CN114332844B - Intelligent classification application method, device, equipment and storage medium of medical image - Google Patents
Intelligent classification application method, device, equipment and storage medium of medical image Download PDFInfo
- Publication number
- CN114332844B CN114332844B CN202210255096.0A CN202210255096A CN114332844B CN 114332844 B CN114332844 B CN 114332844B CN 202210255096 A CN202210255096 A CN 202210255096A CN 114332844 B CN114332844 B CN 114332844B
- Authority
- CN
- China
- Prior art keywords
- medical image
- endoscope
- images
- image
- character
- 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
Landscapes
- Endoscopes (AREA)
- Medical Treatment And Welfare Office Work (AREA)
- Image Analysis (AREA)
Abstract
The application provides a method, a device, equipment and a storage medium for intelligent classification application of medical images, wherein the method comprises the following steps: acquiring an endoscope medical image of an electronic endoscope; inputting the endoscope medical image into a trained character recognition model for character recognition, and outputting character sequence information of the endoscope medical image; analyzing the character sequence information to determine the medical image type of the endoscope medical image; and screening out a target detection program from a plurality of pre-stored detection programs according to the type of the medical image so as to control the electronic endoscope to execute the target detection program. By adopting the method, the classification accuracy and the analysis stability of the endoscope medical image can be improved, the clinical operation process can be simplified, and the detection cost of the artificial intelligence technology in the medical field can be saved.
Description
Technical Field
The embodiment of the application relates to the technical field of computer vision, in particular to an intelligent classification application method, device, equipment and storage medium for medical images.
Background
With the rapid development of artificial intelligence technology, computer vision technology centered on deep learning has attracted attention, and is gradually applied to the field of medicine, in particular to the field of medical endoscopes.
In the field of medical endoscopes, an endoscope (Endoscopy) is an optical instrument, is sent into the body from the outside of the body through a natural orifice of the human body to examine internal diseases, can directly observe pathological changes in the internal cavity of an organ, can determine the pathological change part and range through photography, and greatly improves the accuracy rate of cancerous judgment. However, in practical applications, the types of endoscopes used for examining different parts may be different, and AI detection programs and models for assisting lesion detection are also different.
Therefore, in order to avoid the problems of wrong use of different endoscope models in clinical examinations or wrong switch of AI detection programs, it is necessary to provide a method for assisting clinical examinations to correctly use endoscopes and accurately switch AI programs.
Disclosure of Invention
The application aims to provide an intelligent classification application method, device, equipment and storage medium of medical images, which are used for realizing accurate analysis of image types of the medical images by combining a deep learning technology, further realizing accurate selection of endoscope models and AI (artificial intelligence) detection programs by referring to the image types, and finally improving the analysis stability of the endoscope medical images.
In a first aspect, the present application provides a method for intelligently classifying and applying medical images, including:
acquiring an endoscope medical image of an electronic endoscope;
inputting the endoscope medical image into a trained character recognition model for character recognition, and outputting character sequence information of the endoscope medical image;
analyzing the character sequence information to determine the medical image type of the endoscope medical image;
and screening out a target detection program from a plurality of pre-stored detection programs according to the type of the medical image so as to control the electronic endoscope to execute the target detection program.
In some embodiments of the present application, analyzing the character sequence information to determine a medical image type to which the endoscopic medical image belongs comprises: if the character sequence information contains preset first character information, determining that the medical image type of the endoscope medical image is a first medical image; if the character sequence information contains preset second character information, determining that the medical image type of the endoscope medical image is a second medical image; wherein the first character information includes 'CF-H' and/or 'EC-6', and the second character information includes at least one of 'GIF-H', 'EG-L', and 'EG-7'.
In some embodiments of the present application, the endoscopic medical images include at least two endoscopic medical images, and the method for applying intelligent classification of medical images further includes: determining a first image number of endoscopic medical images belonging to the first medical image; and determining a second number of images of the endoscopic medical image that are attributed to the second medical image; if the number of the first images is larger than or equal to a preset number threshold, determining that the medical image type of the endoscopic medical image is the first medical image; and if the number of the second images is larger than or equal to the number threshold value, determining that the medical image type of the endoscopic medical image is the second medical image.
In some embodiments of the present application, the method for applying intelligent classification to medical images further comprises: and if the number of the first images and the number of the second images are both larger than or equal to the number threshold, or if the number of the first images and the number of the second images are both smaller than the number threshold, generating error prompt information, wherein the error prompt information is used for prompting that the endoscope medical images are obtained again for classification.
In some embodiments of the present application, the detection program includes an enteroscope detection program and a gastroscope detection program, and the target detection program is screened out from a plurality of pre-stored detection programs according to the medical image type to control the electronic endoscope to execute the target detection program, including: if the medical image type is a first medical image, acquiring first residual characters except the first character information in the character sequence information, and screening target enteroscopy detection programs from various enteroscopy detection programs according to the first residual characters to serve as target detection programs; if the medical image type is a second medical image, second residual characters except the second character information in the character sequence information are obtained, and target gastroscope detection programs are screened out from various gastroscope detection programs according to the second residual characters and serve as target detection programs; and controlling the electronic endoscope to execute the target detection program.
In some embodiments of the present application, the trained character recognition model includes a feature extraction network, a sequence labeling network, and a sequence conversion network; wherein, carry out character recognition in inputting scope medical image to the character recognition model that has trained, output scope medical image's character sequence information includes: inputting the endoscope medical image into a trained character recognition model, and performing feature extraction on the endoscope medical image through a feature extraction network to obtain image features; carrying out sequence marking on the image characteristics through a sequence marking network to obtain a character label; and performing sequence conversion on the character tags through a sequence conversion network to obtain character sequence information.
In some embodiments of the present application, before inputting the endoscopic medical image into the trained character recognition model for character recognition, the method further includes: constructing an initial character recognition model; the initial character recognition model is composed of a feature extraction network, a sequence marking network and a sequence conversion network, wherein the feature extraction network comprises ResNet50, and the sequence marking network comprises a long-term and short-term memory network; acquiring a medical sample image set, and dividing the medical sample image set into a training set and a testing set, wherein the medical sample image set comprises a plurality of medical sample images marked with characters; using a training set to carry out primary training on the initial character recognition model to obtain a character recognition model after the primary training; and testing and adjusting the preliminarily trained character recognition model by using the test set to obtain the trained character recognition model.
In some embodiments of the present application, acquiring endoscopic medical images of an electronic endoscope comprises: acquiring an endoscope video of the electronic endoscope in a white light mode; performing frame extraction on an endoscope video to obtain at least two frames of endoscope images; and processing the images of the endoscopes according to a preset interception starting point and a preset screenshot size to obtain medical images of the endoscopes.
In some embodiments of the present application, image processing is performed on each endoscopic image according to a preset capture start point and a preset capture size to obtain an endoscopic medical image, including: according to a preset intercepting starting point and a preset screenshot size, image interception is carried out on each endoscope image to obtain at least two frames of candidate endoscope images; and performing pixel processing on each candidate endoscope image based on a preset binarization method to obtain an endoscope medical image.
In a second aspect, the present application provides an apparatus for intelligently classifying medical images, comprising:
the image acquisition module is used for acquiring an endoscope medical image of the electronic endoscope;
the character recognition module is used for inputting the endoscope medical image into the trained character recognition model for character recognition and outputting character sequence information of the endoscope medical image;
the type determining module is used for analyzing the character sequence information to determine the medical image type of the endoscope medical image;
and the program execution module is used for screening out a target detection program from a plurality of pre-stored candidate detection programs according to the type of the medical image so as to control the electronic endoscope to execute the target detection program.
In a third aspect, the present application further provides a medical image processing apparatus comprising:
one or more processors;
a memory; and one or more application programs, one or more of which are stored in the memory and configured to be executed by the processor to implement the method for intelligent classification application of medical images of the first aspect described above.
In a fourth aspect, the present application further provides a computer readable storage medium having stored thereon a computer program, which is loaded by a processor to perform the steps in the method for applying intelligent classification to medical images.
In a fifth aspect, embodiments of the present application provide a computer program product or a computer program comprising computer instructions stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the method provided by the first aspect.
According to the intelligent classification application method, the intelligent classification application device, the intelligent classification application equipment and the intelligent classification application storage medium of the medical images, the server acquires the endoscope medical images of the electronic endoscope, inputs the endoscope medical images into the trained character recognition model for character recognition, outputs the character sequence information of the endoscope medical images, namely analyzes the character sequence information to determine the medical image type of the endoscope medical images, and finally screens out the target detection program from a plurality of pre-stored detection programs according to the medical image type to control the electronic endoscope to execute the target detection program. According to the method and the device, the image types of the medical images are accurately analyzed by combining a deep learning technology, and then the detection program is accurately selected by referring to the image types, so that the classification accuracy and the analysis stability of the endoscope medical images can be improved, the clinical operation process can be simplified, and the detection cost of the artificial intelligence technology in the medical field is saved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a scene schematic diagram of an intelligent classification application method of medical images provided in an embodiment of the present application;
fig. 2 is a flowchart illustrating a method for applying intelligent classification to medical images provided in an embodiment of the present application;
FIG. 3 is an interface schematic of an endoscopic image provided in an embodiment of the present application;
FIG. 4 is an interface schematic of another endoscopic image provided in an embodiment of the present application;
FIG. 5 is a first schematic diagram of the extraction of endoscopic medical images provided in the embodiments of the present application;
FIG. 6 is a schematic diagram of the extraction of an endoscopic medical image provided in an embodiment of the present application;
FIG. 7 is a first schematic diagram illustrating the character recognition result of an endoscopic medical image provided in an embodiment of the present application;
FIG. 8 is a second schematic diagram illustrating the character recognition result of the endoscopic medical image provided in the embodiment of the present application;
FIG. 9 is a flowchart illustrating an embodiment of a method for applying intelligent classification to medical images;
fig. 10 is a schematic structural diagram of an apparatus for applying intelligent classification of medical images provided in an embodiment of the present application;
fig. 11 is a schematic structural diagram of a medical image processing apparatus provided in an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In the description of the present application, the terms "first", "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implying any number of technical features indicated. Thus, features defined as "first", "second", may explicitly or implicitly include one or more of the described features. In the description of the present application, "a plurality" means two or more unless specifically limited otherwise.
In the description of the present application, the term "for example" is used to mean "serving as an example, instance, or illustration". Any embodiment described herein as "for example" is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is presented to enable any person skilled in the art to make and use the invention. In the following description, details are set forth for the purpose of explanation. It will be apparent to one of ordinary skill in the art that the present invention may be practiced without these specific details. In other instances, well-known structures and processes are not shown in detail to avoid obscuring the description of the invention with unnecessary detail. Thus, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.
The scheme provided by the application relates to a computer vision technology, and is specifically explained by the following embodiments:
computer Vision technology (CV) is a science for researching how to make a machine "see", and further refers to that a camera and a Computer are used to replace human eyes to perform machine Vision such as identification, tracking and measurement on a target, and further image processing is performed, so that the Computer processing becomes an image more suitable for human eyes to observe or is transmitted to an instrument to detect. As a scientific discipline, computer vision research-related theories and techniques attempt to build artificial intelligence systems that can acquire information from images or multidimensional data. Computer vision technologies generally include image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content/behavior recognition, three-dimensional object reconstruction, 3D technologies, virtual reality, augmented reality, synchronous positioning, map construction, and other technologies, and also include common biometric technologies such as face recognition and fingerprint recognition.
The embodiment of the application provides an intelligent classification application method, an intelligent classification application device, intelligent classification application equipment and a storage medium for medical images, which are respectively described in detail below.
Referring to fig. 1, fig. 1 is a scene schematic diagram of an intelligent classification application method for medical images, which can be applied to an intelligent classification application system for medical images. The intelligent classification application system for medical images comprises a terminal 102 and a server 104. The terminal 102 may be a device that includes both receiving and transmitting hardware, i.e., a device having receiving and transmitting hardware capable of performing two-way communication over a two-way communication link. Such a device may include: a cellular or other communication device having a single line display or a multi-line display or a cellular or other communication device without a multi-line display. The terminal 102 may specifically be a desktop terminal or a mobile terminal, and the terminal 102 may also specifically be one of a mobile phone, a tablet computer, a notebook computer, an electronic endoscope, and other medical devices. The server 104 may be an independent server, or may be a server network or a server cluster composed of servers, which includes but is not limited to a computer, a network host, a single network server, a plurality of network server sets, or a cloud server composed of a plurality of servers. Among them, the Cloud server is constituted by a large number of computers or web servers based on Cloud Computing (Cloud Computing). In addition, the terminal 102 and the server 104 establish a communication connection through a network, which may specifically be any one of a wide area network, a local area network, and a metropolitan area network.
Those skilled in the art will appreciate that the application environment shown in fig. 1 is only one application scenario applicable to the present application, and does not constitute a limitation to the application scenario of the present application, and that other application environments may include more or less medical image processing devices than those shown in fig. 1, for example, only 1 server 104 is shown in fig. 1. It is to be understood that the system for applying intelligent classification to medical images may also include one or more other servers, which are not limited herein. In addition, the intelligent classification application system for medical images can further comprise a memory for storing data, such as storing endoscopic medical images.
It should be noted that the scene schematic diagram of the intelligent classification application system for medical images shown in fig. 1 is only an example, and the intelligent classification application system for medical images and the scene described in the embodiment of the present invention are for more clearly illustrating the technical solution of the embodiment of the present invention, and do not form a limitation on the technical solution provided in the embodiment of the present invention.
Referring to fig. 2, an embodiment of the present application provides an intelligent classification application method for medical images, which is mainly exemplified by applying the method to the server 104 in fig. 1, and the method includes steps S201 to S204, which are specifically as follows:
s201, acquiring an endoscope medical image of the electronic endoscope.
The Medical Image Analysis (Medical Image Analysis) according to the embodiment of the present application is a cross field of subjects such as comprehensive Medical images, mathematical modeling, digital Image processing and Analysis, artificial intelligence, and numerical algorithms. Medical images include, but are not limited to: CT images, PET (Positron Emission Tomography) images, MRI (Magnetic Resonance Imaging) images, and the like. The endoscopic medical image mainly refers to a medical image acquired by an electronic endoscope, but the embodiment of the present application does not limit the model, brand, and the like of the electronic endoscope.
In one implementation, the server 104 may acquire endoscopic medical images of the electronic endoscope in one of the following ways, for example: (1) acquiring endoscopic medical images from the terminal 102 or other device; (2) synchronously acquiring medical images of the endoscope from the link point servers of other blocks; (3) the endoscope medical image is obtained by request from an upper server or polling from a lower server. In addition, the endoscopic medical image may be an endoscopic medical image currently obtained by the server 104, or an endoscopic medical image specified by the user through the terminal 102, or an endoscopic medical image obtained by analyzing according to a preset program, and the analysis principle may be that an endoscopic medical image to be detected for a target portion can be obtained for a certain target portion, and the target portion may be any portion of biological tissue, such as a face, a hand, a foot, an intestinal tract, a stomach tube, and the like.
It should be noted that the present application describes exemplary gastrointestinal endoscopes in detail, including a lower gastrointestinal endoscope and an upper gastrointestinal endoscope, wherein the lower gastrointestinal tract refers to the intestinal tract and the upper gastrointestinal tract refers to the stomach, so that endoscopic medical images exist of endoscopic medical images of the lower gastrointestinal tract (as shown in fig. 3) and endoscopic medical images of the upper gastrointestinal tract (as shown in fig. 4). However, if the server 104 is used as a machine device, in practical applications, if one or more frames of endoscopic medical images are received, it is impossible to visually determine whether the images correspond to the upper digestive tract or the lower digestive tract as in a human being, so that the server 104 can be used for classifying the endoscopic medical images in advance, and the endoscope types can be distinguished by identifying the image content, so that the type of the currently received endoscopic medical image can be determined, which type of endoscope should be collected, and detailed analysis steps are described below.
In one embodiment, this step includes: acquiring an endoscope video of the electronic endoscope in a white light mode; performing frame extraction on an endoscope video to obtain at least two frames of endoscope images; and processing the images of the endoscopes according to a preset interception starting point and a preset screenshot size to obtain medical images of the endoscopes.
The endoscope video may be a video acquired by an electronic endoscope, or a video transmitted to the server 104 by other devices, such as the terminal 102, after being acquired by the electronic endoscope, and the video content relates to a target portion to be detected, which has been briefly described above and is not described herein again.
In a specific implementation, before the server 104 acquires the endoscope medical image, if the endoscope video to be detected is acquired, the video may be subjected to frame extraction by using tools such as OpenCV, ffmpeg, via, and the like to obtain the endoscope image with continuous frames. Therefore, the endoscopic image serving as the basis for subsequent analysis can be not only one frame, but also continuous frames; if the frames are consecutive, the analysis will be performed on a frame-by-frame basis.
Further, after the server 104 acquires the endoscopic image, in order to improve the accuracy of subsequent image classification, the endoscopic image may be preprocessed, including capturing an interested parameter region in the endoscopic image, and removing an irrelevant background image for the endoscopic medical image, and the detailed steps will be described below.
It should be noted that the light source mode mentioned in the embodiment of the present application refers to a light source mode activated by an electronic endoscope, and includes a general white light mode, the light source mode may be activated based on a preset program, and the activation program may be set in the terminal 102 or the server 104, and the present application is not limited in particular.
In one embodiment, the image processing of each endoscopic image according to the preset capture starting point and the preset capture size to obtain an endoscopic medical image comprises: according to a preset intercepting starting point and a preset screenshot size, image interception is carried out on each endoscope image to obtain at least two frames of candidate endoscope images; and performing pixel processing on each candidate endoscope image based on a preset binarization method to obtain an endoscope medical image.
The preset capture start point may be an image capture start point preset according to actual business requirements, for example, a coordinate in an endoscopic image before the image is not captured "
", may be used as the starting point for the truncation. It will be appreciated that to explicitly intercept the coordinate values of the starting points, the dimensions of the endoscopic image are first determined "
", so as to determine the specific coordinate value of the interception starting point.
The preset screenshot size may be an image capture size preset according to actual service requirements, for example, the capture size is "
", wherein
，
. Similarly, to explicitly intercept the coordinate value of a dimension, the dimension of the endoscopic image is first determined "
", the particular value of the intercept size can be determined.
In a specific implementation, the image preprocessing step performed by the server 104 includes the following steps: endoscopic image size in white light mode is "
", the embodiment of the application proposes that the interception starting point is based on the preset interception"
", and preset screenshot size"
'intercepting the inner mirror image to ensure the intercepted area to occupy the original size proportion'
，
"and then the captured region image is used as a candidate endoscopic image. Of course, if the scheme is applied to other endoscope models, the interception starting point and the interception size of the image interesting area can be set by analyzing the image content in a labor division manner, and the specific application is not limited.
Furthermore, the candidate endoscope images intercepted in the steps are all RGB color images, and each pixel point has maximum numerical information of '255 x 3', and the information is redundant. Therefore, the embodiment of the present application proposes that an image binarization processing method may be adopted to convert the captured RGB image into a black-and-white image, and each pixel has only two values, i.e., 0 or 255, so as to reduce the image data amount and increase the processing speed of the server 104 on each frame of image.
For example, referring to fig. 3 and fig. 4, fig. 3 is a schematic interface diagram of an endoscopic image according to an embodiment of the present application, that is, an endoscopic medical image acquired by an electronic endoscope with respect to a lower digestive tract; fig. 4 is a schematic interface diagram of another endoscopic medical image proposed in the embodiment of the present application, that is, an electronic endoscope is used to capture an endoscopic image of the upper digestive tract. After the server 104 acquires the endoscope image, image capture may be performed on the acquired endoscope image based on the preset capture start point and the preset capture size, and binarization processing may be performed to obtain the schematic diagrams shown in fig. 5 and/or fig. 6. By contrast, fig. 5 is an endoscopic medical image corresponding to the lower digestive tract, and fig. 6 is an endoscopic medical image corresponding to the upper digestive tract.
S202, inputting the endoscope medical image into the trained character recognition model for character recognition, and outputting character sequence information of the endoscope medical image.
The trained character recognition model can be composed of a deep learning-based CRNN (conditional recovery Neural network) and a ResNet50 network, and is called as a character recognition model because medical images acquired by electronic endoscopes of different models are adopted in advance for iterative training and have a character recognition function.
The character sequence information refers to character sequence information contained in equipment parameters of medical equipment, and the medical equipment comprises an electronic endoscope.
In a specific implementation, the intelligent classification application method for medical images provided by the application is mainly applicable to OLYMPUS (OLYMPUS) endoscope equipment and FUJIFILM (FUJIFILM) endoscope equipment. Since the character identification features of the image parameters acquired by the two endoscope devices are different, the classification of the medical images in the embodiment is substantially directed to the recognition classification of the character identification in the endoscope medical images. For example, referring to fig. 3 and 4, it can be seen that two endoscopic images have a large layout difference, and image classification of the endoscopic medical image in the application link can be realized by collecting a large number of endoscopic images of the two models to construct data sets and performing model training using the data sets. The model training steps involved in this embodiment will be described in detail below.
In one embodiment, before the step, the method further comprises: constructing an initial character recognition model; the initial character recognition model is composed of a feature extraction network, a sequence marking network and a sequence conversion network, wherein the feature extraction network comprises ResNet50, and the sequence marking network comprises a long-term and short-term memory network; acquiring a medical sample image set, and dividing the medical sample image set into a training set and a testing set, wherein the medical sample image set comprises a plurality of medical sample images marked with characters; using the training set to perform preliminary training on the initial character recognition model to obtain a preliminarily trained character recognition model; and testing and adjusting the preliminarily trained character recognition model by using the test set to obtain the trained character recognition model.
In specific implementation, the server 104 may first construct an initial character recognition model in response to a user requirement, where the model may select ResNet50 as a deep learning network structure to perform model training for CRNN-EndoType character recognition, and the loss function employs ctc (connectionist Temporal classification):
wherein the content of the first and second substances,
the length of the corresponding category is indicated,
representing the sum of the probabilities of all paths through node u at time t.
Specifically, referring to fig. 3 and 4, since the parameters shown in the parameter indication area of each of the images captured by the OLYMPUS (OLYMPUS) endoscope apparatus and the FUJIFILM (FUJIFILM) endoscope apparatus are sequences of combinations of english alphabets and numeric symbols, the embodiment of the present application proposes that the labeled image data set for identifying english and numeric symbols can be used as the medical sample image set to perform multiple rounds of iterative training of the character recognition model, so as to obtain the character recognition model with higher recognition accuracy.
Further, the server 104 may obtain a small number of medical sample images marked with characters (english and/or numbers) through the terminal 102, and the marking tool may be labellmg written based on Python language, which is supported in cross-platform operation in Windows, Linux, and the like, and may mark the specified character identifier by frame marking through a visual operation interface. Then, performing data amplification on a small number of medical sample images to obtain a large number of medical sample images to form a medical sample image set, wherein the data amplification method includes but is not limited to: and amplification modes such as optical transformation, affine transformation, Mosaic data enhancement and Mixup data enhancement. Therefore, the server 104 can perform data amplification based on a small number of samples to obtain a large number of samples, so as to realize the construction, training and debugging of the character recognition model, and finally obtain the trained character recognition model.
Still further, the stopping conditions for model training may include: 1. the error is less than some predetermined small value. 2. The weight change between two iterations is already small, a threshold value can be set, and when the weight change is smaller than the threshold value, the training is stopped. 3. A maximum number of iterations is set and training is stopped when the iterations exceed the maximum number, e.g. "273 cycles". 4. The classification accuracy reaches a predetermined large value.
In one embodiment, the trained character recognition model includes a feature extraction network, a sequence labeling network, and a sequence conversion network, and this step includes: inputting the endoscope medical image into a trained character recognition model, and performing feature extraction on the endoscope medical image through a feature extraction network to obtain image features; carrying out sequence marking on the image characteristics through a sequence marking network to obtain a character label; and performing sequence conversion on the character tags through a sequence conversion network to obtain character sequence information.
In specific implementation, in order to accurately distinguish endoscope medical images and further accurately judge how the currently used electronic endoscope model is switched and how the detection program is switched, the embodiment of the application can identify and distinguish the currently used endoscope model by identifying a parameter area projected by the electronic endoscope. In combination with the explanation of the above embodiment, since the endoscope models used for different part examinations are different and the imaging contents are also different, the endoscope models can be distinguished first based on image content identification, and then different detection programs are switched for different endoscope models, so that efficient, accurate and intelligent classification application of medical images can be realized.
Therefore, as a priority, the embodiment of the application proposes that a CRNN model is selected as a deep learning network structure to perform CRNN-EndoType model training, wherein a feature extraction part adopts ResNet50, a sequence marking part adopts a Long Short-Term Memory network (LSTM), so that image feature extraction can be performed on a currently acquired endoscopic medical image by using ResNet50 to obtain image features, then sequence marking is performed on the image features by using the LSTM, and a predicted label (true value) distribution is output to obtain a character label, and finally a series of label distributions obtained from an LSTM layer are converted into a final label sequence by using CTC loss through a sequence conversion network, so that character sequence information can be obtained.
For example, as shown in fig. 7 and fig. 8, which are a result of predicting a lower gastrointestinal tract endoscopic screenshot by a character recognition model and a result of predicting an upper gastrointestinal tract endoscopic screenshot by a character recognition model in an embodiment respectively, a sequence of an illustration result composed of english characters and numbers is character sequence information. Both the lower and upper gastrointestinal endoscopy shots are endoscopic medical images obtained by analysis according to the above embodiment.
And S203, analyzing the character sequence information to determine the medical image type of the endoscope medical image.
The medical image type includes a first medical image and a second medical image, the first medical image may refer to an endoscopic medical image corresponding to a lower digestive tract, and the second medical image may refer to an endoscopic medical image corresponding to an upper digestive tract.
In a specific implementation, the server 104 may call a trained character recognition model to implement character detection on the endoscopic medical image, where the character recognition model may detect and recognize character sequence information included in the endoscopic medical image. The character sequence information contains the model information of the endoscope, and the reason is that targeted image content is obtained when the endoscope medical image is obtained, namely the obtained endoscope medical image necessarily contains characters for identifying the endoscope model, so that the character sequence information is analyzed, and the image classification of the endoscope medical image can be realized.
In one embodiment, this step includes: if the character sequence information contains preset first character information, determining that the medical image type of the endoscope medical image is a first medical image; if the character sequence information contains preset second character information, determining that the medical image type of the endoscope medical image is a second medical image; wherein the first character information includes 'CF-H' and/or 'EC-6', and the second character information includes at least one of 'GIF-H', 'EG-L', and 'EG-7'.
In a specific implementation, the first medical image should be an endoscopic medical image containing first character information such as "CF-H" or "EC-6", and the second medical image should be an endoscopic medical image containing second character information such as "GIF-H", "EG-L" or "EG-7". Therefore, once it is analyzed which of the above character information is included in the character sequence information, the image type of the endoscopic medical image can be inquired and determined.
For example, the server 104 receives a frame of endoscope medical image currently, analyzes the frame of endoscope medical image to obtain the character sequence information included in the endoscope medical image, as shown in fig. 5, since the character sequence information shown in fig. 5 includes "CF-H2901", that is, includes the first character information "CF-H", the endoscope medical image should be determined as the first medical image, that is, the endoscope medical image should be determined as the lower gastrointestinal endoscope medical image.
For another example, the server 104 receives a frame of endoscope medical image currently, analyzes the frame of endoscope medical image to obtain the character sequence information included in the endoscope medical image, as shown in fig. 6, since the character sequence information shown in fig. 6 includes "GIF-HQ 290", that is, includes the second character information "GIF-H", the endoscope medical image should be determined as the second medical image, that is, as the upper gastrointestinal endoscope medical image.
In one embodiment, the endoscopic medical image comprises at least two endoscopic medical images, and the method for applying intelligent classification of medical images further comprises: determining a first image number of endoscopic medical images belonging to the first medical image; and determining a second number of images of the endoscopic medical image that are attributed to the second medical image; if the number of the first images is larger than or equal to a preset number threshold, determining that the medical image type of the endoscopic medical image is the first medical image; and if the number of the second images is larger than or equal to the number threshold, determining the medical image type of the endoscopic medical image as the second medical image.
In a specific implementation, the previous embodiment explains a case of judging an image type when an endoscopic medical image is a frame, and this embodiment will complement a case when multiple frames exist in the endoscopic medical image. That is, when the endoscopic medical image includes a plurality of frames, the judgment of the image type of a single frame is not absolutely valid, and the judgment result of a single frame cannot be a final result, but the judgment results of all images should be comprehensively analyzed.
Specifically, the server 104 may perform detection and analysis on the character sequence information of consecutive multiple frames of endoscopic medical images, and further determine the medical image type of each frame of endoscopic medical image, for example, if there are 7 frames of first medical images and 3 frames of second medical images in consecutive 10 frames of endoscopic medical images, the number of the first images is "7", the number of the second images is "3", and at this time, since the preset number threshold is "7", the medical image type of the consecutive 10 frames of endoscopic medical images is the first medical image. On the contrary, if there are 3 frames of the first medical image and 7 frames of the second medical image in the consecutive 10 frames of frame mirror medical images, the number of the first images is "3", the number of the second images is "7", and at this time, since the preset number threshold is "7", the type of the medical image of the consecutive 10 frames of frame mirror medical images is the second medical image. Of course, the quantity threshold can be set according to actual service requirements, but the larger the value is, the higher the accuracy is.
In one embodiment, the method for applying intelligent classification of medical images further comprises: and if the number of the first images and the number of the second images are both larger than or equal to the number threshold, or if the number of the first images and the number of the second images are both smaller than the number threshold, generating error prompt information, wherein the error prompt information is used for prompting that the endoscope medical images are obtained again for classification.
In a specific implementation, the above embodiment only illustrates a case of successful judgment, that is, the type of medical image of the multi-frame endoscopic medical image can be sequentially judged. The case of non-sequential determination will be supplemented in this embodiment. That is, if both the first image number and the second image number are greater than or equal to the number threshold, or both are less than the number threshold, the abnormality prompt message is generated to re-acquire the endoscopic medical image.
For example, if there are 6 first medical images and 4 second medical images in consecutive 10 frame endoscope medical images, the number of the first images is "6", the number of the second images is "4", and at this time, since the preset number threshold is "7", the medical image type of the consecutive 10 frame endoscope medical images cannot be determined, and the endoscope medical images need to be acquired again for analysis.
And S204, screening a target detection program from a plurality of pre-stored detection programs according to the type of the medical image so as to control the electronic endoscope to execute the target detection program.
The detection program may include an enteroscope detection program, a gastroscope detection program, and other digestive tract detection programs, and may also include detection programs of other parts, which is not specifically limited in the present application.
In a specific implementation, after analyzing the medical image type of the endoscopic medical image, the server 104 determines the endoscope model of the electronic endoscope corresponding to the endoscopic medical image, that is, when determining that the medical image type of the endoscopic medical image is the first image type, the electronic endoscope corresponding to the image is the lower gastrointestinal endoscope; when the medical image type of the endoscopic medical image is judged to be the second image type, the electronic endoscope corresponding to the image is the upper gastrointestinal endoscope. Thus, after determining the model of the currently used electronic endoscope, the server 104 can call up a detection program for the detection portion for the endoscope model.
For example, referring to fig. 9, a flowchart of an embodiment of an intelligent classification application of endoscopic medical images is shown, which illustrates a classification application process of endoscopic medical images.
In one embodiment, the testing procedure comprises an enteroscopy testing procedure and a gastroscopic testing procedure, the steps comprising: if the medical image type is a first medical image, acquiring first residual characters except the first character information in the character sequence information, and screening target enteroscopy detection programs from various enteroscopy detection programs according to the first residual characters to serve as target detection programs; if the medical image type is a second medical image, second residual characters except the second character information in the character sequence information are obtained, and target gastroscope detection programs are screened out from various gastroscope detection programs according to the second residual characters and serve as target detection programs; and controlling the electronic endoscope to execute the target detection program.
Referring to fig. 7, if the character sequence information is "CF-H2901", the first character information is "CF-H", and the first remaining character is "2901"; similarly, referring to fig. 8, if the character sequence information is "GIF-HQ 290", the second character information is "GIF-H", and the second remaining character is "Q290".
In specific implementation, if the similar character sequence corresponds to different detection programs, only the medical image type is identified, and the target detection program cannot be accurately obtained, and further analysis of the character sequence information is needed, that is, the remaining characters in the character sequence information except for the preset first character information or the preset second character information are analyzed, so that the target detection program is locked from the plurality of pre-stored detection programs.
For example, "CF-H2901" corresponds to detection program a, and "CF-H2902" corresponds to detection program B, since there are two detection programs but the two character sequence information both include the first character information "CF-H", server 104 can further analyze the remaining characters, i.e., analyze "2901" and "2902", thereby locking the target detection program.
In the method for intelligently classifying and applying medical images in the above embodiment, the server obtains an endoscope medical image of the electronic endoscope, inputs the endoscope medical image into a trained character recognition model for character recognition, and outputs character sequence information of the endoscope medical image, i.e., the character sequence information can be analyzed to determine a medical image type to which the endoscope medical image belongs, and finally, a target detection program is screened out from a plurality of pre-stored detection programs according to the medical image type to control the electronic endoscope to execute the target detection program. According to the method and the device, the image type to which the medical image belongs is accurately analyzed by combining a deep learning technology, and then the endoscope model and the detection program are accurately selected by referring to the image type, so that the classification accuracy and the analysis stability of the medical image of the endoscope can be improved, the clinical operation process can be simplified, and the detection cost of the artificial intelligence technology in the medical field is saved.
It should be understood that, although the steps in the flowchart of fig. 2 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 2 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
In order to better implement the intelligent classification application method for medical images provided by the embodiment of the present application, on the basis of the intelligent classification application method for medical images provided by the present application, an embodiment of the present application further provides an intelligent classification application apparatus for medical images, as shown in fig. 10, the intelligent classification application apparatus 1000 for medical images includes:
an image acquisition module 1010 for acquiring an endoscopic medical image of the electronic endoscope;
a character recognition module 1020, configured to input the endoscopic medical image into a trained character recognition model for character recognition, and output character sequence information of the endoscopic medical image;
a type determining module 1030, configured to analyze the character sequence information to determine a medical image type to which the endoscopic medical image belongs;
the program executing module 1040 is configured to screen out a target detection program from a plurality of pre-stored detection programs according to the medical image type, so as to control the electronic endoscope to execute the target detection program.
In one embodiment, the type determining module 1030 is further configured to determine that the medical image type of the endoscopic medical image is the first medical image if the character sequence information includes preset first character information; if the character sequence information contains preset second character information, determining that the medical image type of the endoscope medical image is a second medical image; wherein the first character information includes 'CF-H' and/or 'EC-6', and the second character information includes at least one of 'GIF-H', 'EG-L', and 'EG-7'.
In one embodiment, the type determination module 1030 is further configured to determine a first number of images of an endoscopic medical image that is attributed to the first medical image; and determining a second number of images of the endoscopic medical image that are attributed to the second medical image; if the number of the first images is larger than or equal to a preset number threshold, determining that the medical image type of the endoscopic medical image is a first medical image; and if the number of the second images is larger than or equal to the number threshold value, determining that the medical image type of the endoscopic medical image is the second medical image.
In one embodiment, the intelligent classification application device 1000 for medical images further includes an error prompt module, configured to generate an error prompt message if the first number of images and the second number of images are both greater than or equal to a number threshold, or if the first number of images and the second number of images are both less than the number threshold, where the error prompt message is used to prompt to retrieve the endoscopic medical images for classification.
In one embodiment, the detection programs include an enteroscopy detection program and a gastroscope detection program, and the program executing module 1040 is further configured to, if the medical image type is the first medical image, obtain a first remaining character in the character sequence information except the first character information, and screen out a target enteroscopy detection program from the respective enteroscopy detection programs as the target detection program according to the first remaining character; if the medical image type is a second medical image, second residual characters except the second character information in the character sequence information are obtained, and target gastroscope detection programs are screened out from various gastroscope detection programs according to the second residual characters and serve as target detection programs; and controlling the electronic endoscope to execute the target detection program.
In one embodiment, the trained character recognition model comprises a feature extraction network, a sequence labeling network and a sequence conversion network; the character recognition module 1020 is further configured to input the endoscope medical image into the trained character recognition model, and perform feature extraction on the endoscope medical image through a feature extraction network to obtain image features; carrying out sequence marking on the image characteristics through a sequence marking network to obtain a character label; and performing sequence conversion on the character tags through a sequence conversion network to obtain character sequence information.
In one embodiment, the apparatus 1000 for applying intelligent classification of medical images further comprises a model construction module for constructing an initial character recognition model; the initial character recognition model consists of a feature extraction network, a sequence marking network and a sequence conversion network, wherein the feature extraction network comprises ResNet50, and the sequence marking network comprises a long-term and short-term memory network; acquiring a medical sample image set, and dividing the medical sample image set into a training set and a testing set, wherein the medical sample image set comprises a plurality of medical sample images marked with characters; using the training set to perform preliminary training on the initial character recognition model to obtain a preliminarily trained character recognition model; and testing and adjusting the preliminarily trained character recognition model by using the test set to obtain the trained character recognition model.
In one embodiment, the image acquisition module 1010 is further configured to acquire endoscope video of the electronic endoscope in a white light mode; performing frame extraction on an endoscope video to obtain at least two frames of endoscope images; and processing the images of the endoscopes according to a preset interception starting point and a preset screenshot size to obtain medical images of the endoscopes.
In one embodiment, the image obtaining module 1010 is further configured to perform image capturing on each endoscope image according to a preset capturing start point and a preset capturing size, so as to obtain at least two frames of candidate endoscope images; and performing pixel processing on each candidate endoscope image based on a preset binarization method to obtain an endoscope medical image.
In the embodiment, because the application provides the technology of combining the deep learning technology to realize the accurate analysis of the image type of the medical image, and then the endoscope model and the detection program are accurately selected by referring to the image type, the classification accuracy and the analysis stability of the endoscope medical image can be improved, the clinical operation process can be simplified, and the detection cost of the artificial intelligence technology in the medical field is saved.
It should be noted that, for specific limitations of the apparatus for applying intelligent classification to medical images, reference may be made to the above limitations of the method for applying intelligent classification to medical images, and details thereof are not repeated herein. The modules in the intelligent medical image classification application device can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent of a processor in the electronic device, or can be stored in a memory in the electronic device in a software form, so that the processor can call and execute operations corresponding to the modules.
In some embodiments of the present application, the intelligent classification application apparatus 1000 for medical images may be implemented in the form of a computer program, which may be run on a computer device as shown in fig. 11. The memory of the computer device may store various program modules constituting the intelligent classification application 1000 of the medical image, such as an image acquisition module 1010, a character recognition module 1020, a type determination module 1030, and a program execution module 1040 shown in fig. 10; the program modules constitute computer programs to make the processor execute the steps of the intelligent classification application method of medical images of the present application described in the present specification. For example, the computer device shown in fig. 11 may execute step S201 through the image acquisition module 1010 in the intelligent classification application apparatus 1000 of medical images as shown in fig. 10. The computer device may perform step S202 through the character recognition module 1020. The computer device may perform step S203 through the type determination module 1030. The computer device may perform step S204 through the program execution module 1040. Wherein the computer device comprises a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external computer device through a network connection. The computer program is executed by a processor to implement a method of intelligent classification application of medical images.
Those skilled in the art will appreciate that the architecture shown in fig. 11 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In some embodiments of the present application, there is provided a computer device comprising one or more processors; a memory; and one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processor for the steps of the method for applying intelligent classification of medical images. Here, the steps of the method for applying intelligent classification of medical images may be the steps of the method for applying intelligent classification of medical images according to the above embodiments.
In some embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, which is loaded by a processor, so that the processor performs the steps of the above-mentioned method for applying intelligent classification to medical images. The steps of the method for applying intelligent classification of medical images herein may be the steps of the method for applying intelligent classification of medical images of the various embodiments described above.
It will be understood by those of ordinary skill in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when executed. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).
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 method, the apparatus, the device and the storage medium for applying intelligent classification to medical images provided by the embodiments of the present application are described in detail above, and a specific example is applied in the present application to explain the principle and the implementation of the present invention, and the description of the above embodiments is only used to help understanding the method and the core idea of the present invention; meanwhile, for those skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (7)
1. An intelligent classification application method for medical images is characterized by comprising the following steps:
acquiring an endoscope medical image of an electronic endoscope; the endoscope medical image is an image contained in a parameter area of interest in an endoscope image acquired by the electronic endoscope; the size of the parameter region of interest is (0.156)w 0, 0.16h 0)，(w 0,h 0) Is the size of the endoscopic image;
inputting the endoscope medical image into a trained character recognition model for character recognition, and outputting character sequence information of the endoscope medical image; wherein the trained character recognition model is composed of a deep learning based CRNN network and a ResNet50 network;
if the character sequence information contains preset first character information, determining that the medical image type of the endoscope medical image is a first medical image; if the character sequence information contains preset second character information, determining that the medical image type of the endoscope medical image is a second medical image; wherein the first character information includes 'CF-H' and/or 'EC-6', and the second character information includes at least one of 'GIF-H', 'EG-L', and 'EG-7';
determining a first number of images of an endoscopic medical image that is attributed to the first medical image;
determining a second number of images of an endoscopic medical image that is attributed to the second medical image;
if the number of the first images is larger than or equal to a preset number threshold, determining that the medical image type of the endoscopic medical image is the first medical image;
if the number of the second images is larger than or equal to the number threshold, determining that the medical image type of the endoscopic medical image is the second medical image;
if the number of the first images and the number of the second images are simultaneously larger than or equal to the number threshold value or simultaneously smaller than the number threshold value, generating abnormal prompt information for prompting to obtain the endoscope medical image again for medical image type judgment;
if the medical image type is the first medical image, acquiring first residual characters except the first character information in the character sequence information, and screening a target enteroscopy detection program from a plurality of prestored enteroscopy detection programs according to the first residual characters to serve as the target detection program;
if the medical image type is the second medical image, acquiring second residual characters in the character sequence information except the second character information, and screening out a target gastroscope detection program from a plurality of prestored gastroscope detection programs according to the second residual characters to serve as the target detection program;
and controlling the electronic endoscope to execute the target detection program.
2. The method of claim 1, wherein the trained character recognition model comprises a feature extraction network, a sequence labeling network, and a sequence conversion network; wherein the content of the first and second substances,
the inputting the endoscope medical image into a trained character recognition model for character recognition and outputting the character sequence information of the endoscope medical image comprises the following steps:
inputting the endoscope medical image into a trained character recognition model, and performing feature extraction on the endoscope medical image through the feature extraction network to obtain image features;
carrying out sequence marking on the image characteristics through the sequence marking network to obtain a character label;
and performing sequence conversion on the character tags through the sequence conversion network to obtain the character sequence information.
3. The method of claim 1, wherein said acquiring endoscopic medical images of an electronic endoscope comprises:
acquiring an endoscope video of the electronic endoscope in a white light mode;
performing frame extraction on the endoscope video to obtain at least two frames of endoscope images;
and processing the images of the endoscopes according to a preset interception starting point and a preset screenshot size to obtain the medical images of the endoscopes.
4. The method of claim 3, wherein said image processing each of said endoscopic images according to a predetermined cut-out starting point and a predetermined cut-out size to obtain said endoscopic medical image comprises:
according to a preset intercepting starting point and a preset screenshot size, image interception is carried out on each endoscope image to obtain at least two frames of candidate endoscope images;
and performing pixel processing on each candidate endoscope image based on a preset binarization method to obtain the endoscope medical image.
5. An apparatus for intelligently classifying and applying medical images, comprising:
the image acquisition module is used for acquiring an endoscope medical image of the electronic endoscope; the endoscope medical image is an image contained in a parameter area of interest in an endoscope image acquired by the electronic endoscope; the size of the parameter region of interest is (0.156)w 0, 0.16h 0)，(w 0,h 0) Is the size of the endoscopic image;
the character recognition module is used for inputting the endoscope medical image into a trained character recognition model for character recognition and outputting character sequence information of the endoscope medical image; wherein the trained character recognition model is composed of a deep learning based CRNN network and a ResNet50 network;
the type determining module is used for determining that the medical image type of the endoscope medical image is a first medical image if the character sequence information contains preset first character information; if the character sequence information contains preset second character information, determining that the medical image type of the endoscope medical image is a second medical image; wherein the first character information includes 'CF-H' and/or 'EC-6', and the second character information includes at least one of 'GIF-H', 'EG-L', and 'EG-7'; determining a first number of images of an endoscopic medical image that is attributed to the first medical image; and determining a second number of images of endoscopic medical images that are attributed to the second medical image; if the number of the first images is larger than or equal to a preset number threshold, determining that the medical image type of the endoscopic medical image is the first medical image; if the number of the second images is larger than or equal to the number threshold, determining that the medical image type of the endoscopic medical image is the second medical image; if the number of the first images and the number of the second images are simultaneously larger than or equal to the number threshold value or simultaneously smaller than the number threshold value, generating abnormal prompt information for prompting to obtain the endoscope medical image again for medical image type judgment;
a program execution module, configured to, if the medical image type is the first medical image, obtain first remaining characters in the character sequence information, except for the first character information, so as to screen a target enteroscopy detection program from a plurality of prestored enteroscopy detection programs according to the first remaining characters, where the target enteroscopy detection program is used as the target detection program; if the medical image type is the second medical image, acquiring second residual characters in the character sequence information except the second character information, and screening a target gastroscope detection program from a plurality of prestored gastroscope detection programs according to the second residual characters to serve as the target detection program; and controlling the electronic endoscope to execute the target detection program.
6. A medical image processing apparatus, characterized by comprising:
one or more processors;
a memory; and one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the intelligent classification application method for medical images of any of claims 1 to 4.
7. A computer-readable storage medium, having stored thereon a computer program which is loaded by a processor for performing the steps of the method for applying intelligent classification of medical images according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210255096.0A CN114332844B (en) | 2022-03-16 | 2022-03-16 | Intelligent classification application method, device, equipment and storage medium of medical image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210255096.0A CN114332844B (en) | 2022-03-16 | 2022-03-16 | Intelligent classification application method, device, equipment and storage medium of medical image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114332844A CN114332844A (en) | 2022-04-12 |
| CN114332844B true CN114332844B (en) | 2022-07-01 |
Family
ID=81034088
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210255096.0A Active CN114332844B (en) | 2022-03-16 | 2022-03-16 | Intelligent classification application method, device, equipment and storage medium of medical image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114332844B (en) |
Family Cites Families (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5675227B2 (en) * | 2010-08-31 | 2015-02-25 | 富士フイルム株式会社 | Endoscopic image processing apparatus, operation method, and program |
| US10169863B2 (en) * | 2015-06-12 | 2019-01-01 | International Business Machines Corporation | Methods and systems for automatically determining a clinical image or portion thereof for display to a diagnosing physician |
| CN106713700A (en) * | 2016-12-08 | 2017-05-24 | 宇龙计算机通信科技(深圳)有限公司 | Picture processing method and apparatus, as well as terminal |
| CN110321755A (en) * | 2018-03-28 | 2019-10-11 | 中移(苏州)软件技术有限公司 | A kind of recognition methods and device |
| WO2020036121A1 (en) * | 2018-08-17 | 2020-02-20 | 富士フイルム株式会社 | Endoscope system |
| CN110495847B (en) * | 2019-08-23 | 2021-10-08 | 重庆天如生物科技有限公司 | Advanced learning-based auxiliary diagnosis system and examination device for early cancer of digestive tract |
| CN110613417A (en) * | 2019-09-24 | 2019-12-27 | 浙江同花顺智能科技有限公司 | Method, equipment and storage medium for outputting upper digestion endoscope operation information |
| CN111666868A (en) * | 2020-06-03 | 2020-09-15 | 阳光保险集团股份有限公司 | Insurance policy identification method and device and computer equipment |
| CN111881943A (en) * | 2020-07-08 | 2020-11-03 | 泰康保险集团股份有限公司 | Method, device, equipment and computer readable medium for image classification |
| CN111931835A (en) * | 2020-07-31 | 2020-11-13 | 中国工商银行股份有限公司 | Image identification method, device and system |
| CN112508039B (en) * | 2020-12-08 | 2024-04-02 | ***股份有限公司 | Image detection method and device |
| CN112418209B (en) * | 2020-12-15 | 2022-09-13 | 润联软件***(深圳)有限公司 | Character recognition method and device, computer equipment and storage medium |
| CN112906699A (en) * | 2020-12-23 | 2021-06-04 | 深圳市信义科技有限公司 | Method for detecting and identifying enlarged number of license plate |
| CN112927776A (en) * | 2021-02-03 | 2021-06-08 | 昆山慧医优策医疗科技有限公司 | Artificial intelligence automatic interpretation system for medical inspection report |
| CN113642562A (en) * | 2021-08-30 | 2021-11-12 | 平安医疗健康管理股份有限公司 | Data interpretation method, device and equipment based on image recognition and storage medium |
| CN113762257A (en) * | 2021-09-30 | 2021-12-07 | 时趣互动（北京）科技有限公司 | Identification method and device for marks in makeup brand images |
| CN113920309B (en) * | 2021-12-14 | 2022-03-01 | 武汉楚精灵医疗科技有限公司 | Image detection method, image detection device, medical image processing equipment and storage medium |
-
2022
- 2022-03-16 CN CN202210255096.0A patent/CN114332844B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN114332844A (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10810735B2 (en) | Method and apparatus for analyzing medical image | |
| JP7058373B2 (en) | Lesion detection and positioning methods, devices, devices, and storage media for medical images | |
| CN110866908B (en) | Image processing method, image processing apparatus, server, and storage medium | |
| CN110288597B (en) | Attention mechanism-based wireless capsule endoscope video saliency detection method | |
| JP2021513435A (en) | Systems and methods for diagnosing gastrointestinal tumors | |
| CN108229288B (en) | Neural network training and clothes color detection method and device, storage medium and electronic equipment | |
| CN113920309B (en) | Image detection method, image detection device, medical image processing equipment and storage medium | |
| EP3998579A1 (en) | Medical image processing method, apparatus and device, medium and endoscope | |
| KR101955919B1 (en) | Method and program for providing tht region-of-interest in image by deep-learing algorithm | |
| CN113129293B (en) | Medical image classification method, medical image classification device, computer equipment and storage medium | |
| JP7420916B2 (en) | Real-time deployment of machine learning systems | |
| CN111814768B (en) | Image recognition method, device, medium and equipment based on AI composite model | |
| CN112419295A (en) | Medical image processing method, apparatus, computer device and storage medium | |
| Zhang et al. | Dual encoder fusion u-net (defu-net) for cross-manufacturer chest x-ray segmentation | |
| CN113255516A (en) | Living body detection method and device and electronic equipment | |
| CN113935943A (en) | Method, device, computer equipment and storage medium for intracranial aneurysm identification detection | |
| CN114445406A (en) | Enteroscopy image analysis method and device and medical image processing equipment | |
| CN111127432B (en) | Medical image detection method, device, equipment and storage medium | |
| CN114332844B (en) | Intelligent classification application method, device, equipment and storage medium of medical image | |
| CN111047582A (en) | Crohn's disease auxiliary diagnosis system under enteroscope based on degree of depth learning | |
| EP4036867A2 (en) | System and method for classifying dermatological images using machine learning | |
| KR20220078495A (en) | Method, apparatus and program to read lesion of small intestine based on capsule endoscopy image | |
| CN112102235A (en) | Human body part recognition method, computer device, and storage medium | |
| CN110390234B (en) | Image processing apparatus and method, and storage medium | |
| KR20210054140A (en) | Medical image diagnosis assistance apparatus and method using a plurality of medical image diagnosis algorithm for endoscope images |
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 |