CN111784686A - Dynamic intelligent detection method, system and readable storage medium for endoscope bleeding area - Google Patents
Dynamic intelligent detection method, system and readable storage medium for endoscope bleeding area Download PDFInfo
- Publication number
- CN111784686A CN111784686A CN202010700705.XA CN202010700705A CN111784686A CN 111784686 A CN111784686 A CN 111784686A CN 202010700705 A CN202010700705 A CN 202010700705A CN 111784686 A CN111784686 A CN 111784686A
- Authority
- CN
- China
- Prior art keywords
- image
- bleeding
- endoscope
- rgb
- area
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30028—Colon; Small intestine
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Endoscopes (AREA)
Abstract
The invention discloses a dynamic intelligent detection method, a system and a readable storage medium for endoscope bleeding areas, wherein the method comprises the following steps: acquiring an endoscope image and converting the image into an RGB image by utilizing a linear interpolation algorithm; preprocessing the RGB image, wherein the preprocessing comprises filtering, denoising and interference area removing; constructing a binary template to perform image segmentation on the preprocessed RGB image; setting a standard bleeding area image pixel value, calculating an average pixel vector of each RGB segmented image, and calculating the similarity of the average pixel vectors of each RGB segmented image of the standard bleeding area image pixel value; and if the similarity is greater than the preset value, inputting the segmented image into a preset bleeding area detection model classifier, and outputting a bleeding detection result. The invention has fast speed and high accuracy for detecting the bleeding area.
Description
Technical Field
The invention relates to the field of medical image processing, in particular to a dynamic intelligent detection method, a dynamic intelligent detection system and a readable storage medium for endoscope bleeding areas.
Background
With the development of economic society, the work pace is accelerated, the environment is polluted, gastrointestinal diseases become common diseases affecting human health, such as gastric cancer, intestinal cancer and the like, and are mostly common diseases and frequently encountered diseases, which have great threat to human health, and capsule endoscope systems are widely applied as effective means for judging gastrointestinal diseases due to the advantages of high safety, high reliability and the like. Many of the gastrointestinal disorders are accompanied by bleeding. By combining the computer technology and the image recognition technology, the bleeding images in the image sequence are accurately detected, and the capsule endoscope system can assist doctors in diagnosis in gastrointestinal tract examination, so that the diagnosis efficiency is improved, and the workload of the doctors is reduced.
The deep learning is an algorithm based on the characterization learning of data, and is very suitable for finding complex structures in high-dimensional data such as images. In the process of training the network, firstly, respectively performing rotation, brightness adjustment, Gaussian blur and Poisson noise on a bleeding image and a non-bleeding image, and then forming a new data set by the converted images and an original image together; secondly, copying all bleeding images in the process of network training to enable the number of the bleeding images to be equal to that of the non-bleeding images, and thus obtaining an amplification data set; then training three deep convolution neural networks of VGGNet, GoogleNet and AlexNet to obtain three nonlinear mapping relations; and finally, performing endoscope image intestinal bleeding detection according to the obtained three deep convolutional neural networks to obtain three detection results, and then obtaining a final detection result according to a majority voting rule.
Currently, the bleeding area detection based on the endoscope has the defects of more processing data, lower speed and unsatisfactory detection effect in a detection algorithm,
therefore, it is highly desirable to develop a dynamic intelligent detection method for endoscope bleeding areas.
Disclosure of Invention
In order to solve at least one technical problem, the invention provides a dynamic intelligent detection method, a system and a readable storage medium for an endoscope bleeding area.
In order to solve the technical problem, a first aspect of the present invention discloses a dynamic intelligent detection method for a bleeding area of an endoscope, comprising:
acquiring an endoscope image and converting the image into an RGB image by utilizing a linear interpolation algorithm;
preprocessing the RGB image, wherein the preprocessing comprises filtering, denoising and interference area removing;
constructing a binary template to perform image segmentation on the preprocessed RGB image;
setting a standard bleeding area image pixel value, calculating an average pixel vector of each RGB segmented image, and calculating the similarity of the average pixel vectors of each RGB segmented image of the standard bleeding area image pixel value;
and if the similarity is greater than the preset value, inputting the segmented image into a preset bleeding area detection model, and outputting a bleeding detection result.
In the scheme, the linear interpolation algorithm is used for converting the image into the RGB image, and the linear interpolation algorithm based on the adjacent domain is adopted.
In this scheme, the filtering process is an average filtering process.
In this scheme, the removing the interference region includes: cavity area, over-exposed area in the endoscopic image.
In the scheme, the image segmentation process of the preprocessed RGB image by constructing the binary template comprises the following steps:
calculating the saturation component of the preprocessed RGB image;
taking 15% of the maximum saturation as a threshold value to generate a binary template;
and multiplying the preprocessed RGB image by the binary template to obtain a segmentation image.
In the scheme, a standard bleeding area image pixel value A is set, an average pixel vector X of each RGB (red, green and blue) segmented image is calculated, the similarity of the average pixel vectors of each RGB segmented image of the standard bleeding area image pixel value is calculated, and a similarity value expression is used
In the scheme, the bleeding area detection model is a preset bleeding area detection model based on a neural network, and the output result is a bleeding image or a non-bleeding image.
The invention provides a dynamic intelligent detection system for endoscope bleeding areas, which comprises a memory and a processor, wherein the memory comprises a dynamic intelligent detection method program for endoscope bleeding areas, and the dynamic intelligent detection method program for endoscope bleeding areas realizes the following steps when being executed by the processor:
acquiring an endoscope image and converting the image into an RGB image by utilizing a linear interpolation algorithm;
preprocessing the RGB image, wherein the preprocessing comprises filtering, denoising and interference area removing;
constructing a binary template to perform image segmentation on the preprocessed RGB image;
setting a standard bleeding area image pixel value, calculating an average pixel vector of each RGB segmented image, and calculating the similarity of the average pixel vectors of each RGB segmented image of the standard bleeding area image pixel value;
and if the similarity is greater than the preset value, inputting the segmented image into a preset bleeding area detection model, and outputting a bleeding detection result.
In the scheme, the linear interpolation algorithm is used for converting the image into the RGB image, and the linear interpolation algorithm based on the adjacent domain is adopted.
The third aspect of the present invention provides a computer readable storage medium, which includes a program of a method for dynamically and intelligently detecting an endoscope bleeding area, and when the program of the method for dynamically and intelligently detecting an endoscope bleeding area is executed by a processor, the steps of the method for dynamically and intelligently detecting an endoscope bleeding area are implemented.
The invention discloses a dynamic intelligent detection method, a system and a readable storage medium for endoscope bleeding areas.
Drawings
Fig. 1 shows a flow chart of a dynamic intelligent detection method for endoscope bleeding areas according to the invention.
Fig. 2 shows a flow chart of image segmentation of an RGB image in the present invention.
Fig. 3 shows a block diagram of a dynamic intelligent detection system for endoscope bleeding areas.
Detailed description of the invention
In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings. It should be noted that the embodiments and features of the embodiments of the present application may be combined with each other without conflict.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those specifically described herein, and therefore the scope of the present invention is not limited by the specific embodiments disclosed below.
The dynamic intelligent detection method for the endoscope bleeding area is mainly suitable for medical endoscope systems, such as a hard tube endoscope, a fiber endoscope and an electronic endoscope, and is used for preprocessing endoscope images, further segmenting the images, calculating the similarity between the average pixel vector of the segmented images and the pixel value of the standard bleeding area image, and further detecting the bleeding area through a preset bleeding area detection model. Of course, the present invention is not limited to the type of endoscope used in the room, and any technical solution adopting the present invention will fall into the protection scope of the present invention.
Fig. 1 shows a flow chart of a dynamic intelligent detection method for endoscope bleeding areas according to the invention.
As shown in fig. 1, a first aspect of the present invention discloses a dynamic intelligent detection method for endoscope bleeding areas, including:
s102, acquiring an endoscope image and converting the image into an RGB image by using a linear interpolation algorithm;
s104, preprocessing the RGB image, wherein the preprocessing comprises filtering, denoising and interference area removing;
s106, constructing a binary template to carry out image segmentation on the preprocessed RGB image;
s108, setting pixel values of images in the standard bleeding area, calculating average pixel vectors of RGB segmented images, and calculating the similarity of the average pixel vectors of the RGB segmented images in the pixel values of the images in the standard bleeding area;
and S110, if the similarity is greater than the preset value, inputting the segmented image into a preset bleeding area detection model, and outputting a bleeding detection result.
In a specific embodiment, the endoscopic images may be obtained from a variety of endoscopes, such as hard-tube endoscopes, fiberscopes, electronic endoscopes, which: the endoscope consists of an endoscope body and a cold light source, wherein two optical fiber bundles are arranged in the endoscope body, one optical fiber bundle is called as a light beam, the other optical fiber bundle is called as an image bundle, one end of the optical fiber bundle is aligned to an ocular, the other end of the optical fiber bundle is aligned to the surface of an observed object through an objective lens, and a doctor can visually see the surface condition of the visceral organs through the ocular, so that the condition of the visceral organs can be diagnosed conveniently and accurately.
The electronic endoscope is characterized in that: the electronic endoscope is not used for image transmission any more, but is replaced by a photosensitive integrated circuit camera system, and not only has good image quality, strong brightness and large image, but also can detect finer pathological changes, and the electronic endoscope has thinner outer diameter, clearer and more visual image and convenient operation.
In the scheme, the linear interpolation algorithm is used for converting the image into the RGB image, and the linear interpolation algorithm based on the adjacent domain is adopted.
It should be noted that the image obtained by the endoscope needs to be further converted into a standard RGB color image, for example, the image obtained by the endoscope is a bayer color array, and the missing color component needs to be inserted into each pixel, and a common difference method includes: the method comprises a neighborhood method, a linear interpolation method, a bilinear interpolation method, a cubic spline interpolation method and the like.
In this scheme, the filtering process is an average filtering process.
After the obtained endoscopic image is formatted, a great deal of noise exists in the image, and the noise comprises: white noise, Gaussian noise, salt and pepper noise and the like, so that a new pixel is generated in the center coordinate of the neighborhood by filtering and denoising, the value of the pixel can be obtained by a predefined budget, and the center of the filter traverses each pixel to generate a denoised image.
In this scheme, the removing the interference region includes: cavity area, over-exposed area in the endoscopic image.
It should be noted that, in the present invention, the interference region further includes a peripheral region of the image detection region of interest, in a specific embodiment, the center of the image may be used as a center of a circle, the set pixel length is used as a diameter to make a circle, and the portion outside the circumference is used as a peripheral region, a cavity region in the endoscopic image is generally displayed as black in the endoscopic image, and does not contain any useful information, the overexposure region is a light reflection region with very high brightness, and cannot be used for respectively providing effective information, it should be noted that, by removing the interference region, the influence of the interference region on the detection can be effectively reduced, and at the same time, the image data and the calculation amount can be reduced.
It should be noted that, in the endoscopic image to be detected, a normal tissue is a background region, and a bleeding region and the background region are clearly distinguished in saturation.
Fig. 2 shows a flow chart of image segmentation for RGB images.
In the scheme, the image segmentation process of the preprocessed RGB image by constructing the binary template comprises the following steps:
s202, calculating the saturation component of the preprocessed RGB image;
s204, taking 15% of the maximum saturation as a threshold value to generate a binary template;
s206, multiplying the preprocessed RGB image by the binary template to obtain a segmentation image.
In the present invention, pixels larger than the threshold value in the RGB image are set to 1, that is, displayed as white, and pixels smaller than the threshold value are set to 0, and displayed as black.
In the scheme, a standard bleeding area image pixel value A is set, an average pixel vector X of each RGB (red, green and blue) segmented image is calculated, the similarity of the average pixel vector of each RGB segmented image of the standard bleeding area image pixel value is calculated, and a similarity value expression is obtained
In the scheme, the bleeding area detection model is a preset bleeding area detection model based on a neural network, and the output result is a bleeding image or a non-bleeding image.
It should be noted that, in a specific embodiment, the bleeding area detection model based on the neural network includes: the system comprises a feature extraction module, a feature fusion module, a preliminary detection module and a classification output module, wherein the feature extraction module is used for respectively extracting features of input images, the feature fusion module is used for fusing the extracted features to construct fusion feature vectors, the preliminary detection module detects bleeding areas of the images and outputs the bleeding areas, and the classification output module classifies and outputs preliminary detection results. The method comprises the steps of pre-training a bleeding area detection model based on a neural network, firstly obtaining endoscope images to construct an endoscope image data set, amplifying data of the endoscope image data set, and dividing the amplified endoscope data set into a training set and a test set according to a preset proportion; and training the model by using the training set and the test set respectively.
As shown in fig. 3, a block diagram of a system for dynamic intelligent detection of an endoscope bleeding area, a second aspect of the present invention provides a system for dynamic intelligent detection of an endoscope bleeding area, including a memory 31 and a processor 32, where the memory includes a program for a method for dynamic intelligent detection of an endoscope bleeding area, and when executed by the processor, the program for a method for dynamic intelligent detection of an endoscope bleeding area implements the following steps:
acquiring an endoscope image and converting the image into an RGB image by utilizing a linear interpolation algorithm;
preprocessing the RGB image, wherein the preprocessing comprises filtering, denoising and interference area removing;
constructing a binary template to perform image segmentation on the preprocessed RGB image;
setting a pixel value of an image in a standard bleeding area, calculating an average pixel vector of each RGB segmented image, and calculating the similarity of the average pixel vectors of each segmented image of the RGB pixel value in the standard bleeding area;
and if the similarity is greater than the preset value, inputting the segmented image into a preset bleeding area detection model, and outputting a bleeding detection result.
In a specific embodiment, the endoscopic images may be obtained from a variety of endoscopes, such as hard-tube endoscopes, fiberscopes, electronic endoscopes, which: the endoscope consists of an endoscope body and a cold light source, wherein two optical fiber bundles are arranged in the endoscope body, one optical fiber bundle is called as a light beam, the other optical fiber bundle is called as an image bundle, one end of the optical fiber bundle is aligned to an ocular, the other end of the optical fiber bundle is aligned to the surface of an observed object through an objective lens, and a doctor can visually see the surface condition of the visceral organs through the ocular, so that the condition of the visceral organs can be diagnosed conveniently and accurately.
The electronic endoscope is characterized in that: the electronic endoscope is not used for image transmission any more, but is replaced by a photosensitive integrated circuit camera system, and not only has good image quality, strong brightness and large image, but also can detect finer pathological changes, and the electronic endoscope has thinner outer diameter, clearer and more visual image and convenient operation.
In the scheme, the linear interpolation algorithm is used for converting the image into the RGB image, and the linear interpolation algorithm based on the adjacent domain is adopted.
It should be noted that the image obtained by the endoscope needs to be further converted into a standard RGB color image, for example, the image obtained by the endoscope is a bayer color array, and the missing color component needs to be inserted into each pixel, and a common difference method includes: the method comprises a neighborhood method, a linear interpolation method, a bilinear interpolation method, a cubic spline interpolation method and the like.
In this scheme, the filtering process is an average filtering process.
After the obtained endoscopic image is formatted, a great deal of noise exists in the image, and the noise comprises: white noise, Gaussian noise, salt and pepper noise and the like, so that a new pixel is generated in the center coordinate of the neighborhood by filtering and denoising, the value of the pixel can be obtained by a predefined budget, and the center of the filter traverses each pixel to generate a denoised image.
In this scheme, the removing the interference region includes: cavity area, over-exposed area in the endoscopic image.
It should be noted that, in the present invention, the interference region further includes a peripheral region of the image detection region of interest, in a specific embodiment, the center of the image may be used as a center of a circle, the set pixel length is used as a diameter to make a circle, and the portion outside the circumference is used as a peripheral region, a cavity region in the endoscopic image is generally displayed as black in the endoscopic image, and does not contain any useful information, the overexposure region is a light reflection region with very high brightness, and cannot be used for respectively providing effective information, it should be noted that, by removing the interference region, the influence of the interference region on the detection can be effectively reduced, and at the same time, the image data and the calculation amount can be reduced.
It should be noted that, in the endoscopic image to be detected, the normal tissue is the background, and the bleeding area and the background area are clearly distinguished in saturation, and the invention uses the saturation as a template to perform image segmentation.
In the scheme, the image segmentation process of the preprocessed RGB image by constructing the binary template comprises the following steps:
calculating the saturation component of the preprocessed RGB image;
taking 15% of the maximum saturation as a threshold value to generate a binary template;
and multiplying the preprocessed RGB image by the binary template to obtain a segmentation image.
In the present invention, pixels larger than the threshold value in the RGB image are set to 1, that is, displayed as white, and pixels smaller than the threshold value are set to 0, and displayed as black.
In the scheme, a standard bleeding area image pixel value A is set, an average pixel vector X of each RGB segmentation image is calculated, and a standard bleeding area is calculatedSimilarity of average pixel vectors of respective divided images of image pixel values RGB, similarity value expression
In the scheme, the bleeding area detection model is a preset bleeding area detection model based on a neural network, and the output result is a bleeding image or a non-bleeding image.
It should be noted that, in a specific embodiment, the bleeding area detection model based on the neural network includes: the system comprises a feature extraction module, a feature fusion module, a preliminary detection module and a classification output module, wherein the feature extraction module is used for respectively extracting features of input images, the feature fusion module is used for fusing the extracted features to construct fusion feature vectors, the preliminary detection module detects bleeding areas of the images and outputs the bleeding areas, and the classification output module classifies and outputs preliminary detection results. The method comprises the steps of pre-training a bleeding area detection model based on a neural network, firstly obtaining endoscope images to construct an endoscope image data set, amplifying data of the endoscope image data set, and dividing the amplified endoscope data set into a training set and a test set according to a preset proportion; and training the model by using the training set and the test set respectively.
The third aspect of the present invention provides a computer readable storage medium, which includes a program of a method for dynamically and intelligently detecting an endoscope bleeding area, and when the program of the method for dynamically and intelligently detecting an endoscope bleeding area is executed by a processor, the steps of the method for dynamically and intelligently detecting an endoscope bleeding area are implemented.
The dynamic intelligent detection method for the endoscope bleeding area comprises the following steps:
acquiring an endoscope image and converting the image into an RGB image by utilizing a linear interpolation algorithm;
preprocessing the RGB image, wherein the preprocessing comprises filtering, denoising and interference area removing;
constructing a binary template to perform image segmentation on the preprocessed RGB image;
setting a standard bleeding area image pixel value, calculating an average pixel vector of each RGB segmented image, and calculating the similarity of the average pixel vectors of each RGB segmented image of the standard bleeding area image pixel value;
and if the similarity is greater than the preset value, inputting the segmented image into a preset bleeding area detection model, and outputting a bleeding detection result.
In a specific embodiment, the endoscopic images may be obtained from a variety of endoscopes, such as hard-tube endoscopes, fiberscopes, electronic endoscopes, which: the endoscope consists of an endoscope body and a cold light source, wherein two optical fiber bundles are arranged in the endoscope body, one optical fiber bundle is called as a light beam, the other optical fiber bundle is called as an image bundle, one end of the optical fiber bundle is aligned to an ocular, the other end of the optical fiber bundle is aligned to the surface of an observed object through an objective lens, and a doctor can visually see the surface condition of the visceral organs through the ocular, so that the condition of the visceral organs can be diagnosed conveniently and accurately.
The electronic endoscope is characterized in that: the electronic endoscope is not used for image transmission any more, but is replaced by a photosensitive integrated circuit camera system, and not only has good image quality, strong brightness and large image, but also can detect finer pathological changes, and the electronic endoscope has thinner outer diameter, clearer and more visual image and convenient operation.
In the scheme, the linear interpolation algorithm is used for converting the image into the RGB image, and the linear interpolation algorithm based on the adjacent domain is adopted.
It should be noted that the image obtained by the endoscope needs to be further converted into a standard RGB color image, for example, the image obtained by the endoscope is a bayer color array, and the missing color component needs to be inserted into each pixel, and a common difference method includes: the method comprises a neighborhood method, a linear interpolation method, a bilinear interpolation method, a cubic spline interpolation method and the like.
In this scheme, the filtering process is an average filtering process.
After the obtained endoscopic image is formatted, a great deal of noise exists in the image, and the noise comprises: white noise, Gaussian noise, salt and pepper noise and the like, so that a new pixel is generated in the center coordinate of the neighborhood by filtering and denoising, the value of the pixel can be obtained by a predefined budget, and the center of the filter traverses each pixel to generate a denoised image.
In this scheme, the removing the interference region includes: cavity area, over-exposed area in the endoscopic image.
It should be noted that, in the present invention, the interference region further includes a peripheral region of the image detection region of interest, in a specific embodiment, the center of the image may be used as a center of a circle, the set pixel length is used as a diameter to make a circle, and the portion outside the circumference is used as a peripheral region, a cavity region in the endoscopic image is generally displayed as black in the endoscopic image, and does not contain any useful information, the overexposure region is a light reflection region with very high brightness, and cannot be used for respectively providing effective information, it should be noted that, by removing the interference region, the influence of the interference region on the detection can be effectively reduced, and at the same time, the image data and the calculation amount can be reduced.
It should be noted that, in the endoscopic image to be detected, the normal tissue is the background, and the bleeding area and the background area are clearly distinguished in saturation, and the invention uses the saturation as a template to perform image segmentation.
In the scheme, the image segmentation process of the preprocessed RGB image by constructing the binary template comprises the following steps:
calculating the saturation component of the preprocessed RGB image;
taking 15% of the maximum saturation as a threshold value to generate a binary template;
and multiplying the preprocessed RGB image by the binary template to obtain a segmentation image.
In the present invention, pixels larger than the threshold value in the RGB image are set to 1, that is, displayed as white, and pixels smaller than the threshold value are set to 0, and displayed as black.
In the scheme, a standard bleeding area image pixel value A is set, the average pixel direction X of each RGB (red, green and blue) segmented image is calculated, the similarity of the average pixel vector of each RGB segmented image of the standard bleeding area image pixel value is calculated, and the similarity value expression is used for expressing
In the scheme, the bleeding area detection model is a preset bleeding area detection model based on a neural network, and the output result is a bleeding image or a non-bleeding image.
It should be noted that, in a specific embodiment, the bleeding area detection model based on the neural network includes: the system comprises a feature extraction module, a feature fusion module, a preliminary detection module and a classification output module, wherein the feature extraction module is used for respectively extracting features of input images, the feature fusion module is used for fusing the extracted features to construct fusion feature vectors, the preliminary detection module detects bleeding areas of the images and outputs the bleeding areas, and the classification output module classifies and outputs preliminary detection results. The method comprises the steps of pre-training a bleeding area detection model based on a neural network, firstly obtaining endoscope images to construct an endoscope image data set, amplifying data of the endoscope image data set, and dividing the amplified endoscope data set into a training set and a test set according to a preset proportion; and training the model by using the training set and the test set respectively.
In order to better explain the technical solution of the present invention, the following will describe in detail the specific steps of the indoor teaching action analysis and correction method through several embodiments.
The invention discloses a dynamic intelligent detection method, a system and a readable storage medium for endoscope bleeding areas.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described device embodiments are merely illustrative, for example, the division of the unit is only a logical functional division, and there may be other division ways in actual implementation, such as: multiple units or components may be combined, or may be integrated into another system, or some features may be omitted, or not implemented. In addition, the coupling, direct coupling or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or units may be electrical, mechanical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units; can be located in one place or distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, all the functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately regarded as one unit, or two or more units may be integrated into one unit; the integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.
Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: a mobile storage device, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Alternatively, the integrated unit of the present invention may be stored in a computer-readable storage medium if it is implemented in the form of a software functional module and sold or used as a separate product. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or a part contributing to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, a ROM, a RAM, a magnetic or optical disk, or various other media that can store program code.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (10)
1. A dynamic intelligent detection method for endoscope bleeding areas is characterized by comprising the following steps:
acquiring an endoscope image and converting the image into an RGB image by utilizing a linear interpolation algorithm;
preprocessing the RGB image, wherein the preprocessing comprises filtering, denoising and interference area removing;
constructing a binary template to perform image segmentation on the preprocessed RGB image;
setting a standard bleeding area image pixel value, calculating an average pixel vector of each RGB segmented image, and calculating the similarity of the average pixel vectors of each RGB segmented image of the standard bleeding area image pixel value;
and if the similarity is greater than the preset value, inputting the segmented image into a preset bleeding area detection model, and outputting a bleeding detection result.
2. The method as claimed in claim 1, wherein the converting of the image into the RGB image by the linear interpolation algorithm is performed by a clinical-domain-based linear interpolation algorithm.
3. The method according to claim 1, wherein the filtering process is a mean filtering process.
4. The method according to claim 1, wherein said removing the interference area comprises: cavity area, over-exposed area in the endoscopic image.
5. The method for dynamically and intelligently detecting the endoscope bleeding area according to claim 1, wherein the process of constructing the binary template to perform image segmentation on the preprocessed RGB image comprises the following steps:
calculating the saturation component of the preprocessed RGB image;
taking 15% of the maximum saturation as a threshold value to generate a binary template;
and multiplying the preprocessed RGB image by the binary template to obtain a segmentation image.
6. The method as claimed in claim 1, wherein the method comprises setting a pixel value A of an image in a standard bleeding region, calculating an average pixel vector X of each of RGB segmented images, calculating a similarity of the average pixel vectors of the RGB segmented images, and expressing the similarity value
7. The method as claimed in claim 1, wherein the bleeding area detection model is a predetermined bleeding area detection model based on a neural network, and the output result is a bleeding image or a non-bleeding image.
8. A system for dynamic intelligent detection of an endoscope bleeding area, comprising a memory and a processor, wherein the memory includes a program for the method for dynamic intelligent detection of an endoscope bleeding area, and when the program for the method for dynamic intelligent detection of an endoscope bleeding area is executed by the processor, the following steps are implemented:
acquiring an endoscope image and converting the image into an RGB image by utilizing a linear interpolation algorithm;
preprocessing the RGB image, wherein the preprocessing comprises filtering, denoising and interference area removing;
constructing a binary template to perform image segmentation on the preprocessed RGB image;
setting a standard bleeding area image pixel value, calculating an average pixel vector of each segmented image of the RGB image, and calculating the similarity of the average pixel vectors of each segmented image of the RGB image pixel value of the standard bleeding area;
and if the similarity is greater than the preset value, inputting the segmented image into a preset bleeding area detection model classifier, and outputting a bleeding detection result.
9. The system of claim 8, wherein the linear interpolation algorithm is used to convert the image into RGB image, and the linear interpolation algorithm based on clinical domain is used. .
10. A computer-readable storage medium, characterized in that the computer-readable storage medium includes a program of a method for dynamic intelligent detection of an endoscope bleeding area, and when the program of the method is executed by a processor, the steps of the method for dynamic intelligent detection of an endoscope bleeding area according to any one of claims 1 to 7 are implemented.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010700705.XA CN111784686A (en) | 2020-07-20 | 2020-07-20 | Dynamic intelligent detection method, system and readable storage medium for endoscope bleeding area |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010700705.XA CN111784686A (en) | 2020-07-20 | 2020-07-20 | Dynamic intelligent detection method, system and readable storage medium for endoscope bleeding area |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111784686A true CN111784686A (en) | 2020-10-16 |
Family
ID=72764381
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010700705.XA Pending CN111784686A (en) | 2020-07-20 | 2020-07-20 | Dynamic intelligent detection method, system and readable storage medium for endoscope bleeding area |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111784686A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112842245A (en) * | 2020-12-30 | 2021-05-28 | 中原工学院 | Method and system for detecting friction injury between endoscopic clamp and intestinal tissue |
| CN113469996A (en) * | 2021-07-16 | 2021-10-01 | 四川大学华西医院 | Endoscope mucous membrane image reflection area detection and repair system |
| CN115153647A (en) * | 2022-07-05 | 2022-10-11 | 四川轻化工大学 | Intelligent pancreatic cancer detection method and platform based on ultrasonic endoscope |
| CN115937085A (en) * | 2022-06-28 | 2023-04-07 | 哈尔滨学院 | Nuclear cataract image processing method based on neural network learning |
-
2020
- 2020-07-20 CN CN202010700705.XA patent/CN111784686A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112842245A (en) * | 2020-12-30 | 2021-05-28 | 中原工学院 | Method and system for detecting friction injury between endoscopic clamp and intestinal tissue |
| CN113469996A (en) * | 2021-07-16 | 2021-10-01 | 四川大学华西医院 | Endoscope mucous membrane image reflection area detection and repair system |
| CN113469996B (en) * | 2021-07-16 | 2023-06-20 | 四川大学华西医院 | Endoscope mucous membrane image reflection of light region detects and repair system |
| CN115937085A (en) * | 2022-06-28 | 2023-04-07 | 哈尔滨学院 | Nuclear cataract image processing method based on neural network learning |
| CN115153647A (en) * | 2022-07-05 | 2022-10-11 | 四川轻化工大学 | Intelligent pancreatic cancer detection method and platform based on ultrasonic endoscope |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021036616A1 (en) | Medical image processing method, medical image recognition method and device | |
| US20220207728A1 (en) | Quality assessment in video endoscopy | |
| CN111784686A (en) | Dynamic intelligent detection method, system and readable storage medium for endoscope bleeding area | |
| US11514270B2 (en) | Speckle contrast analysis using machine learning for visualizing flow | |
| JP5113841B2 (en) | Computer-aided analysis using video from an endoscope | |
| CN110288597B (en) | Attention mechanism-based wireless capsule endoscope video saliency detection method | |
| US9754189B2 (en) | Detection device, learning device, detection method, learning method, and information storage device | |
| CN112446880B (en) | Image processing method, electronic device and readable storage medium | |
| CN111062947A (en) | Deep learning-based X-ray chest radiography focus positioning method and system | |
| US20130016198A1 (en) | Image processing device, endoscope apparatus, information storage device, and image processing method | |
| CN110473176B (en) | Image processing method and device, fundus image processing method and electronic equipment | |
| Ramaraj et al. | Homomorphic filtering techniques for WCE image enhancement | |
| You et al. | A simple and effective multi-focus image fusion method based on local standard deviations enhanced by the guided filter | |
| CN113850299B (en) | Gastrointestinal capsule endoscope video key frame extraction method with self-adaptive threshold | |
| DE112015002614T5 (en) | Image processing device, image processing method and image processing program | |
| CN110110750B (en) | Original picture classification method and device | |
| JP2005198890A (en) | Abnormal shadow detecting method, abnormal shadow detecting apparatus and program for the same | |
| WO2022250905A1 (en) | Specular reflection reduction in endoscope visualization | |
| Zhao et al. | An abnormality based WCE video segmentation strategy | |
| CN114663424A (en) | Endoscope video auxiliary diagnosis method, system, equipment and medium based on edge cloud cooperation | |
| Gadermayr et al. | Getting one step closer to fully automatized celiac disease diagnosis | |
| Vats et al. | SURF-SVM based identification and classification of gastrointestinal diseases in wireless capsule endoscopy | |
| Lim | Robust specular reflection removal and visibility enhancement of endoscopic images using 3-channel thresholding technique and image inpainting | |
| Flores et al. | Identifying precursory cancer lesions using temporal texture analysis | |
| CN117036905A (en) | Capsule endoscope image focus identification method based on HSV color space color attention |
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 |