CN112767502B - Image processing method and device based on medical image model - Google Patents
Image processing method and device based on medical image model Download PDFInfo
- Publication number
- CN112767502B CN112767502B CN202110024946.1A CN202110024946A CN112767502B CN 112767502 B CN112767502 B CN 112767502B CN 202110024946 A CN202110024946 A CN 202110024946A CN 112767502 B CN112767502 B CN 112767502B
- Authority
- CN
- China
- Prior art keywords
- convolution
- processing
- encoder
- pooling
- result
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
-
- 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
-
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Multimedia (AREA)
- Public Health (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention relates to an image processing method and device based on a medical image model, which perform convolution pooling on a medical image to be processed through a first encoder, perform convolution processing on a convolution pooling processing result, and perform up-sampling convolution processing on a convolution processing result of the first encoder through a first decoder. And performing feature map splicing on the convolution pooling processing result of the first encoder through the second encoder, performing convolution and/or maximum pooling operation on the splicing processing result, performing convolution pooling processing on the processing result of the second encoder through the second decoder, and obtaining a combined feature map according to the convolution pooling processing result of the second decoder. Based on the above, the generalization capability of the medical image model is improved through the two encoding and decoding processes. Meanwhile, based on the processing basis of the medical image model, the obtained combined feature map is the target area of the medical image to be processed, so that medical workers can quickly identify tissues or focuses of the target area, and the working efficiency is improved.
Description
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to an image processing method and apparatus based on a medical image model.
Background
Medical imaging refers to the technique and process of obtaining images of internal tissues of a human body or a part of the human body in a non-invasive manner for medical treatment or medical research. Generally, medical images contain the following two relatively independent directions of study: medical imaging systems and medical image processing. After the imaging system in the hospital completes the imaging of the image, the image needs to be further processed, so as to facilitate the subsequent related processing such as medical diagnosis or data processing.
At present, the conventional medical image processing mainly aims at the processing of image characteristics, such as image enhancement on medical images for medical staff to observe or data compression on medical images for cross-domain transmission. Medical image processing greatly facilitates the development complexity of medical care work on an image level. However, in the imaging process of medical images, a great number of objective interference factors are naturally caused by individual differences of human bodies. Taking medical images as chest X-ray for imaging the lung as an example, due to the individual difference of imaging objects, the obtained lung images are different in size, so that lung tissues are difficult to clearly show, and the efficiency of medical care work such as diagnosis is influenced.
Disclosure of Invention
Therefore, it is necessary to provide an image processing method and an image processing device based on a medical image model for overcoming the defect that the efficiency of medical care work is affected by the fact that the traditional medical image also clearly shows relevant information such as a focus.
An image processing method based on medical image model is applied to the medical image model comprising a first encoder, a second encoder, a first decoder and a second decoder, and comprises the following steps:
acquiring a medical image to be processed;
performing convolution pooling on the medical image to be processed through a first encoder, and performing convolution processing on a convolution pooling processing result;
performing up-sampling convolution processing on a convolution processing result of the first encoder through a first decoder;
performing characteristic diagram splicing processing on the convolution pooling processing result of the first encoder through a second encoder, and performing convolution and/or maximum pooling operation on the splicing processing result;
and performing convolution pooling on the processing result of the second encoder through a second decoder, and obtaining a merging characteristic diagram according to the convolution pooling result of the second decoder.
According to the image processing method based on the medical image model, after the medical image to be processed is obtained, the medical image to be processed is subjected to convolution pooling through the first encoder, convolution processing is carried out on the result of the convolution pooling, and up-sampling convolution processing is carried out on the result of the convolution processing of the first encoder through the first decoder. And further, performing feature map splicing on the convolution pooling processing result of the first encoder through a second encoder, performing convolution and/or maximum pooling on the splicing processing result, performing convolution pooling processing on the processing result of the second encoder through a second decoder, and obtaining a combined feature map according to the convolution pooling processing result of the second decoder. Based on the above, the generalization capability of the medical image model is improved through the two encoding and decoding processes. Meanwhile, based on the processing basis of the medical image model, the obtained combined feature map is the target area of the medical image to be processed, so that medical workers can quickly identify tissues or focuses of the target area, and the working efficiency is improved.
In one embodiment, the process of performing convolution pooling on the medical image to be processed by the first encoder and performing convolution processing on the result of the convolution pooling includes the steps of:
sequentially carrying out three-wheel convolution pooling processing on medical images to be processed through a first encoder; wherein each round of convolution pooling process comprises two convolution operations of 3x3 and one maximum pooling of 2x 2.
In one embodiment, the process of upsampling convolution processing by the first decoder on the convolution processing result of the first encoder comprises the steps of:
performing three rounds of up-sampling convolution processing on the convolution processing result of the first encoder through the first decoder; wherein each round of upsampling convolution processing comprises 2x2 upsampling and two 3x3 convolution operations.
In one embodiment, the process of performing feature map splicing on the convolution pooling processing result of the first encoder and performing convolution and/or maximum pooling on the splicing processing result by the second encoder comprises the following steps;
performing primary feature map splicing processing on the first round of convolution pooling processing results through a second encoder, and performing primary 2x2 maximum pooling on the primary feature map splicing processing results;
performing second feature map splicing treatment on the second round of convolution pooling treatment results and the maximum pooling result of 2x2 of the first feature map splicing treatment results through a second encoder, and performing two times of convolution operation of 3x3 and one time of maximum pooling of 2x2 on the second feature map splicing treatment results;
and splicing the 2x2 maximum pooling result of the first round convolution pooling result and the second feature map splicing result.
In one embodiment, the process of performing convolutional pooling on the processing result of the second encoder by the second decoder comprises the steps of:
performing three-wheel convolution pooling processing on the processing result of the second encoder through a second decoder; wherein, each round of convolution pooling comprises two convolution operations of 3x3 and one maximum pooling of 2x 2;
three rounds of convolution pooling were performed two more times with 3x3 convolution operations.
In one embodiment, the process of obtaining the merged feature map according to the result of the convolutional pooling process of the second decoder comprises the steps of:
and carrying out convolution operation of 1x1 once on the convolution pooling processing result of the second decoder to obtain a merged feature map.
In one embodiment, the medical image model includes an M-ResNet neural network.
An image processing apparatus based on medical image model, applied to a medical image model including a first encoder, a second encoder, a first decoder and a second decoder, comprises:
the image acquisition module is used for acquiring a medical image to be processed;
the first processing module is used for performing convolution pooling on the medical image to be processed through the first encoder and performing convolution processing on a convolution pooling result;
the second processing module is used for performing up-sampling convolution processing on the convolution processing result of the first encoder through the first decoder;
the third processing module is used for performing characteristic diagram splicing processing on the convolution pooling processing result of the first encoder through the second encoder and performing convolution and/or maximum pooling operation on the splicing processing result;
and the fourth processing module is used for performing convolution pooling processing on the processing result of the second encoder through the second decoder and obtaining a combined feature map according to the convolution pooling processing result of the second decoder.
After the medical image to be processed is acquired, the image processing device based on the medical image model performs convolution pooling on the medical image to be processed through the first encoder, performs convolution processing on a convolution pooling result, and performs up-sampling convolution processing on a convolution processing result of the first encoder through the first decoder. And further, performing feature map splicing on the convolution pooling processing result of the first encoder through a second encoder, performing convolution and/or maximum pooling operation on the splicing processing result, performing convolution pooling processing on the processing result of the second encoder through a second decoder, and obtaining a combined feature map according to the convolution pooling processing result of the second decoder. Based on the above, the generalization capability of the medical image model is improved through the two encoding and decoding processes. Meanwhile, based on the processing basis of the medical image model, the obtained combined feature map is the target area of the medical image to be processed, so that medical workers can quickly identify tissues or focuses of the target area, and the working efficiency is improved.
A computer storage medium, on which computer instructions are stored, and the computer instructions, when executed by a processor, implement the image processing method based on medical image model according to any of the above embodiments.
After the medical image to be processed is obtained, the computer storage medium performs convolution pooling processing on the medical image to be processed through the first encoder, performs convolution processing on a convolution pooling processing result, and performs up-sampling convolution processing on a convolution processing result of the first encoder through the first decoder. And further, performing feature map splicing on the convolution pooling processing result of the first encoder through a second encoder, performing convolution and/or maximum pooling operation on the splicing processing result, performing convolution pooling processing on the processing result of the second encoder through a second decoder, and obtaining a combined feature map according to the convolution pooling processing result of the second decoder. Based on the above, the generalization capability of the medical image model is improved through the two encoding and decoding processes. Meanwhile, based on the processing basis of the medical image model, the obtained combined feature map is the target area of the medical image to be processed, so that medical workers can quickly identify tissues or focuses of the target area, and the working efficiency is improved.
A computer device includes a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the image processing method based on the medical image model according to any of the above embodiments.
After the computer device acquires the medical image to be processed, performing convolution pooling on the medical image to be processed through the first encoder, performing convolution processing on a convolution pooling processing result, and performing up-sampling convolution processing on a convolution processing result of the first encoder through the first decoder. And further, performing feature map splicing on the convolution pooling processing result of the first encoder through a second encoder, performing convolution and/or maximum pooling operation on the splicing processing result, performing convolution pooling processing on the processing result of the second encoder through a second decoder, and obtaining a combined feature map according to the convolution pooling processing result of the second decoder. Based on the above, the generalization capability of the medical image model is improved through the two encoding and decoding processes. Meanwhile, based on the processing basis of the medical image model, the obtained combined feature map is the target area of the medical image to be processed, so that medical workers can quickly identify tissues or focuses of the target area, and the working efficiency is improved.
Drawings
FIG. 1 is a schematic diagram of a medical image model structure;
FIG. 2 is a flowchart of an embodiment of an image processing method based on a medical image model;
FIG. 3 is a flowchart illustrating an image processing method based on a medical image model according to another embodiment;
FIG. 4 is a diagram illustrating a structure of a medical image model according to another embodiment;
fig. 5 is a block diagram of an image processing apparatus based on a medical image model according to an embodiment.
Detailed Description
For better understanding of the objects, technical solutions and effects of the present invention, the present invention will be further explained with reference to the accompanying drawings and examples. Meanwhile, the following described examples are only for explaining the present invention, and are not intended to limit the present invention.
The embodiment of the invention provides an image processing method based on a medical image model.
The image processing method based on the medical image model is based on the medical image model. The medical image model is constructed based on a neural network. As a preferred embodiment, the medical image model includes an M-ResNet (Residual Neural Network) Neural Network.
Fig. 1 is a schematic structural diagram of a medical image model, and as shown in fig. 1, the medical image model includes a first encoder Encode _1, a first decoder Decode _1, a second encoder Encode _2, and a second decoder Decode _2.
Fig. 2 is a flowchart illustrating an embodiment of an image processing method based on a medical image model, and as shown in fig. 2, the image processing method based on the medical image model includes steps S100 to S104:
s100, acquiring a medical image to be processed;
the medical image to be processed comprises raw data collected by a medical imaging device or raw data which is not processed by relevant medical and technical processing.
The medical image to be processed is input into the medical image model as a feature vector, and the vector level processing is completed sequentially by a first encoder Encode _1, a first decoder Decode _1, a second encoder Encode _2 and a second decoder Decode _2 of the medical image model.
S101, performing convolution pooling on the medical image to be processed through a first encoder Encode _1, and performing convolution processing on a convolution pooling result;
and performing convolution pooling on the medical image to be processed by the first encoder Encode _1, and performing convolution processing on a convolution pooling result after the convolution pooling processing to finish the first encoding operation of the first encoder Encode _ 1.
In one embodiment, fig. 3 is a flowchart of an image processing method based on a medical image model according to another embodiment, and as shown in fig. 3, a process of performing convolution pooling on a medical image to be processed by a first encoder Encode _1 in step S101 and performing convolution processing on a result of the convolution pooling includes step S200:
s200, sequentially carrying out three-wheel convolution pooling on medical images to be processed through a first encoder Encode _ 1; wherein each round of convolution pooling process comprises two convolution operations of 3x3 and one maximum pooling of 2x 2.
Fig. 4 is a schematic structural diagram of a medical image model according to another embodiment, and as shown in fig. 4, the first encoder Encode _1 performs three rounds of convolution pooling, and performs 2x2 maximal pooling after every 3x3 convolution operations. After three rounds of convolution pooling, 3x3 convolution operations were performed twice more.
As shown in fig. 4, the arrow pointing to the right indicates a 3 × 3 convolutional layer and ReLu activation for feature extraction; the up-pointing arrow represents the 2x2 max pooling layer for narrowing down the feature map; the downward pointing arrow represents a 2x2 upsampling for enlarging the feature map.
S102, performing up-sampling convolution processing on a convolution processing result of a first encoder Encode _1 through a first decoder Decode _ 1;
the convolution processing result of the first encoder Encode _1 is subjected to upsampling convolution processing by the first decoder Decode _1, and the first decoding operation of the first decoder Decode _1 is completed.
In one embodiment, as shown in fig. 3, the process of performing the upsampling convolution processing on the convolution processing result of the first encoder Encode _1 by the first decoder Decode _1 in step S102 includes step S300:
s300, performing three rounds of up-sampling convolution processing on the convolution processing result of the first encoder Encode _1 through the first decoder Decode _ 1; wherein each round of upsampling convolution processing comprises 2x2 upsampling and two 3x3 convolution operations.
As shown in fig. 4, the first decoder Decode _1 performs convolution operations of 3 × 3 two times after performing up-sampling of 2 × 2 once, and performs up-sampling convolution three times in total.
S103, performing feature map splicing processing on the convolution pooling processing result of the first encoder Encode _1 through the second encoder Encode _2, and performing convolution and/or maximum pooling operation on the splicing processing result;
and performing characteristic diagram splicing processing on the convolution pooling processing result of the first encoder Encode _1 through the second encoder Encode _2, and performing convolution and/or maximum pooling operation on the splicing processing result to complete second encoding operation of the second encoder Encode _2.
In one embodiment, the process of performing feature map splicing on the result of convolution pooling processing of the first encoder by the second encoder and performing convolution and/or maximum pooling on the result of splicing processing in step S103 includes steps S400 to S402:
s400, performing first feature map splicing processing on the first round of convolution pooling processing results through a second encoder, and performing 2x2 maximal pooling on the first feature map splicing processing results;
s401, performing second feature map splicing processing on the second round of convolution pooling processing results and the 2x2 maximum pooling result of the first feature map splicing processing results through a second encoder, and performing two times of convolution operation of 3x3 and one time of maximum pooling of 2x2 on the second feature map splicing processing results;
and S402, splicing the 2x2 maximum pooling result of the first round convolution pooling result and the second feature map splicing result.
As shown in fig. 4, the first encoder Encode _1 connects the feature map of the last convolution of each round to the second encoder Encode _2 by copying. The second encoder Encode _2 is spliced with the feature map copied in the first round of the first encoder Encode _1, then maximal pooling of 2x2 is carried out once, and then the second encoder Encode _2 is spliced with the feature map copied in the second round of the first encoder. And after 3x3 convolution operation is performed twice, performing 2x2 maximum pooling once, and splicing with the feature map copied by the first encoder Encode _1, wherein two rounds of convolution pooling and feature map splicing are performed in total.
S104, performing convolution pooling on the processing result of the second encoder Encode _2 through the second decoder Decode _2, and obtaining a merging feature map according to the convolution pooling result of the second decoder Decode _2.
And performing convolution pooling on the processing result of the second encoder Encode _2 through the second decoder Decode _2, obtaining a merging feature map according to the convolution pooling result of the second decoder Decode _2, and completing second decoding processing of the second decoder Decode _2. And performing coding and decoding processing twice to obtain a combined feature map, namely the segmented target area image of the medical image to be processed.
In one embodiment, as shown in fig. 3, the process of performing convolution pooling processing on the processing result of the second encoder Encode _2 by the second decoder Decode _2 in step S104 includes steps S500 and S501:
s500, performing three-wheel convolution pooling processing on the processing result of the second encoder Encode _2 through the second decoder Decode _ 2; wherein, each round of convolution pooling comprises two times of convolution operation of 3x3 and one time of maximal pooling of 2x 2;
and S501, performing convolution operation of 3x3 twice on the results of the three-wheel convolution pooling processing.
As shown in fig. 4, the second decoder Decode _2 performs the maximum pooling for 2x2 once every 3x3 convolution operations, performs a total of three rounds of convolution pooling, and performs the convolution for 3x3 twice.
In one embodiment, as shown in fig. 3, the process of obtaining the merged feature map according to the result of the convolutional pooling process of the second decoder Decode _2 in step S104 includes step S600:
s600, performing convolution operation of 1x1 once on the convolution pooling processing result of the second decoder Decode _2 to obtain a merged feature map.
As shown in fig. 4, after the convolution operation of 3 × 3 is finally performed in step S401, the feature map obtained by the convolution operation of 3 × 3 is subjected to convolution operation of 1 × 1 to obtain a merged feature map.
In the image processing method based on the medical image model according to any of the embodiments, after the medical image to be processed is acquired, the medical image to be processed is subjected to convolution pooling processing by the first encoder Encode _1, a convolution processing result of the convolution pooling processing is subjected to convolution processing, and an up-sampling convolution processing result of the first encoder Encode _1 is subjected to up-sampling convolution processing by the first decoder Decode _ 1. Furthermore, feature map splicing processing is performed on the convolution pooling processing result of the first encoder Encode _1 through a second encoder Encode _2, convolution and/or maximum pooling operation is performed on the splicing processing result, finally, convolution pooling processing is performed on the processing result of the second encoder Encode _2 through a second decoder Decode _2, and a combined feature map is obtained according to the convolution pooling processing result of the second decoder Decode _2. Based on the above, the generalization capability of the medical image model is improved through the two encoding and decoding processes. Meanwhile, based on the processing basis of the medical image model, the obtained combined feature map is the target area of the medical image to be processed, so that medical workers can quickly identify tissues or focuses of the target area, and the working efficiency is improved.
The embodiment of the invention also provides an image processing device based on the medical image model.
Fig. 5 is a block diagram of an embodiment of an image processing apparatus based on a medical image model, and as shown in fig. 5, the image processing apparatus based on a medical image model according to an embodiment includes a module 100, a module 101, a module 102, a module 103, and a module 104:
the image acquisition module is used for acquiring a medical image to be processed;
the first processing module is used for performing convolution pooling processing on the medical image to be processed through the first encoder and performing convolution processing on a convolution pooling processing result;
the second processing module is used for performing up-sampling convolution processing on the convolution processing result of the first encoder through the first decoder;
the third processing module is used for performing characteristic diagram splicing processing on the convolution pooling processing result of the first encoder through the second encoder and performing convolution and/or maximum pooling operation on the splicing processing result;
and the fourth processing module is used for performing convolution pooling processing on the processing result of the second encoder through the second decoder and obtaining a combined feature map according to the convolution pooling processing result of the second decoder.
After the medical image to be processed is acquired, the image processing device based on the medical image model performs convolution pooling on the medical image to be processed through the first encoder, performs convolution processing on a convolution pooling result, and performs up-sampling convolution processing on a convolution processing result of the first encoder through the first decoder. And further, performing feature map splicing on the convolution pooling processing result of the first encoder through a second encoder, performing convolution and/or maximum pooling operation on the splicing processing result, performing convolution pooling processing on the processing result of the second encoder through a second decoder, and obtaining a combined feature map according to the convolution pooling processing result of the second decoder. Based on the above, the generalization capability of the medical image model is improved through the two encoding and decoding processes. Meanwhile, based on the processing basis of the medical image model, the obtained combined feature map is the target area of the medical image to be processed, so that medical workers can quickly identify tissues or focuses of the target area, and the working efficiency is improved.
The embodiment of the present invention further provides a computer storage medium, on which computer instructions are stored, and when the instructions are executed by a processor, the image processing method based on a medical image model according to any one of the above embodiments is implemented.
Those skilled in the art will understand that: all or part of the steps for implementing the method embodiments may be implemented by hardware related to program instructions, and the program may be stored in a computer readable storage medium, and when executed, the program performs the steps including the method embodiments; and the aforementioned storage medium includes: various media that can store program codes, such as a removable Memory device, a Random Access Memory (RAM), a Read-Only Memory (ROM), a magnetic disk, and an optical disk.
Alternatively, the integrated unit of the present invention may be stored in a computer-readable storage medium if it is implemented in the form of a software functional module and sold or used as a separate product. Based on such understanding, the technical solutions of the embodiments of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a terminal, or a network device) to execute all or part of the methods of the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, a RAM, a ROM, a magnetic or optical disk, or various other media that can store program code.
Corresponding to the computer storage medium, in an embodiment, a computer device is further provided, where the computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the program, the image processing method based on the medical image model in any one of the above embodiments is implemented.
After the medical image to be processed is obtained, the computer device performs convolution pooling processing on the medical image to be processed through the first encoder, performs convolution processing on a convolution pooling processing result, and performs up-sampling convolution processing on a convolution processing result of the first encoder through the first decoder. And further, performing feature map splicing on the convolution pooling processing result of the first encoder through a second encoder, performing convolution and/or maximum pooling on the splicing processing result, performing convolution pooling processing on the processing result of the second encoder through a second decoder, and obtaining a combined feature map according to the convolution pooling processing result of the second decoder. Based on the above, the generalization capability of the medical image model is improved through the two encoding and decoding processes. Meanwhile, based on the processing basis of the medical image model, the obtained combined feature map is the target area of the medical image to be processed, so that medical workers can quickly identify tissues or focuses of the target area, and the working efficiency is improved.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only show some 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 should be subject to the appended claims.
Claims (10)
1. An image processing method based on medical image model is applied to the medical image model comprising a first encoder, a second encoder, a first decoder and a second decoder, and comprises the following steps:
acquiring a medical image to be processed;
performing convolution pooling on the medical image to be processed through the first encoder, and performing convolution processing on a convolution pooling processing result;
performing, by the first decoder, upsampling convolution processing on a convolution processing result of the first encoder;
performing feature map splicing processing on the convolution pooling processing result of the first encoder and the result of the up-sampling convolution processing through the second encoder, and performing convolution and/or maximum pooling on the splicing processing result;
and performing convolution pooling on the processing result of the second encoder through the second decoder, and obtaining a combined feature map according to the convolution pooling result of the second decoder.
2. The medical image model-based image processing method as claimed in claim 1, wherein said procedure of performing convolution pooling on said medical image to be processed by said first encoder and performing convolution processing on the result of said convolution pooling comprises the steps of:
sequentially carrying out three-wheel convolution pooling processing on the medical image to be processed through the first encoder; wherein each round of convolution pooling process comprises two convolution operations of 3x3 and one maximum pooling of 2x 2.
3. The method as claimed in claim 1, wherein the step of performing the upsampling convolution processing on the convolution processing result of the first encoder by the first decoder comprises the steps of:
performing three rounds of upsampling convolution processing on the convolution processing result of the first encoder through the first decoder; wherein each round of upsampling convolution processing comprises 2x2 upsampling and two 3x3 convolution operations.
4. The medical image model-based image processing method according to claim 2, wherein the process of performing feature map stitching processing on the result of convolution pooling processing of the first encoder and the result of convolution processing of the upsampling by the second encoder and performing convolution and/or max-pooling on the result of stitching processing by the second encoder comprises the steps of:
performing first feature map splicing processing on a first round of convolution pooling processing result and the result of the up-sampling convolution processing through the second encoder, and performing 2x2 maximum pooling on the first feature map splicing processing result;
performing second feature map splicing processing on the result of the second round of convolution pooling processing and the maximum pooling result of 2x2 of the result of the first feature map splicing processing through the second encoder, and performing two times of convolution operation of 3x3 and one time of maximum pooling of 2x2 on the result of the second feature map splicing processing;
and splicing the first round convolution pooling processing result and the maximum pooling result of 2x2 of the second feature map splicing processing result.
5. The method of claim 1, wherein the convolution pooling of the processing result of the second encoder by the second decoder comprises:
performing three rounds of convolution pooling on the processing result of the second encoder by the second decoder; wherein, each round of convolution pooling comprises two convolution operations of 3x3 and one maximum pooling of 2x 2;
and performing 3x3 convolution operation twice on the results of the three rounds of convolution pooling processing.
6. The method as claimed in claim 5, wherein the step of obtaining the merged feature map according to the result of the convolutional pooling process of the second decoder comprises:
and carrying out convolution operation of 1x1 once on the convolution pooling processing result of the second decoder to obtain the merging characteristic diagram.
7. The medical image model-based image processing method according to any one of claims 1 to 6, wherein the medical image model comprises an M-ResNet neural network.
8. An image processing apparatus based on a medical image model, applied to a medical image model including a first encoder, a second encoder, a first decoder and a second decoder, comprises:
the image acquisition module is used for acquiring a medical image to be processed;
the first processing module is used for performing convolution pooling processing on the medical image to be processed through the first encoder and performing convolution processing on the convolution pooling processing result;
a second processing module, configured to perform, by the first decoder, upsampling convolution processing on a convolution processing result of the first encoder;
the third processing module is used for performing characteristic map splicing processing on the convolution pooling processing result of the first encoder and the result of the up-sampling convolution processing through the second encoder, and performing convolution and/or maximum pooling on the splicing processing result;
and the fourth processing module is used for performing convolution pooling processing on the processing result of the second encoder through the second decoder and obtaining a combined feature map according to the convolution pooling processing result of the second decoder.
9. A computer storage medium having computer instructions stored thereon, wherein the computer instructions, when executed by a processor, implement the medical image model-based image processing method according to any one of claims 1 to 7.
10. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the image processing method based on the medical image model according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110024946.1A CN112767502B (en) | 2021-01-08 | 2021-01-08 | Image processing method and device based on medical image model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110024946.1A CN112767502B (en) | 2021-01-08 | 2021-01-08 | Image processing method and device based on medical image model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112767502A CN112767502A (en) | 2021-05-07 |
| CN112767502B true CN112767502B (en) | 2023-04-07 |
Family
ID=75701092
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110024946.1A Active CN112767502B (en) | 2021-01-08 | 2021-01-08 | Image processing method and device based on medical image model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112767502B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114601560B (en) * | 2022-05-11 | 2022-08-19 | 中国科学院深圳先进技术研究院 | Minimally invasive surgery assisting method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111325751A (en) * | 2020-03-18 | 2020-06-23 | 重庆理工大学 | CT image segmentation system based on attention convolution neural network |
| CN111681252A (en) * | 2020-05-30 | 2020-09-18 | 重庆邮电大学 | Medical image automatic segmentation method based on multipath attention fusion |
| CN112150428A (en) * | 2020-09-18 | 2020-12-29 | 青岛大学 | Medical image segmentation method based on deep learning |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3994616A1 (en) * | 2019-07-01 | 2022-05-11 | Optimum Semiconductor Technologies, Inc. | Feedbackward decoder for parameter efficient semantic image segmentation |
| CN111311629B (en) * | 2020-02-21 | 2023-12-01 | 京东方科技集团股份有限公司 | Image processing method, image processing device and equipment |
| CN111627012B (en) * | 2020-05-28 | 2021-12-21 | 华北电力大学（保定） | Deep neural network surface defect detection method based on feature fusion |
-
2021
- 2021-01-08 CN CN202110024946.1A patent/CN112767502B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111325751A (en) * | 2020-03-18 | 2020-06-23 | 重庆理工大学 | CT image segmentation system based on attention convolution neural network |
| CN111681252A (en) * | 2020-05-30 | 2020-09-18 | 重庆邮电大学 | Medical image automatic segmentation method based on multipath attention fusion |
| CN112150428A (en) * | 2020-09-18 | 2020-12-29 | 青岛大学 | Medical image segmentation method based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112767502A (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114612479B (en) | Medical image segmentation method and device based on global and local feature reconstruction network | |
| CN112602114B (en) | Image processing method and device, neural network, training method and storage medium | |
| CN116433914A (en) | Two-dimensional medical image segmentation method and system | |
| CN116309648A (en) | Medical image segmentation model construction method based on multi-attention fusion | |
| CN110674824A (en) | Finger vein segmentation method and device based on R2U-Net and storage medium | |
| CN111161269A (en) | Image segmentation method, computer device and readable storage medium | |
| KR20200014396A (en) | Medical Image De-Identification System, Method and Computer Readable Recording Medium | |
| CN112767502B (en) | Image processing method and device based on medical image model | |
| CN116912503B (en) | Multi-mode MRI brain tumor semantic segmentation method based on hierarchical fusion strategy | |
| KR102419270B1 (en) | Apparatus and method for segmenting medical image using mlp based architecture | |
| CN114187181B (en) | Dual-path lung CT image super-resolution method based on residual information refining | |
| CN116935051B (en) | Polyp segmentation network method, system, electronic equipment and storage medium | |
| US20210074034A1 (en) | Methods and apparatus for neural network based image reconstruction | |
| CN112767403A (en) | Medical image segmentation model training method, medical image segmentation method and device | |
| CN113177938B (en) | Method and device for segmenting brain glioma based on circular convolution kernel and related components | |
| Huang et al. | Anatomical‐functional image fusion based on deep convolution neural networks in local Laplacian pyramid domain | |
| CN116823851A (en) | Feature reconstruction-based unsupervised domain self-adaptive OCT image segmentation method and system | |
| CN114299010A (en) | Method and device for segmenting brain tumor image, computer equipment and storage medium | |
| CN111626972B (en) | CT image reconstruction method, model training method and equipment | |
| CN115100050A (en) | Method, device and equipment for removing ring artifacts of CT image and storage medium | |
| CN113379770A (en) | Nasopharyngeal carcinoma MR image segmentation network construction method, image segmentation method and device | |
| CN113139964A (en) | Multi-modal image segmentation method and device, electronic equipment and storage medium | |
| CN113327221A (en) | Image synthesis method and device fusing ROI (region of interest), electronic equipment and medium | |
| Thamm et al. | Detection of large vessel occlusions using deep learning by deforming vessel tree segmentations | |
| CN118297961A (en) | Medical image segmentation method based on CNN-transducer parallel encoder |
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 |