may implement the above method embodiments, and the implementation principle and technical effect are similar, which are not described herein again.
On the basis of the above embodiment, the regression network model is labeled with a label of a corresponding human body part in advance, and optionally, the apparatus further includes: a second determination module.
Specifically, the second determining module is configured to determine the regression network model corresponding to the human body part according to a correspondence between the human body part to which the medical image belongs and the label of the regression network model.
The human body part recognition device provided in this embodiment may implement the above method embodiments, and the implementation principle and technical effect are similar, which are not described herein again.
On the basis of the foregoing embodiment, optionally, the apparatus further includes: the device comprises a second acquisition module, a third acquisition module, a fourth acquisition module and a first training module.
Specifically, the second acquisition module is used for acquiring a first sample medical image;
the third acquisition module is used for inputting the first sample medical image into a corresponding preset initial regression network model to obtain a sample mark of a human body part in the first sample medical image;
the fourth acquisition module is used for acquiring a loss function value of the initial regression network model according to the sample mark and the mark of the human body part in the first sample medical image in advance;
and the first training module is used for training the initial regression network model by using the value of the loss function of the initial regression network model to obtain the regression network model.
Optionally, the loss function of the initial regression network model is any one of the following functions: a mean square error function; an average absolute value error function; a smoothed average absolute error function.
The human body part recognition device provided in this embodiment may implement the above method embodiments, and the implementation principle and technical effect are similar, which are not described herein again.
On the basis of the foregoing embodiment, optionally, the first training module includes an obtaining unit and a training unit.
Specifically, the obtaining unit is configured to obtain a value of a gaussian kernel weighted loss function of the initial regression network model according to the value of the loss function of the initial regression network model;
and the training unit is used for training the initial regression network model by utilizing the value of the Gaussian kernel weighted loss function to obtain the regression model.
Wherein, the Gaussian kernel weighted loss functions of the regression network models corresponding to different human body parts are different.
The human body part recognition device provided in this embodiment may implement the above method embodiments, and the implementation principle and technical effect are similar, which are not described herein again.
On the basis of the foregoing embodiment, optionally, the apparatus further includes a fifth obtaining module.
Specifically, the fifth obtaining module is configured to divide the human body part according to a preset division rule, so as to obtain a mark of the human body part in the first sample medical image in advance.
The human body part recognition device provided in this embodiment may implement the above method embodiments, and the implementation principle and technical effect are similar, which are not described herein again.
On the basis of the foregoing embodiment, optionally, the apparatus includes a sixth obtaining module, a third determining module, a seventh obtaining module, and a second training module.
Specifically, the sixth acquiring module is configured to acquire a second sample medical image;
the third determining module is used for inputting the second sample medical image into a preset initial classification network model and determining the human body part to which the second sample medical image belongs;
the seventh obtaining module is used for obtaining the value of the loss function of the initial classification network model according to the human body part to which the second sample medical image belongs and the marking result of the second sample medical image in advance;
and the second training module is used for training the initial classification network model according to the value of the loss function of the initial classification network model to obtain the classification network model.
The human body part recognition device provided in this embodiment may implement the above method embodiments, and the implementation principle and technical effect are similar, which are not described herein again.
For specific definition of the human body part recognition device, reference may be made to the above definition of the human body part recognition method, which is not described herein again. All or part of the modules in the human body part recognition device can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring a medical image to be analyzed;
inputting the medical image into a classification network model, and determining a human body part to which the medical image belongs;
and inputting the medical image into the regression network model corresponding to the human body part to obtain the mark corresponding to the human body part in the medical image.
The implementation principle and technical effect of the computer device provided by the above embodiment are similar to those of the above method embodiment, and are not described herein again.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring a medical image to be analyzed;
inputting the medical image into a classification network model, and determining a human body part to which the medical image belongs;
and inputting the medical image into the regression network model corresponding to the human body part to obtain the mark corresponding to the human body part in the medical image.
The implementation principle and technical effect of the computer-readable storage medium provided by the above embodiments are similar to those of the above method embodiments, and are not described herein again.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method for identifying a human body part, the method comprising:
acquiring a medical image to be analyzed;
inputting the medical image into a classification network model, and determining a human body part to which the medical image belongs;
and inputting the medical image into the regression network model corresponding to the human body part to obtain a mark corresponding to the human body part in the medical image.
2. The method according to claim 1, wherein the regression network model is labeled with labels of corresponding human body parts in advance, and before the medical image is input into the regression network model corresponding to the human body parts and labels corresponding to the human body parts in the medical image are obtained, the method further comprises:
and determining a regression network model corresponding to the human body part according to the corresponding relation between the human body part to which the medical image belongs and the label of the regression network model.
3. The method of claim 1, wherein the training process of the regression network model comprises:
acquiring a first sample medical image;
inputting the first sample medical image into a corresponding preset initial regression network model to obtain a sample mark of a human body part in the first sample medical image;
obtaining a value of a loss function of the initial regression network model according to the sample mark and the mark of the human body part in the first sample medical image in advance;
and training the initial regression network model by using the value of the loss function of the initial regression network model to obtain the regression network model.
4. The method of claim 3, wherein the training the initial regression network model using the value of the loss function of the initial regression network model to obtain the regression model comprises:
obtaining a value of a Gaussian kernel weighted loss function of the initial regression network model according to the value of the loss function of the initial regression network model;
and training the initial regression network model by using the value of the Gaussian kernel weighted loss function to obtain the regression model.
5. The method of claim 4, wherein the Gaussian kernel weighted loss functions of the regression network model are different for different human body parts.
6. The method of claim 3 or 4, wherein the loss function of the initial regression network model comprises any of the following functions: a mean square error function; mean absolute error function.
7. The method of claim 3, wherein before obtaining the value of the loss function of the initial regression network model based on the sample labels and the labels of the human body parts in the first sample medical image in advance, the method further comprises:
and according to a preset division rule, dividing the human body part to obtain the mark of the human body part in the first sample medical image in advance.
8. The method of claim 1, wherein the training process of the classification network model comprises:
acquiring a second sample medical image;
inputting the second sample medical image into a preset initial classification network model, and determining a human body part to which the second sample medical image belongs;
obtaining a loss function value of the initial classification network model according to the human body part to which the second sample medical image belongs and a marking result of the second sample medical image in advance;
and training the initial classification network model according to the value of the loss function of the initial classification network model to obtain the classification network model.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method according to any of claims 1-8.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911249073.3A CN111080594B (en) | 2019-12-09 | 2019-12-09 | Slice marker determination method, computer device, and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911249073.3A CN111080594B (en) | 2019-12-09 | 2019-12-09 | Slice marker determination method, computer device, and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111080594A true CN111080594A (en) | 2020-04-28 |
| CN111080594B CN111080594B (en) | 2024-03-26 |
Family
ID=70313261
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911249073.3A Active CN111080594B (en) | 2019-12-09 | 2019-12-09 | Slice marker determination method, computer device, and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111080594B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190065897A1 (en) * | 2017-08-28 | 2019-02-28 | Boe Technology Group Co., Ltd. | Medical image analysis method, medical image analysis system and storage medium |
| CN109461161A (en) * | 2018-10-22 | 2019-03-12 | 北京连心医疗科技有限公司 | A method of human organ in medical image is split based on neural network |
| CN109753859A (en) * | 2017-11-08 | 2019-05-14 | 佳能株式会社 | The device and method and image processing system of human part are detected in the picture |
| CN110136103A (en) * | 2019-04-24 | 2019-08-16 | 平安科技（深圳）有限公司 | Medical image means of interpretation, device, computer equipment and storage medium |
| CN110309888A (en) * | 2019-07-11 | 2019-10-08 | 南京邮电大学 | A kind of image classification method and system based on layering multi-task learning |
-
2019
- 2019-12-09 CN CN201911249073.3A patent/CN111080594B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190065897A1 (en) * | 2017-08-28 | 2019-02-28 | Boe Technology Group Co., Ltd. | Medical image analysis method, medical image analysis system and storage medium |
| CN109753859A (en) * | 2017-11-08 | 2019-05-14 | 佳能株式会社 | The device and method and image processing system of human part are detected in the picture |
| CN109461161A (en) * | 2018-10-22 | 2019-03-12 | 北京连心医疗科技有限公司 | A method of human organ in medical image is split based on neural network |
| CN110136103A (en) * | 2019-04-24 | 2019-08-16 | 平安科技（深圳）有限公司 | Medical image means of interpretation, device, computer equipment and storage medium |
| CN110309888A (en) * | 2019-07-11 | 2019-10-08 | 南京邮电大学 | A kind of image classification method and system based on layering multi-task learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111080594B (en) | 2024-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109472213B (en) | Palm print recognition method and device, computer equipment and storage medium | |
| CN109754396B (en) | Image registration method and device, computer equipment and storage medium | |
| CN110211076B (en) | Image stitching method, image stitching equipment and readable storage medium | |
| CN110674712A (en) | Interactive behavior recognition method and device, computer equipment and storage medium | |
| CN110599526B (en) | Image registration method, computer device, and storage medium | |
| CN111080573B (en) | Rib image detection method, computer device and storage medium | |
| CN111178128B (en) | Image recognition method, device, computer equipment and storage medium | |
| CN110210519B (en) | Classification method, computer device, and storage medium | |
| CN111832581B (en) | Lung feature recognition method and device, computer equipment and storage medium | |
| CN111144398A (en) | Target detection method, target detection device, computer equipment and storage medium | |
| CN112102235B (en) | Human body part recognition method, computer device, and storage medium | |
| CN114549462A (en) | Focus detection method, device, equipment and medium based on visual angle decoupling Transformer model | |
| CN111583184A (en) | Image analysis method, network, computer device, and storage medium | |
| CN110717449A (en) | Vehicle annual inspection personnel behavior detection method and device and computer equipment | |
| CN110738702B (en) | Three-dimensional ultrasonic image processing method, device, equipment and storage medium | |
| CN110490841B (en) | Computer-aided image analysis method, computer device and storage medium | |
| CN110309855B (en) | Training method for image segmentation, computer device and storage medium | |
| CN111178126A (en) | Target detection method, target detection device, computer equipment and storage medium | |
| CN111027469B (en) | Human body part recognition method, computer device, and readable storage medium | |
| CN110310314B (en) | Image registration method and device, computer equipment and storage medium | |
| CN109063601B (en) | Lip print detection method and device, computer equipment and storage medium | |
| CN111145152A (en) | Image detection method, computer device, and storage medium | |
| CN111080594A (en) | Human body part recognition method, computer device and readable storage medium | |
| CN115147792A (en) | Vision-based positioning method and device, computer equipment and storage medium | |
| CN110766652B (en) | Network training method, device, segmentation method, computer 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |