CN112435242A - Lung image processing method and device, electronic equipment and storage medium - Google Patents
Lung image processing method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN112435242A CN112435242A CN202011345196.XA CN202011345196A CN112435242A CN 112435242 A CN112435242 A CN 112435242A CN 202011345196 A CN202011345196 A CN 202011345196A CN 112435242 A CN112435242 A CN 112435242A
- Authority
- CN
- China
- Prior art keywords
- historical
- image processing
- chest
- lung image
- images
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/143—Segmentation; Edge detection involving probabilistic approaches, e.g. Markov random field [MRF] modelling
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
-
- 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
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30096—Tumor; Lesion
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Radiology & Medical Imaging (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Quality & Reliability (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention discloses a lung image processing method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a lung image to be processed; inputting the lung image to be processed into a trained lung image processing model to obtain a probability value determined as tuberculosis and a corresponding region of interest thermodynamic diagram in the lung image to be processed; the lung image processing is trained based on a plurality of groups of historical images, and the historical images comprise historical chest orthophotos images and historical annotation information corresponding to the historical chest orthophotos images. The pulmonary tuberculosis screening method has the advantages of improving the pulmonary tuberculosis omission phenomenon in the pulmonary tuberculosis screening process, reducing the manual participation in the pulmonary tuberculosis screening process, improving the pulmonary tuberculosis screening efficiency and assisting a doctor in completing the diagnosis of pulmonary tuberculosis.
Description
Technical Field
The embodiments of the present invention relate to medical image processing technologies, and in particular, to a lung image processing method and apparatus, an electronic device, and a storage medium.
Background
In the medical field, tuberculosis is one of the infectious diseases with the highest fatality rate at present, and the tuberculosis screening operation is completed by positioning and identifying a focus related to the tuberculosis in a target positive chest radiography image by a doctor through naked eyes at present.
In the process of pulmonary tuberculosis screening, only obvious lesions in the chest radiograph can be screened generally, tiny lesions in the chest radiograph cannot be screened, and the pulmonary tuberculosis omission phenomenon is easy to occur. Meanwhile, the screening scheme has extremely high requirements on medical quality of doctors, corresponding screening operation needs to be completed manually by the doctors, and the overall screening efficiency is not high.
Disclosure of Invention
The embodiment of the invention provides a pulmonary image processing method, a pulmonary image processing device, electronic equipment and a storage medium, which are used for improving the pulmonary tuberculosis omission phenomenon in a pulmonary tuberculosis screening process, reducing the artificial participation degree in the pulmonary tuberculosis screening process, improving the pulmonary tuberculosis screening efficiency and assisting a doctor in completing the diagnosis effect of pulmonary tuberculosis.
In a first aspect, an embodiment of the present invention provides a lung image processing method, including:
acquiring a lung image to be processed;
inputting the lung image to be processed into a trained lung image processing model to obtain a probability value determined as tuberculosis and a corresponding region of interest thermodynamic diagram in the lung image to be processed; the lung image processing is trained based on a plurality of groups of historical images, and the historical images comprise historical chest orthophotos images and historical annotation information corresponding to the historical chest orthophotos images.
In a second aspect, an embodiment of the present invention further provides a lung image processing apparatus, including:
the image acquisition module is used for acquiring a lung image to be processed;
a pulmonary tuberculosis determining module, configured to input the to-be-processed lung image into a trained lung image processing model, so as to obtain a probability value determined as pulmonary tuberculosis and a corresponding thermodynamic diagram of a region of interest in the to-be-processed lung image; the lung image processing model is obtained based on training of multiple groups of historical images, and the historical images comprise historical chest orthophotos images and historical annotation information corresponding to the historical chest orthophotos images.
In a third aspect, an embodiment of the present invention further provides an electronic device, where the electronic device includes:
one or more processors;
storage means for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the lung image processing method of any of the embodiments of the present invention.
In a fourth aspect, embodiments of the present invention further provide a storage medium containing computer-executable instructions, which when executed by a computer processor, are configured to perform the lung image processing method according to any one of the embodiments of the present invention.
According to the technical scheme of the embodiment of the invention, the obtained lung image to be processed is input into the trained lung image processing model to obtain the probability value determined as the pulmonary tuberculosis, so that the probability value of the pulmonary tuberculosis in the lung image to be processed is rapidly and accurately obtained, the differential diagnosis of the pulmonary tuberculosis is realized, the phenomenon of missed detection of the pulmonary tuberculosis in the pulmonary tuberculosis screening process in the prior art is solved, the artificial participation in the pulmonary tuberculosis screening process is reduced, the pulmonary tuberculosis screening efficiency is improved, and the effect of assisting a doctor in completing the diagnosis of the pulmonary tuberculosis is realized. The trained lung image processing model can also output a corresponding region-of-interest thermodynamic diagram in the to-be-processed lung image, so that the pulmonary tuberculosis in the to-be-processed lung image can be accurately and quickly positioned, and the position of the pulmonary tuberculosis can be found.
Drawings
FIG. 1 is a flow chart of a method for processing lung images according to one embodiment of the present invention;
fig. 2 is a schematic structural diagram of ResNet in the first embodiment of the present invention;
FIG. 3 is a diagram of a residual block structure according to a first embodiment of the present invention;
FIG. 4 is a schematic diagram of history annotation information according to a first embodiment of the present invention;
FIG. 5 is a schematic diagram of a labeling mask of history labeling information according to a first embodiment of the present invention;
FIG. 6 is a flowchart of a lung image processing method according to a second embodiment of the present invention;
FIG. 7 is a schematic diagram of feature map generation in the second embodiment of the present invention;
FIG. 8 is a flowchart illustrating a lung image processing method according to a second embodiment of the present invention;
FIG. 9 is a schematic diagram illustrating the determination of logic values according to a second embodiment of the present invention;
FIG. 10 is a schematic diagram illustrating the generation of a logic diagram according to a second embodiment of the present invention;
FIG. 11 is a schematic diagram showing comparison between history annotation information and a thermodynamic diagram of a region of interest in a second embodiment of the present invention;
FIG. 12 is a schematic diagram of an ROC curve of an evaluation lung image processing model according to a second embodiment of the present invention;
FIG. 13 is a flowchart of a method for processing images of lungs according to a third embodiment of the present invention;
fig. 14 is a schematic structural diagram of a lung image processing apparatus according to a fourth embodiment of the present invention;
fig. 15 is a schematic structural diagram of an electronic device in a fifth embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
Example one
Fig. 1 is a flowchart of a lung image processing method according to an embodiment of the present invention, where the embodiment is applicable to processing a lung image and screening tuberculosis, the method may be executed by a lung image processing apparatus, the lung image processing apparatus may be implemented by software and/or hardware, and the lung image processing apparatus may be configured on an electronic computing device, and specifically includes the following steps:
and S110, acquiring a lung image to be processed.
For example, the lung image to be processed may be an image to be identified whether or not there is a tuberculosis lesion in the image.
Alternatively, the lung image to be processed may be a chest orthophoto image.
S120, inputting the lung image to be processed into a trained lung image processing model to obtain a probability value determined as tuberculosis and a corresponding thermodynamic diagram of an interested area in the lung image to be processed; the lung image processing model is obtained based on training of multiple groups of historical images, and the historical images comprise historical chest positive images and historical marking information corresponding to the historical chest positive images.
For example, the lung image processing model may be a model trained based on a plurality of sets of historical images, and may be a neural network model, for example. Specific examples of the method include a classic classification convolutional neural Network (Deep Residual Network, ResNet34) with Residual learning capability and an FPN model in a CNN model, such as a structural diagram of ResNet shown in fig. 2 and a Residual block diagram shown in fig. 3.
In the ResNet network structure of fig. 2, it is mainly composed of 4 stages, each of which is composed of a plurality of residual blocks, and the residual block structure is as shown in fig. 2.
The historical images may include historical chest positive slice images and historical annotation information corresponding to the historical chest positive slice images.
The historical chest positive image may be a chest positive image of a previously scanned patient.
Referring to the schematic diagram of the history annotation information described in fig. 4, the history annotation information may be JSON format information (specifically, a JSON file) for determining whether a lesion region (e.g., tuberculosis) exists in the history chest positive slice image, the JSON file includes a disease category in the history chest positive slice image that has been determined, and a polygonal lesion frame for outlining and labeling the lesion region in the image, where the frame is composed of a plurality of sets of coordinate point lists, that is, coordinate point information of the frame.
In fig. 4, a region a is a lesion region, and a doctor correspondingly marks the lesion region in the historical chest orthophoto image.
It should be noted that, after labeling a focus region, a doctor generates a corresponding labeling mask (as shown in fig. 5, a region a in fig. 5 is a focus region, that is, a tuberculosis region) from a focus region coordinate point list in history labeling information by using a preset algorithm, and when performing iterative training on a lung image processing model to be trained, the mask is used as a training label. The focus position in the mask is 1, and the other positions are 0, that is, the polygonal focus frame is internally set to 1, and the other positions are set to 0.
It should be noted that the preset algorithm here may be any algorithm that can generate a corresponding labeling mask from the lesion area coordinate point list in the historical labeling information, for example, the preset algorithm may use a polygon rendering function in OpenCV, which is an open source library, to complete the conversion from the coordinate point to the labeling mask. Without limitation, any algorithm that can generate a corresponding labeling mask for a lesion area in the history labeling information is within the protection scope of the embodiment of the present invention.
And performing iterative training on the lung image processing model to be trained by using a plurality of groups of historical images to obtain the trained lung image processing model. The training process for the lung image processing model is further described in the following embodiments.
Inputting the acquired lung image to be processed into a trained lung image processing model to obtain a probability value of the pulmonary tuberculosis in the lung image to be processed, comparing the probability value with a preset threshold, and if the probability value is greater than the preset threshold, proving that the pulmonary tuberculosis exists in the lung image to be processed.
Therefore, the probability value of the pulmonary tuberculosis in the pulmonary image to be processed is quickly and accurately obtained, the differential diagnosis of the pulmonary tuberculosis is realized, the missing detection phenomenon of the pulmonary tuberculosis in the pulmonary tuberculosis screening process in the prior art is solved, the manual participation in the pulmonary tuberculosis screening process is reduced, the pulmonary tuberculosis screening efficiency is improved, and the effect of assisting a doctor in completing the diagnosis of the pulmonary tuberculosis is realized.
The region of interest may be a region of interest in the lung image to be processed, such as a lesion region, and in particular a tuberculosis region.
After the acquired to-be-processed lung image is input into the trained lung image processing model, besides the probability value of the pulmonary tuberculosis in the to-be-processed lung image, a corresponding region-of-interest thermodynamic diagram, namely a pulmonary tuberculosis thermodynamic diagram, in the to-be-processed lung image can be obtained. The position of the tuberculosis in the image of the lung to be processed can be visually and clearly seen from the thermodynamic diagram.
Therefore, the pulmonary tuberculosis in the lung image to be processed can be accurately and quickly positioned, and the position of the pulmonary tuberculosis can be found.
According to the technical scheme of the embodiment of the invention, the obtained lung image to be processed is input into the trained lung image processing model to obtain the probability value determined as the pulmonary tuberculosis, so that the probability value of the pulmonary tuberculosis in the lung image to be processed is rapidly and accurately obtained, the differential diagnosis of the pulmonary tuberculosis is realized, the phenomenon of missed detection of the pulmonary tuberculosis in the pulmonary tuberculosis screening process in the prior art is solved, the artificial participation in the pulmonary tuberculosis screening process is reduced, the pulmonary tuberculosis screening efficiency is improved, and the effect of assisting a doctor in completing the diagnosis of the pulmonary tuberculosis is realized. The trained lung image processing model can also output a corresponding region-of-interest thermodynamic diagram in the to-be-processed lung image, so that the pulmonary tuberculosis in the to-be-processed lung image can be accurately and quickly positioned, and the position of the pulmonary tuberculosis can be found.
Example two
Fig. 6 is a flowchart of a lung image processing method according to a second embodiment of the present invention, which may be combined with various alternatives of the above embodiments. In an embodiment of the present invention, optionally, the training method of the lung image processing model includes: acquiring a plurality of groups of historical chest positive position images and historical marking information corresponding to the historical chest positive position images, wherein the historical marking information is information obtained after marking the tuberculosis in the historical chest positive position images; and inputting the plurality of groups of historical chest positive images and the historical marking information corresponding to the historical chest positive images into the lung image processing model to be trained, and performing iterative training on the lung image processing model to be trained to obtain the trained lung image processing model.
As shown in fig. 6, the method of the embodiment of the present invention specifically includes the following steps:
s210, acquiring a plurality of groups of historical chest positive images and historical marking information corresponding to the historical chest positive images, wherein the historical marking information is information obtained after the pulmonary tuberculosis in the historical chest positive images is marked.
For example, before training the lung image processing model to be trained, a training set is first obtained, where the training set may be the historical chest orthophotos image and the historical annotation information corresponding to the historical chest orthophotos image.
The historical chest radiograph image may be an image of a historical patient obtained from an image database of a hospital or other means, and the historical labeling information corresponding to the historical chest radiograph image may be information obtained by labeling a lesion region in the historical chest radiograph image, that is, JSON format information for determining whether a lesion region (e.g., tuberculosis) exists in the historical chest radiograph image.
S220, inputting a plurality of groups of historical chest positive images and historical marking information corresponding to the historical chest positive images into the lung image processing model to be trained, and performing iterative training on the lung image processing model to be trained to obtain the trained lung image processing model.
Illustratively, prior to inputting the historical chest orthophoto images and the historical annotation information into the lung image processing model to be trained, the historical annotation information needs to be correspondingly processed. The specific processing procedure is as described in the above embodiment, and generates a corresponding labeling mask for the lesion area in the history labeling information by using a preset algorithm.
And inputting the historical chest positive image, the historical marking information and the marking mask into a lung image processing model to be trained, and performing iterative training on the lung image processing model to be trained to obtain the trained lung image processing model.
Therefore, a trained lung image processing model can be obtained, so that a lung image to be processed can be processed subsequently based on the trained lung image processing model to obtain a probability value of the tuberculosis and a corresponding thermodynamic diagram of the region of interest.
Optionally, the iterative training is performed on the lung image processing model to be trained to obtain a trained lung image processing model, which may specifically be: for any group of historical chest positive images input into the lung image processing model to be trained and historical annotation information corresponding to the historical chest positive images, the following steps are executed, and the lung image processing model to be trained is iteratively trained: obtaining a feature map corresponding to the historical chest orthophoto image based on the historical chest orthophoto image; calculating the characteristic value of each channel in the characteristic diagram to obtain a target characteristic value of each channel; carrying out weighted summation on the target characteristic values to obtain a logic value of the characteristic diagram, and determining a probability value of tuberculosis based on the logic value; determining a first loss function of a lung image processing model to be trained based on the probability value, a preset probability threshold and historical labeling information corresponding to the historical chest orthophoto image; and when the first loss function is smaller than a first preset loss threshold value, determining that the training of the lung image processing model to be trained is finished.
For example, for any historical chest orthophoto image, the corresponding feature map (feature map) may be an image obtained by feature extraction on the historical chest orthophoto image.
The feature extraction of the historical chest orthophoto image can be realized based on a deep learning model, so that a feature map corresponding to the historical chest orthophoto image can be obtained quickly.
Referring to the schematic feature map generation diagram described in fig. 7, the leftmost image in fig. 7 is an acquired historical chest orthophoto image, and the specific feature extraction process may be: the historical chest positive slice image is a single-channel image with a certain size, namely the historical chest positive slice image is a two-dimensional matrix image with the shape of H x W, wherein H and W represent the size of the historical chest positive slice image. After the historical breast normal position sheet image is input into the feature extraction model of the deep learning, an image of a multidimensional matrix, namely a feature map, is obtained, the feature map is a multidimensional matrix, the shape of the feature map is C x h w, C represents the number of channels, h and w represent the size of the feature map, as can be seen from FIG. 7, the number of channels is larger and larger as the depth of the model is increased, and the corresponding sizes of h and w are smaller and smaller (generally, h is w, which is also equal in the feature extraction model in the embodiment of the present invention).
As shown in fig. 7, H and W of the acquired original chest positive slice image (historical chest positive slice image) are inconsistent with H and W of the feature obtained after feature extraction, and the size of the feature map is generally smaller than that of the original chest positive slice image.
In the embodiment of the invention, H is H/32, W is W/32, that is, the size of the feature map is reduced by 32 times compared with the original breast normal position image.
The target feature value may be a value obtained by calculating a feature value of each channel in the feature map.
Referring to the execution flow chart of the lung image processing method illustrated in fig. 8, after a feature map corresponding to the historical chest orthophoto image is obtained, feature values of each channel in the feature map are calculated to obtain target feature values of each channel, weighted summation is performed on each target feature value to obtain a logit value corresponding to the feature map, and based on the logit value, a probability value of tuberculosis can be obtained.
Specifically, the logic value determination diagram described in fig. 9 may be referred to for obtaining the logic value according to the feature map, where fig. 9 takes 4 channels as an example, the uppermost graph in fig. 9 is a feature map (the feature map is a simplified diagram of the feature map generated in fig. 7, and is not in the form in fig. 9 for explaining the logic value calculation process), and rectangles counted from left to right in the feature map in fig. 9 are respectively a channel, such as the first channel, the second channel, the third channel, and the fourth channel in fig. 9. Calculating the feature value on the rightmost channel through a global pooling layer to obtain the maximum value (or average value) of the feature value on the channel, namely obtaining the value of a block 1 (the target feature value of the rightmost channel), calculating the feature value on the second channel from the right through the global pooling layer to obtain the value of a block 2 (the target feature value of the second channel from the right), calculating the target feature value of each channel (such as the values of the block 1, the block 2, the block 3 and the block 4 in fig. 9) by analogy, and performing weighted summation on the target feature values of each channel to obtain the logit value (namely the value of the block 5) corresponding to the feature map.
It should be noted that, when the global pooling layer is calculated, whether the maximum value or the average value of the feature values on each channel is obtained is determined according to the structure of the global pooling layer. If the structure of the global pooling layer is a maximum value structure, the maximum value of the feature values on each channel is obtained, and if the structure of the global pooling layer is an average value structure, the average value of the feature values on each channel is obtained. The specific structure of the global pooling layer may be set according to the user's needs, and is not limited herein.
After the logic value corresponding to the characteristic diagram is obtained, the logic value is normalized by a sigmoid function, so that a value between 0 and 1 can be obtained, and the value between 0 and 1 is the probability value of the tuberculosis.
The preset probability threshold may be a preset probability threshold determined as the pulmonary tuberculosis, and when the probability value of the pulmonary tuberculosis obtained based on the to-be-trained pulmonary image processing model is greater than the preset probability threshold, it is determined that the pulmonary tuberculosis focus exists in the to-be-processed pulmonary image.
The first loss function may be a loss function of the pulmonary image processing model to be trained, which is obtained according to a probability value of the pulmonary tuberculosis obtained by the pulmonary image processing model to be trained, a preset probability threshold value and historical labeling information (specifically, a labeling mask).
The first preset loss threshold may be a preset threshold of the first loss function, and when the value of the first loss function is smaller than the first preset loss threshold, it is proved that the training of the lung image processing model to be trained is completed.
After obtaining the probability value of the pulmonary tuberculosis, according to the probability value, a preset probability threshold and historical labeling information (specifically, a labeling mask), a first loss function of the pulmonary image processing model to be trained can be obtained. And when the first loss function is smaller than a first preset loss threshold value, the training of the lung image processing model to be trained is proved to be finished.
Therefore, the probability value of the pulmonary tuberculosis in the lung image to be processed can be quickly and accurately obtained based on the trained lung image processing model.
Optionally, after obtaining the feature map corresponding to the historical chest orthophoto image, the method further includes: performing channel compression on the feature diagram to obtain a logic diagram corresponding to the feature diagram; calculating each region on the logic diagram to obtain a probability value of each region, and obtaining a thermodynamic diagram of the region of interest in the historical chest orthophoto image based on the probability value of each region; determining a second loss function of a pulmonary image processing model to be trained based on the region of interest thermodynamic diagram and a historical region of interest thermodynamic diagram corresponding to the historical chest orthophoto image; and when the second loss function is smaller than a second preset loss threshold value, determining that the training of the lung image processing model to be trained is finished.
Illustratively, the logic map (logitmap) may be an image formed after channel compression of the feature map.
Specifically, the feature map is input into a full-connected layer (fc layer) of the lung image model to be trained, so that a logitmap after channel compression is obtained.
In a specific process, referring to the schematic diagram generated by the logic diagram illustrated in fig. 10, fig. 10 takes 4 channels as an example, the leftmost diagram in fig. 10 is a feature diagram, and rectangles counted from left to right in the feature diagram in fig. 10 are respectively different channels, such as the first channel, the second channel, the third channel, and the fourth channel in fig. 10. The 1 st data (i.e., the data at the upper left corner of each channel, such as the data of the dashed line in fig. 10) on each channel is weighted and summed to obtain the first block (block M) in the subsequent logtmap, and so on, the corresponding data of each channel is weighted and summed to obtain the subsequent logtmap.
In an embodiment of the present invention, each small square in the logitmap represents a small region in the historical chest orthophoto image.
After obtaining the logitmap, each region (i.e. each small square in fig. 10) in the logitmap is calculated to obtain a probability value of each region, wherein the probability value represents the probability value that the region in the historical chest orthophoto image corresponding to the region (small square) is tuberculosis.
And obtaining a region-of-interest thermodynamic diagram (tuberculosis thermodynamic diagram) of the historical chest orthophoto image according to the probability values of the regions.
In the embodiment of the invention, the thermodynamic diagram can be drawn through the thermodynamic diagram rendering module according to the probability value of each region. In the thermodynamic diagram rendering process, the higher the probability value of the region, the more red the color of the region is.
In another embodiment, instead of obtaining the logit value in the manner described above with reference to FIG. 9, the logit value can also be obtained using a logmap. Specifically, the maximum value in logitmap obtained in fig. 10 may be used as the logit value.
The two ways of obtaining the logic value can be selected according to the user requirement, and are not limited herein.
The second loss function may be a loss function of the pulmonary image processing model to be trained obtained from a region-of-interest thermodynamic diagram (tuberculosis thermodynamic diagram) obtained from the pulmonary image processing model to be trained and historical labeling information.
The second preset loss threshold may be a preset threshold of a second loss function, and when the value of the second loss function is smaller than the second preset loss threshold, it is proved that the training of the lung image processing model to be trained is completed.
After obtaining an interesting region thermodynamic diagram corresponding to the historical chest orthophoto image, determining a second loss function of the lung image processing model to be trained according to the interesting region thermodynamic diagram and historical annotation information corresponding to the historical chest orthophoto image; and when the second loss function is smaller than a second preset loss threshold value, determining that the training of the lung image processing model to be trained is finished.
Referring to the schematic diagram for comparing the historical labeling information and the thermodynamic diagram of the region of interest described in fig. 11, the left image in fig. 11 is the historical labeling information, the right image is the thermodynamic diagram of the region of interest corresponding to the historical labeling information, and the region a in fig. 11 is the region of interest, that is, the tuberculosis region.
And comparing the two images in the image 11, calculating a second loss function, and determining that the training of the lung image processing model to be trained is completed when the second loss function is smaller than a second preset loss threshold. Therefore, the thermodynamic diagram of the region of interest of the lung image to be processed can be quickly and accurately obtained based on the trained lung image processing model.
It should be noted that, in the embodiment of the present invention, the first loss function and the second loss function are used together to optimize the lung image processing model. And (3) calculating loss (namely a first loss function and a second loss function) simultaneously, and only when the loss functions of the two are respectively smaller than the corresponding loss threshold values, the completion of the training of the lung image processing model can be proved.
In another embodiment, two loss functions (i.e., the first loss function and the second loss function) may be added to obtain a total loss function, and the total loss function is smaller than a preset loss threshold, which proves that the training of the lung image processing model is complete.
In the embodiment of the invention, after the training of the lung image processing model is completed, the trained lung image processing model can be evaluated to evaluate whether the lung image processing model is feasible or not.
In the embodiment of the present invention, when evaluating the trained lung image processing model, the available evaluation indexes are: accuracy, sensitivity and specificity. When the sensitivity is greater than the preset sensitivity threshold and the specificity is greater than the preset specificity threshold, for example, the sensitivity is greater than or equal to 90%, and the specificity is greater than 80%, the lung image processing model can be considered to have a good evaluation effect, and can be well used for subsequent image processing. The training mode is as follows: the historical images are divided into a training set and a verification set according to a ratio of 9:1, and the lung image processing model is verified by using the verification set. Specifically, the ROC curve shown in fig. 12 can be used to evaluate the lung image processing model.
And S230, acquiring a lung image to be processed.
S240, inputting the lung image to be processed into the trained lung image processing model to obtain a probability value determined as the pulmonary tuberculosis and a corresponding thermodynamic diagram of the region of interest in the lung image to be processed; the lung image processing model is obtained based on training of multiple groups of historical images, and the historical images comprise historical chest positive images and historical marking information corresponding to the historical chest positive images.
According to the technical scheme of the embodiment of the invention, the obtained multiple groups of historical chest positive bit images and the historical marking information corresponding to the historical chest positive bit images are input into the lung image processing model to be trained, and the lung image processing model to be trained is subjected to iterative training to obtain the trained lung image processing model, so that the lung image to be processed can be processed based on the trained lung image processing model to obtain the probability value of tuberculosis and the corresponding thermodynamic diagram of the region of interest.
EXAMPLE III
Fig. 13 is a flowchart of a lung image processing method according to a third embodiment of the present invention, and the third embodiment of the present invention may be combined with various alternatives in the foregoing embodiments. In this embodiment of the present invention, optionally, before the inputting the plurality of sets of historical chest orthophotos images and the historical annotation information corresponding to the historical chest orthophotos images into the lung image processing model to be trained, the method further includes: for any group of acquired historical chest orthophoto images and historical annotation information corresponding to the historical chest orthophoto images, the following steps are executed: preprocessing the historical chest positive image and the historical marking information to obtain a target historical chest positive image and target historical marking information; correspondingly, the inputting the plurality of groups of historical chest positive images and the historical marking information corresponding to the historical chest positive images into the lung image processing model to be trained comprises: and inputting a plurality of groups of historical chest positive images of the target and historical marking information of the target corresponding to the historical chest positive images of the target into a lung image processing model to be trained.
As shown in fig. 13, the method of the embodiment of the present invention specifically includes the following steps:
s310, acquiring a plurality of groups of historical chest positive images and historical annotation information corresponding to the historical chest positive images, wherein the historical annotation information is information obtained after the pulmonary tuberculosis in the historical chest positive images is annotated.
S320, preprocessing the historical chest orthophoto image and the historical annotation information corresponding to the historical chest orthophoto image for any group of acquired historical chest orthophoto images and historical annotation information to obtain a target historical chest orthophoto image and target historical annotation information.
Illustratively, for any group of historical chest orthophoto images and historical annotation information corresponding to the historical chest orthophoto images, preprocessing is required to ensure that the images input into the lung image processing model to be trained are images meeting the requirements of the lung image processing model to be trained.
Optionally, the preprocessing is performed on the historical chest orthophoto image and the historical annotation information, and specifically, the preprocessing may be: converting the data format of the historical chest orthophoto image into a target data format required by a lung image processing model to be trained; carrying out histogram equalization on the history chest orthophoto image after format conversion; respectively zooming the equalized historical chest orthophoto image and the historical annotation information to a target size; converting the channel number of the zoomed historical chest orthophoto image into a target channel number required by a lung image processing model to be trained; and subtracting a preset pixel value from the pixel value of each pixel point of the history chest positive image after channel conversion to obtain a target history chest positive image and target history annotation information.
Illustratively, the target data format may be the data format of the images input into it as required by the lung image processing model to be trained.
The target size may be the size of the image input into the lung image processing model to be trained.
The target number of channels may be the number of channels of images input into the lung image processing model to be trained that are required.
The target historical chest positive image may be an image resulting from pre-processing the historical chest positive image.
The target historical annotation information may be an image obtained by preprocessing the historical annotation information.
The preset pixel value may be a preset pixel value, and the preset pixel value may be within a middle value range of the pixel values. For example, the pixel value of the image ranges from 0 to 255, and may be 125 at a preset pixel value.
For any group of historical chest positive image and historical annotation information corresponding to the historical chest positive image, firstly, the data format of the historical chest positive image needs to be converted into a target data format required by a lung image processing model to be trained.
Secondly, histogram equalization is carried out on the historical chest positive image after format conversion, so that the contrast of the historical chest positive image after format conversion is adjusted, the contrast is clear, and the image is clear.
It should be noted that whether the histogram equalization is required to be performed or not can be set according to the user's requirement, because the contrast of some historical chest radiograph images is good and the histogram equalization is not required, the histogram equalization may not be performed on the historical chest radiograph images after format conversion, and the histogram equalization may be specifically selected according to the requirement, which is not limited herein.
And then, scaling the historical chest orthophoto image and the historical annotation information after histogram equalization to the size of the image input into the lung image processing model to be trained. For example, it may be scaled to 1024 x 1024.
It should be noted that, the chest orthophoto image is generally rectangular and has a long side and a short side, and it is necessary to ensure that the chest orthophoto image input into the lung image processing model is square, so it is necessary to perform a compensation operation on the short side portion, and after the compensation operation is completed, the chest orthophoto image and the labeling mask are scaled to ensure that the sizes of all images input into the lung image processing model are consistent.
After the historical labeling information is zoomed, the target historical labeling information which is finally input into the lung image processing model can be obtained.
And then, converting the channel number of the zoomed historical chest orthophoto image into the channel number of the image input into the lung image processing model to be trained. For example, it may be that a single channel of the scaled historical chest orthophoto image is converted into three channels, respectively.
And finally, subtracting a preset pixel value from the pixel value of each pixel point on the historical chest positive image after channel conversion to obtain the target historical chest positive image.
In the embodiment of the invention, the preset pixel value is subtracted from the pixel value of each pixel point on the historical chest orthophoto image after channel conversion, so that the average value of the historical chest orthophoto image pixels after channel conversion is ensured to be near 0, and the problem of overlarge parameter gradient during deep learning optimization caused by overlarge image pixel average value is prevented.
The historical chest positive image and the historical marking information are preprocessed, so that the target historical chest positive image and the target historical marking information which have better quality and meet the requirement of the lung image processing model to be trained can be obtained, and the training quality of the lung image processing model is further ensured.
S330, inputting a plurality of groups of historical chest positive images of the target and historical target labeling information corresponding to the historical chest positive images of the target into a lung image processing model to be trained, and performing iterative training on the lung image processing model to be trained to obtain the trained lung image processing model.
Illustratively, after the historical chest orthophoto images and the historical target annotation information are obtained, a plurality of groups of historical chest orthophoto images and historical target annotation information are input into a lung image processing model to be trained, and the lung image processing model to be trained is subjected to iterative training to obtain a trained lung image processing model.
And S340, acquiring a lung image to be processed.
S350, inputting the lung image to be processed into the trained lung image processing model to obtain a probability value determined as the pulmonary tuberculosis and a corresponding thermodynamic diagram of the region of interest in the lung image to be processed; the lung image processing model is obtained based on training of multiple groups of historical images, and the historical images comprise historical chest positive images and historical marking information corresponding to the historical chest positive images.
In another embodiment, the lung image processing method of the embodiment of the present invention may be implemented based on a PyTorch framework, and the lung image processing method is deployed in a browser cloud, so that an end user only needs to meet a browser version requirement and a network bandwidth requirement to implement the lung image processing method of the embodiment of the present invention.
In another embodiment, in the process of implementing the technical solution of the embodiment of the present invention, if an error occurs, for example, an error occurs in the execution of an algorithm or the like, an error prompt may be fed back, so that an engineer may take corresponding measures according to the error prompt.
According to the technical scheme of the embodiment of the invention, the historical chest positive film image and the historical marking information are preprocessed by any group of acquired historical chest positive film images and the historical marking information corresponding to the historical chest positive film images, so that the target historical chest positive film image and the target historical marking information which have better quality and meet the requirement of the lung image processing model to be trained can be obtained, and the training quality of the lung image processing model is further ensured.
Example four
Fig. 14 is a schematic structural diagram of a lung image processing apparatus according to a fourth embodiment of the present invention, as shown in fig. 14, the apparatus includes: an image acquisition module 31 and a tuberculosis determination module 32.
The image acquiring module 31 is configured to acquire a lung image to be processed;
a tuberculosis determining module 32, configured to input the to-be-processed lung image into a trained lung image processing model, so as to obtain a probability value determined as tuberculosis and a corresponding region of interest thermodynamic diagram in the to-be-processed image; the lung image processing model is obtained based on training of multiple groups of historical images, and the historical images comprise historical chest orthophotos images and historical annotation information corresponding to the historical chest orthophotos images.
On the basis of the technical scheme of the embodiment of the invention, the device also comprises:
the system comprises a historical image acquisition module, a data processing module and a data processing module, wherein the historical image acquisition module is used for acquiring a plurality of groups of historical chest positive images and historical marking information corresponding to the historical chest positive images, and the historical marking information is information obtained after marking the tuberculosis in the historical chest positive images;
and the model training module is used for inputting a plurality of groups of historical chest positive images and historical marking information corresponding to the historical chest positive images into the lung image processing model to be trained, and performing iterative training on the lung image processing model to be trained to obtain the trained lung image processing model.
On the basis of the technical scheme of the embodiment of the invention, the model training module comprises:
the feature map determining unit is used for obtaining a feature map corresponding to a historical chest orthophoto image based on the historical chest orthophoto image for any group of historical chest orthophoto images input into a lung image processing model to be trained and historical annotation information corresponding to the historical chest orthophoto images;
the target characteristic value determining unit is used for calculating the characteristic value of each channel in the characteristic diagram to obtain the target characteristic value of each channel for any group of historical chest positive images input into the lung image processing model to be trained and the historical marking information corresponding to the historical chest positive images;
a tuberculosis probability value determining unit, configured to perform weighted summation on each target feature value for any group of historical chest positive slice images input into the lung image processing model to be trained and historical labeling information corresponding to the historical chest positive slice images to obtain a logic value of the feature map, and determine a probability value of tuberculosis based on the logic value;
the first loss function determining unit is used for determining a first loss function of the lung image processing model to be trained based on the probability value, a preset probability threshold and historical marking information corresponding to the historical chest orthophoto image;
and the first determination model training completion unit is used for determining that the training of the lung image processing model to be trained is completed when the first loss function is smaller than a first preset loss threshold value.
On the basis of the technical scheme of the embodiment of the invention, the tuberculosis probability value determining unit comprises:
and the normalization subunit is used for normalizing the logic value and determining the normalized logic value as the probability value of the tuberculosis.
On the basis of the technical scheme of the embodiment of the invention, the model training module comprises:
the logic diagram determining unit is used for performing channel compression on the feature diagram for any group of historical chest positive slice images input into the lung image processing model to be trained and historical marking information corresponding to the historical chest positive slice images to obtain a logic diagram corresponding to the feature diagram;
the interesting region thermodynamic diagram determining unit is used for calculating each region on the logic diagram according to any group of historical chest orthophoric images input into the lung image processing model to be trained and historical marking information corresponding to the historical chest orthophoric images to obtain a probability value of each region, and obtaining an interesting region thermodynamic diagram in the historical chest orthophoric images based on the probability value of each region;
a second loss function determination unit, configured to determine a second loss function of the lung image processing model to be trained based on the region of interest thermodynamic diagram and historical annotation information corresponding to the historical chest orthophoto image;
and the second determination model training completion unit is used for determining that the training of the lung image processing model to be trained is completed when the second loss function is smaller than a second preset loss threshold value.
On the basis of the technical scheme of the embodiment of the invention, the device also comprises:
the preprocessing module is used for preprocessing the historical chest positive film images and the historical marking information corresponding to any group of acquired historical chest positive film images to obtain target historical chest positive film images and target historical marking information;
correspondingly, the model training module is specifically configured to:
and inputting a plurality of groups of historical chest positive images of the target and the historical target labeling information corresponding to the historical chest positive images of the target into a lung image processing model to be trained, and performing iterative training on the lung image processing model to be trained to obtain the trained lung image processing model.
On the basis of the technical scheme of the embodiment of the invention, the preprocessing module comprises:
the data format conversion unit is used for converting the data format of the historical chest orthophoto image into a target data format required by a lung image processing model to be trained;
the histogram equalization unit is used for performing histogram equalization on the historical chest orthophoto image after format conversion;
the scaling unit is used for scaling the equalized historical chest orthophoto image and the historical annotation information to a target size respectively;
the channel number conversion unit is used for converting the channel number of the zoomed historical chest orthophoto image into the target channel number required by the lung image processing model to be trained;
and the pixel value determining unit is used for subtracting a preset pixel value from the pixel value of each pixel point of the history chest positive film image after channel conversion to obtain a target history chest positive film image and target history marking information.
The lung image processing device provided by the embodiment of the invention can execute the lung image processing method provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method.
EXAMPLE five
Fig. 15 is a schematic structural diagram of an electronic apparatus according to a fifth embodiment of the present invention, as shown in fig. 15, the electronic apparatus includes a processor 70, a memory 71, an input device 72, and an output device 73; the number of the processors 70 in the electronic device may be one or more, and one processor 70 is taken as an example in fig. 15; the processor 70, the memory 71, the input device 72 and the output device 73 in the electronic apparatus may be connected by a bus or other means, and the bus connection is exemplified in fig. 15.
The memory 71 serves as a computer-readable storage medium for storing software programs, computer-executable programs, and modules, such as program instructions/modules (e.g., the image acquisition module 31 and the tuberculosis determination module 32) corresponding to the lung image processing method in the embodiment of the present invention. The processor 70 executes various functional applications of the electronic device and data processing, i.e., implements the lung image processing method described above, by executing software programs, instructions, and modules stored in the memory 71.
The memory 71 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory 71 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, the memory 71 may further include memory located remotely from the processor 70, which may be connected to the electronic device through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input device 72 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function controls of the electronic apparatus. The output device 73 may include a display device such as a display screen.
EXAMPLE six
An embodiment of the present invention also provides a storage medium containing computer-executable instructions for performing a method of lung image processing when executed by a computer processor.
Of course, the embodiment of the present invention provides a storage medium containing computer-executable instructions, and the computer-executable instructions are not limited to the operations of the method described above, and may also perform related operations in the lung image processing method provided in any embodiment of the present invention.
From the above description of the embodiments, it is obvious for those skilled in the art that the present invention can be implemented by software and necessary general hardware, and certainly, can also be implemented by hardware, but the former is a better embodiment in many cases. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, and includes instructions for enabling a computer electronic device (which may be a personal computer, a server, or a network device) to execute the methods according to the embodiments of the present invention.
It should be noted that, in the embodiment of the lung image processing apparatus, the units and modules included in the embodiment are merely divided according to the functional logic, but are not limited to the above division as long as the corresponding functions can be implemented; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (10)
1. A method of lung image processing, comprising:
acquiring a lung image to be processed;
inputting the lung image to be processed into a trained lung image processing model to obtain a probability value determined as tuberculosis and a corresponding region of interest thermodynamic diagram in the lung image to be processed; the lung image processing model is obtained based on training of multiple groups of historical images, and the historical images comprise historical chest orthophotos images and historical annotation information corresponding to the historical chest orthophotos images.
2. The method of claim 1, wherein the training method of the lung image processing model comprises:
acquiring a plurality of groups of historical chest positive images and historical marking information corresponding to the historical chest positive images, wherein the historical marking information is information obtained after marking the tuberculosis in the historical chest positive images;
and inputting a plurality of groups of historical chest positive images and historical marking information corresponding to the historical chest positive images into the lung image processing model to be trained, and performing iterative training on the lung image processing model to be trained to obtain the trained lung image processing model.
3. The method of claim 2, wherein iteratively training the lung image processing model to be trained to obtain a trained lung image processing model comprises:
for any group of historical chest positive images input into the lung image processing model to be trained and historical annotation information corresponding to the historical chest positive images, the following steps are executed, and the lung image processing model to be trained is subjected to iterative training:
obtaining a feature map corresponding to the historical chest orthophoto image based on the historical chest orthophoto image;
calculating the characteristic value of each channel in the characteristic diagram to obtain a target characteristic value of each channel;
carrying out weighted summation on the target characteristic values to obtain a logic value of the characteristic diagram, and determining a probability value of tuberculosis based on the logic value;
determining a first loss function of a lung image processing model to be trained based on the probability value, a preset probability threshold and historical labeling information corresponding to the historical chest orthophoto image;
and when the first loss function is smaller than a first preset loss threshold value, determining that the training of the lung image processing model to be trained is finished.
4. The method of claim 3, wherein determining a probability value for tuberculosis based on the logical value comprises:
and normalizing the logic value, and determining the normalized logic value as the probability value of the tuberculosis.
5. The method of claim 3, wherein after the obtaining the feature map corresponding to the historical chest orthophoto image, the method further comprises:
performing channel compression on the feature diagram to obtain a logic diagram corresponding to the feature diagram;
calculating each region on the logic diagram to obtain a probability value of each region, and obtaining a thermodynamic diagram of the region of interest in the historical chest orthophoto image based on the probability value of each region;
determining a second loss function of the lung image processing model to be trained based on the region of interest thermodynamic diagram and historical annotation information corresponding to the historical chest orthophoto image;
and when the second loss function is smaller than a second preset loss threshold value, determining that the training of the lung image processing model to be trained is finished.
6. The method of claim 2, wherein prior to the inputting the sets of historical chest orthophotos images and historical annotation information corresponding to the historical chest orthophotos images into the lung image processing model to be trained, the method further comprises:
for any group of acquired historical chest orthophoto images and historical annotation information corresponding to the historical chest orthophoto images, the following steps are executed:
preprocessing the historical chest positive image and the historical marking information to obtain a target historical chest positive image and target historical marking information;
correspondingly, the inputting the plurality of groups of historical chest positive images and the historical marking information corresponding to the historical chest positive images into the lung image processing model to be trained comprises:
and inputting the plurality of groups of historical chest positive images and the historical target labeling information corresponding to the historical chest positive images into a lung image processing model to be trained.
7. The method of claim 6, wherein the pre-processing the historical chest orthophoto images and the historical annotation information comprises:
converting the data format of the historical chest orthophoto image into a target data format required by a lung image processing model to be trained;
carrying out histogram equalization on the history chest orthophoto image after format conversion;
respectively zooming the equalized historical chest orthophoto image and the historical annotation information to a target size;
converting the channel number of the zoomed historical chest orthophoto image into a target channel number required by a lung image processing model to be trained;
and subtracting a preset pixel value from the pixel value of each pixel point of the history chest positive image after channel conversion to obtain a target history chest positive image and target history annotation information.
8. A lung image processing apparatus characterized by comprising:
the image acquisition module is used for acquiring a lung image to be processed;
a pulmonary tuberculosis determining module, configured to input the to-be-processed lung image into a trained lung image processing model, so as to obtain a probability value determined as pulmonary tuberculosis and a corresponding thermodynamic diagram of a region of interest in the to-be-processed lung image; the lung image processing model is obtained based on training of multiple groups of historical images, and the historical images comprise historical chest orthophotos images and historical annotation information corresponding to the historical chest orthophotos images.
9. An electronic device, characterized in that the electronic device comprises:
one or more processors;
storage means for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the method of image processing of the lungs of any one of claims 1-7.
10. A storage medium containing computer-executable instructions for performing the method of lung image processing according to any one of claims 1-7 when executed by a computer processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011345196.XA CN112435242A (en) | 2020-11-25 | 2020-11-25 | Lung image processing method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011345196.XA CN112435242A (en) | 2020-11-25 | 2020-11-25 | Lung image processing method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112435242A true CN112435242A (en) | 2021-03-02 |
Family
ID=74699219
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011345196.XA Pending CN112435242A (en) | 2020-11-25 | 2020-11-25 | Lung image processing method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112435242A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104050451A (en) * | 2014-06-18 | 2014-09-17 | 西北农林科技大学 | Robust target tracking method based on multi-channel Haar-like characteristics |
| CN108510482A (en) * | 2018-03-22 | 2018-09-07 | 姚书忠 | Cervical carcinoma detection method, device, equipment and medium based on gynecatoptron image |
| CN109858429A (en) * | 2019-01-28 | 2019-06-07 | 北京航空航天大学 | A kind of identification of eye fundus image lesion degree and visualization system based on convolutional neural networks |
| US20200012904A1 (en) * | 2018-07-03 | 2020-01-09 | General Electric Company | Classification based on annotation information |
| CN110930373A (en) * | 2019-11-06 | 2020-03-27 | 天津大学 | Pneumonia recognition device based on neural network |
| CN111325739A (en) * | 2020-02-28 | 2020-06-23 | 北京推想科技有限公司 | Method and device for detecting lung focus and training method of image detection model |
-
2020
- 2020-11-25 CN CN202011345196.XA patent/CN112435242A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104050451A (en) * | 2014-06-18 | 2014-09-17 | 西北农林科技大学 | Robust target tracking method based on multi-channel Haar-like characteristics |
| CN108510482A (en) * | 2018-03-22 | 2018-09-07 | 姚书忠 | Cervical carcinoma detection method, device, equipment and medium based on gynecatoptron image |
| US20200012904A1 (en) * | 2018-07-03 | 2020-01-09 | General Electric Company | Classification based on annotation information |
| CN109858429A (en) * | 2019-01-28 | 2019-06-07 | 北京航空航天大学 | A kind of identification of eye fundus image lesion degree and visualization system based on convolutional neural networks |
| CN110930373A (en) * | 2019-11-06 | 2020-03-27 | 天津大学 | Pneumonia recognition device based on neural network |
| CN111325739A (en) * | 2020-02-28 | 2020-06-23 | 北京推想科技有限公司 | Method and device for detecting lung focus and training method of image detection model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7058373B2 (en) | Lesion detection and positioning methods, devices, devices, and storage media for medical images | |
| CN110232719B (en) | Medical image classification method, model training method and server | |
| US10810735B2 (en) | Method and apparatus for analyzing medical image | |
| CN111325739B (en) | Method and device for detecting lung focus and training method of image detection model | |
| US11900647B2 (en) | Image classification method, apparatus, and device, storage medium, and medical electronic device | |
| CN110232383B (en) | Focus image recognition method and focus image recognition system based on deep learning model | |
| CN111524106A (en) | Skull fracture detection and model training method, device, equipment and storage medium | |
| CN110176002B (en) | Focus detection method of X-ray image and terminal device | |
| WO2023137914A1 (en) | Image processing method and apparatus, electronic device, and storage medium | |
| CN108510489B (en) | Pneumoconiosis detection method and system based on deep learning | |
| WO2022179083A1 (en) | Image detection method and apparatus, and device, medium and program | |
| CN112884782B (en) | Biological object segmentation method, apparatus, computer device, and storage medium | |
| CN111754453A (en) | Pulmonary tuberculosis detection method and system based on chest radiography image and storage medium | |
| CN110766670A (en) | Mammary gland molybdenum target image tumor localization algorithm based on deep convolutional neural network | |
| CN111062947A (en) | Deep learning-based X-ray chest radiography focus positioning method and system | |
| CN112836653A (en) | Face privacy method, device and apparatus and computer storage medium | |
| CN110807409A (en) | Crowd density detection model training method and crowd density detection method | |
| CN111784686A (en) | Dynamic intelligent detection method, system and readable storage medium for endoscope bleeding area | |
| US20230115927A1 (en) | Systems and methods for plaque identification, plaque composition analysis, and plaque stability detection | |
| CN112435242A (en) | Lung image processing method and device, electronic equipment and storage medium | |
| CN115880358A (en) | Construction method of positioning model, positioning method of image mark points and electronic equipment | |
| WO2022227193A1 (en) | Liver region segmentation method and apparatus, and electronic device and storage medium | |
| CN112750124B (en) | Model generation method, image segmentation method, model generation device, image segmentation device, electronic equipment and storage medium | |
| CN113643263A (en) | Identification method and system for upper limb bone positioning and forearm bone fusion deformity | |
| CN110852377A (en) | Method for identifying pneumonia according to lung X-ray film |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210302 |
|
| RJ01 | Rejection of invention patent application after publication |