CN113436183A - Image correlation analysis device - Google Patents
Image correlation analysis device Download PDFInfo
- Publication number
- CN113436183A CN113436183A CN202110779993.7A CN202110779993A CN113436183A CN 113436183 A CN113436183 A CN 113436183A CN 202110779993 A CN202110779993 A CN 202110779993A CN 113436183 A CN113436183 A CN 113436183A
- Authority
- CN
- China
- Prior art keywords
- image
- color
- edge
- pathological
- glandular
- 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
- 238000010219 correlation analysis Methods 0.000 title claims abstract description 15
- 230000001575 pathological effect Effects 0.000 claims abstract description 64
- 230000000762 glandular Effects 0.000 claims abstract description 56
- 238000004364 calculation method Methods 0.000 claims abstract description 11
- 238000005070 sampling Methods 0.000 claims description 23
- 230000011218 segmentation Effects 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 17
- 230000007170 pathology Effects 0.000 claims description 12
- 238000012937 correction Methods 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 9
- 238000003708 edge detection Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 238000003711 image thresholding Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000003745 diagnosis Methods 0.000 abstract description 8
- 238000004458 analytical method Methods 0.000 abstract description 7
- 238000010586 diagram Methods 0.000 description 9
- 238000000034 method Methods 0.000 description 7
- 239000013598 vector Substances 0.000 description 4
- 206010028980 Neoplasm Diseases 0.000 description 3
- 201000011510 cancer Diseases 0.000 description 3
- 210000004907 gland Anatomy 0.000 description 3
- 230000003321 amplification Effects 0.000 description 2
- 230000003902 lesion Effects 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000005513 bias potential Methods 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000799 fluorescence microscopy Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010827 pathological analysis Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10068—Endoscopic image
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Endoscopes (AREA)
Abstract
The application discloses image correlation analysis device, the device includes: the glandular tube color characteristic acquisition module is used for acquiring the color characteristic of the glandular tube in the endoscope image; the edge color feature acquisition module is used for acquiring color features of an edge zone in the pathological image; the similarity value calculation module is used for calculating the similarity value between the color feature of the duct and the color feature of the edge zone; and the correlation analysis module is used for analyzing the correlation between the endoscope image and the pathological image according to the calculated similarity value. The device effectively combines an endoscope image and a pathological image to carry out combined analysis, analyzes the pathological correlation of the endoscope and the pathological color characteristic similarity value, can assist doctors to diagnose complicated and various focuses, and improves the diagnosis precision.
Description
Technical Field
The present invention relates to the field of image analysis, and in particular, to an apparatus for analyzing image correlation.
Background
The early cancer detection generally adopts NBI (negative fluorescence imaging) amplified endoscopic images and pathological images for examination and analysis, and the current examination and analysis mode is mainly diagnosis according to personal experience of doctors.
However, physicians can make different judgments about the differences existing in the endoscope and pathological tables and the complicated and various lesions due to different personal experiences, resulting in different diagnosis results.
Therefore, how to improve the diagnosis precision is a technical problem to be solved urgently by those skilled in the art.
Disclosure of Invention
In view of the above, an object of the present invention is to provide an image correlation analysis device that can assist a physician in diagnosing complicated and various lesions and improve the accuracy of diagnosis. The specific scheme is as follows:
an apparatus for analyzing correlation of images, comprising:
the glandular tube color characteristic acquisition module is used for acquiring the color characteristic of the glandular tube in the endoscope image;
the edge color feature acquisition module is used for acquiring color features of an edge zone in the pathological image;
a similarity value calculation module for calculating a similarity value between the color feature of the duct and the color feature of the marginal zone;
and the correlation analysis module is used for analyzing the correlation between the endoscope image and the pathological image according to the calculated similarity value.
Preferably, in the apparatus for analyzing correlation between images provided in an embodiment of the present invention, the ductal color feature obtaining module includes:
the glandular tube Mask extracting unit is used for extracting the glandular tube Mask in the endoscopic image;
the endoscope image processing unit is used for carrying out image preprocessing on the endoscope image and extracting a saturation space;
the glandular duct part extracting unit is used for extracting the glandular duct part through which the straight line passes in the saturation space according to the calibrated straight line and the extracted glandular duct Mask;
and the saturation calculating unit is used for calculating a point with the highest saturation value in the region where each segment of the glandular part where the straight line passes through is, and the saturation value of the point is used as the color characteristic of the segment of the glandular pipe.
Preferably, in the apparatus for analyzing correlation of images according to an embodiment of the present invention, the ductal Mask extracting unit includes:
the segmentation model creating subunit is used for creating and training a U2Net segmentation model; the U2Net segmentation model comprises an encoder, a decoder and a feature map fusion module connected with the decoder and the last-stage encoder;
and the glandular tube Mask extraction subunit is used for reading the endoscope image through the trained U2Net segmentation model and outputting the glandular tube Mask in the endoscope image.
Preferably, in the apparatus for analyzing correlation between images according to an embodiment of the present invention, the endoscopic image processing unit includes:
the color correction subunit is used for performing color correction on the endoscope image by adopting a gray world white balance algorithm;
and the denoising processing subunit is used for denoising the endoscope image after color correction.
Preferably, in the apparatus for analyzing correlation of an image according to an embodiment of the present invention, the edge color feature obtaining module includes:
the pathological image processing unit is used for carrying out filtering denoising and graying processing on the pathological image;
an edge band extraction unit for extracting an edge band in the processed pathological image;
and the gray value calculating unit is used for intercepting the edge band at equal distance, taking the point with the lowest gray value in each intercepting block as a sampling point, and taking the difference value between 255 and the gray value of the sampling point as the color characteristic of the sampling point.
Preferably, in the apparatus for analyzing correlation between images according to an embodiment of the present invention, the edge band extracting unit includes:
the thresholding subunit is used for carrying out image thresholding on the processed pathological image to obtain a pathological binary image;
the edge detection subunit is used for carrying out edge detection on the pathology binary image to obtain an edge profile;
and the edge band extraction subunit is used for cutting the edge contour into an upper part and a lower part, calculating the edge average gray value of the two parts, and taking the half part contour with lower edge gray value as the edge band in the pathological image.
Preferably, in the apparatus for analyzing correlation between images according to an embodiment of the present invention, the similarity value calculating module includes:
the list mean value sampling unit is used for carrying out mean value sampling on the list corresponding to the color features of the edge zone to enable the length of the list corresponding to the color features of the edge zone to be consistent with that of the list corresponding to the color features of the duct;
and the cosine similarity value calculation unit is used for calculating the cosine similarity value between the lists corresponding to the two color features with the same length.
As can be seen from the above technical solutions, an image correlation analysis apparatus provided by the present invention includes: the glandular tube color characteristic acquisition module is used for acquiring the color characteristic of the glandular tube in the endoscope image; the edge color feature acquisition module is used for acquiring color features of an edge zone in the pathological image; the similarity value calculation module is used for calculating the similarity value between the color feature of the duct and the color feature of the edge zone; and the correlation analysis module is used for analyzing the correlation between the endoscope image and the pathological image according to the calculated similarity value.
The device provided by the invention effectively combines the endoscope image and the pathological image for combined analysis, analyzes the pathological correlation of the endoscope and the pathological color characteristic similarity value, can assist doctors in diagnosing complicated and various focuses, and improves the diagnosis precision.
Drawings
In order to more clearly illustrate the embodiments of the present invention or technical solutions in related arts, the drawings used in the description of the embodiments or related arts will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a schematic structural diagram of an apparatus for analyzing image correlation according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a U2Net segmentation model provided in an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an RSU module according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the step of extracting the gland Mask provided by the embodiment of the invention;
FIG. 5 is a pre-processed endoscopic image provided in accordance with an embodiment of the present invention;
FIG. 6 is a schematic diagram illustrating an endoscope image after being sampled according to an embodiment of the present invention;
FIG. 7 is a preprocessed pathology image provided in accordance with an embodiment of the present invention;
FIG. 8 is a binary chart of a pathology provided by an embodiment of the present invention;
FIG. 9 is an edge profile corresponding to a pathology binary image provided in accordance with an embodiment of the present invention;
fig. 10 is an edge band extracted from a pathological image according to an embodiment of the present invention;
fig. 11 is a schematic diagram of a pathological image after being sampled according to an embodiment of the present invention;
fig. 12 is a schematic diagram of mean sampling of a pathology gray value list according to an embodiment of the present invention;
fig. 13 is a length matching graph of the endoscope and pathological color feature point list according to the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, 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 invention.
The present invention provides an image correlation analysis device, as shown in fig. 1, comprising:
the glandular tube color characteristic acquisition module 11 is used for acquiring the color characteristic of the glandular tube in the endoscope image;
an edge color feature obtaining module 12, configured to obtain color features of an edge band in a pathological image;
a similarity value calculation module 13 for calculating a similarity value between the color feature of the duct and the color feature of the edge band;
and a correlation analysis module 14 for analyzing the correlation between the endoscopic image and the pathological image according to the calculated similarity value.
In the image correlation analysis device provided by the embodiment of the invention, the interaction of the four modules is effectively combined with an endoscope image and a pathological image for combined analysis, the pathological correlation of the endoscope and the pathological color feature similarity is analyzed, a doctor can be assisted in diagnosing complicated and various focuses, and the diagnosis precision is improved.
It should be noted that, the invention adopts the image characteristics of the endoscope image and the pathological image to carry out the combined analysis, and specifically, the color characteristic of the glandular tube in the endoscope image is required to be obtained, and the color of the glandular tube in the general cancer area is more red than that of the glandular tube in the non-cancer area; meanwhile, the nuclear pulp ratio color feature of the edge zone in the pathological image is acquired, in the reduced pathological image, the area with high nuclear pulp ratio has darker color and higher gray value due to more nuclear pulp ratio, and the area with low nuclear pulp ratio has lighter color and lower gray value due to more pulp ratio.
After extracting the glandular ducts of the endoscope image, sequentially collecting the color characteristic values of the glandular duct regions on the straight line according to the manually set straight line; then extracting edges by adopting a pathological image, sampling acquisition points at equal intervals, and counting the color characteristics of each point; and finally, calculating similarity values of the two color characteristic curves, and analyzing the correlation between the endoscope image and the pathological image according to the similarity values.
Further, in a specific implementation, in the apparatus for analyzing correlation between images provided in the embodiment of the present invention, the ductal color feature obtaining module 11 may include:
the glandular tube Mask extracting unit is used for extracting the glandular tube Mask in the endoscopic image;
the endoscope image processing unit is used for carrying out image preprocessing on an endoscope image and extracting a saturation space;
the glandular duct part extracting unit is used for extracting a glandular duct part through which the straight line passes in a saturation space according to the calibrated straight line and the extracted glandular duct Mask;
and the saturation calculating unit is used for calculating a point with the highest saturation value in the region where each segment of the glandular part where the straight line passes through is, and the saturation value of the point is used as the color characteristic of the segment of the glandular part.
In a specific implementation manner, in the apparatus for analyzing correlation between images according to an embodiment of the present invention, the ductal Mask extracting unit may include:
the segmentation model creating subunit is used for creating and training a U2Net segmentation model; the U2Net segmentation model comprises an encoder, a decoder and a feature map fusion module connected with the decoder and the last-stage encoder;
and the glandular duct Mask extraction subunit is used for reading an endoscope image through the trained U2Net segmentation model and outputting the glandular duct Mask in the endoscope image.
Specifically, the endoscopic ductal image is used for extracting a ductal Mask based on a U2Net segmentation model.As shown in fig. 2, U2Net is a network structure proposed based on Unet, and is a deep learning segmentation model similar to a U-Net model of an encoding-decoding (Encoder-Decoder) structure. In the U2Net segmentation model, each stage (stage) is composed of an RSU module (residual U-block), and a nested U structure can more effectively extract multi-scale features in the stage and multi-level features in an aggregation stage. As shown in fig. 3, the RSU module is mainly composed of three parts: the first part is the input convolutional layer: extracting partial feature of the common convolution layer, inputting feature diagram x (H × W × C)in) Is converted into a compound having CoutChannel number intermediate feature map F1(x) (ii) a The second part is U-Blovk: inputting an intermediate characteristic diagram F by adopting a Unet network structure1(x) Learning to extract and encode multiscale context information U (F)1(x) Multi-scale features are extracted from the gradient down-sampling feature map and are coded into the high-resolution feature map through devices such as gradual up-sampling, merging and convolution, and detail loss caused by large-scale direct up-sampling is reduced; the third part is the summation operation: f1(x)+U(F1(x) ) fuse local features and multi-scale features.
In practical application, before training a U2Net segmentation model, a large number of NBI (negative bias potential indicator) amplification endoscope images need to be collected, model training is carried out according to the gland Mask marked by a professional physician, and the model with a good storage effect is used for segmenting and extracting the gland Mask in the NBI amplification endoscope. As shown in fig. 4, after a U2Net segmentation model is created and the weight parameters in the trained model are loaded, the model reads an endoscope image, starts calculation processing, and calculates an output glandular tube Mask (that is, a black area is a background and a white area is a glandular tube), where the Mask is a binary image with the same size as the input image.
In a specific implementation manner, in the device for analyzing correlation between images according to an embodiment of the present invention, the endoscopic image processing unit may include:
the color correction subunit is used for performing color correction on the endoscope image by adopting a gray world white balance algorithm;
and the denoising processing subunit is used for denoising the endoscope image after the color correction.
Figure 5 shows an endoscopic image after preprocessing. The gray world white balance algorithm comprises the following specific steps:
Then, the gain coefficients for the three channels of image R, G, B are calculated:
the converted images R1, G1, B1 are:
then, extracting a saturation channel S from the endoscope image subjected to image preprocessing; according to the manually calibrated line segment and the extracted glandular duct Mask, as shown in fig. 6, extracting a glandular duct part through which the line segment passes in the saturation space S; calculating the point with the highest saturation value in each segment of glandular tract region where the straight line passes through, taking the point as the characteristic of the segment of glandular tract, and obtaining a glandular tract saturation value list with the length of n, namely a list corresponding to the color characteristic of the glandular tract, wherein the straight line passes through n glandular tracts:
[x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 … xn-3 xn-2 xn-1 xn]
the saturation is the purity of the color, and the higher the color is, the more pure the color is, and the lower the color is, the gray is gradually formed. In the digital image, the saturation value range is 0-255. The higher the saturation value, the purer the color, which means that the duct color is closer to red; the smaller the saturation value, the closer the duct color is to white.
Further, in a specific implementation, in the apparatus for analyzing correlation between images provided in the embodiment of the present invention, the edge color feature obtaining module 12 may include:
the pathological image processing unit is used for carrying out filtering denoising and graying processing on the pathological image;
an edge band extraction unit for extracting an edge band in the processed pathological image;
and the gray value calculating unit is used for intercepting the edge zones at equal intervals, taking the point with the lowest gray value in each intercepting block as a sampling point, and taking the difference value between 255 and the gray value of the sampling point as the color characteristic of the sampling point. .
In a specific implementation manner, in the apparatus for analyzing correlation between images according to an embodiment of the present invention, the edge band extracting unit may include:
the thresholding subunit is used for carrying out image thresholding on the processed pathological image to obtain a pathological binary image;
the edge detection subunit is used for carrying out edge detection on the pathological binary image to obtain an edge profile;
and the edge band extraction subunit is used for cutting the edge contour into an upper part and a lower part, calculating the edge average gray value of the two parts, and taking the half part contour with lower edge gray value as the edge band in the pathological image.
Taking the pathology image after the graying processing shown in fig. 7 as an example, the image thresholding is performed to obtain a pathology binary map shown in fig. 8; performing edge detection on the pathology binary image to obtain an edge profile shown in fig. 9; the contour is cut into an upper part and a lower part, the average gray value of the two parts is calculated, and as shown in fig. 10, the half part contour with lower edge gray value is taken as the edge zone of the research.
The next step is to sample the feature list: when the transverse length of the pathological image is w and the pathological edge zone is transversely intercepted by num pixels at equal intervals, taking the point with the lowest gray value in each intercepting block as a sampling point; taking the gray value of the sampling point, and taking the (255-gray value) as the color characteristic value of the point; each cut segment of edge is sampled in turn, as shown in fig. 11, to obtain a pathological edge gray list with length m ═ w/num, that is, a list corresponding to the color features of the edge band:
[y1 y2 y3 y4 y5 y6 y7 y8 y9 y10 … ym-3 ym-2 ym-1 ym]
it should be noted that, the black pixel value in the image is 0, the white is 255, the darker the pixel color is, the smaller the gray value is, and in the pathological analysis, the deeper the purple is, the smaller the gray value is, and in order to match the walking potential of the curve with the intention of the endoscope, the 255-gray value is adopted as the color gray value of the point in the pathological map.
Further, in a specific implementation, in the apparatus for analyzing correlation between images according to an embodiment of the present invention, the similarity value calculating module 13 may include:
the list mean value sampling unit is used for carrying out mean value sampling on the list corresponding to the color features of the edge zone, so that the length of the list corresponding to the color features of the edge zone (namely a pathological gray value list) is consistent with that of the list corresponding to the color features of the glandular tube (namely an endoscope saturation value list);
and the cosine similarity value calculation unit is used for calculating the cosine similarity value between the lists corresponding to the two color features with the same length.
It should be noted that, because the length of the endoscope image is significantly different from the length of the pathological image, the length of the sampled pathological gray value list is also different from the length of the endoscope saturation value list, and in order to calculate the cosine similarity between the pathological gray value list and the endoscope saturation value list, the pathological gray value list needs to be downsampled to make the pathological gray value list consistent with the endoscope saturation value list.
As shown in fig. 12 and 13, when the length of the endoscope saturation value list is n and the length of the pathology gray value list is m, the pathology gray value list is subjected to mean sampling by using a step length of L ═ m/n, so as to obtain a pathology edge gray value list with the length of n.
In one embodiment, the correlation analysis module 14 can use the cosine similarity value to evaluate the similarity between the color feature of the endoscope and the gray scale feature of the pathological edge. The cosine similarity is evaluated by calculating the cosine value of the included angle of two vectors. When the directions of the two vectors are the same, the cosine similarity value is 1; when the included angle of the two vectors is 90 degrees, the cosine similarity value is 0; when the two vectors are in opposite directions, the cosine similarity has a value of-1. The cosine similarity value ranges from-1 to 1, and the closer the value is to 1, the closer the directions of the two lists are, namely the more similar the two lists are.
Specifically, the cosine similarity value calculation formula is as follows:
where similarity is the calculated similarity value, xiIs the color characteristic of the glandular duct, yiIs the color feature of the marginal zone, and n is the length of the list corresponding to the color feature of the glandular tube and the list corresponding to the color feature of the marginal zone after treatment.
The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts among the embodiments are referred to each other.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
To sum up, an apparatus for analyzing image correlation according to an embodiment of the present invention includes: the glandular tube color characteristic acquisition module is used for acquiring the color characteristic of the glandular tube in the endoscope image; the edge color feature acquisition module is used for acquiring color features of an edge zone in the pathological image; the similarity value calculation module is used for calculating the similarity value between the color feature of the duct and the color feature of the edge zone; and the correlation analysis module is used for analyzing the correlation between the endoscope image and the pathological image according to the calculated similarity value. The device effectively combines the endoscope image and the pathological image for combined analysis, analyzes the pathological correlation of the endoscope and the pathological color characteristic similarity value, can assist doctors in diagnosing complicated and various focuses, and improves the diagnosis precision.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The image correlation analysis device provided by the present invention is described in detail above, and the principle and the implementation of the present invention are explained in detail herein by applying specific examples, and the description of the above examples is only used to help understanding the method and the core idea of the present invention; meanwhile, for a person 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 apparatus for analyzing correlation of images, comprising:
the glandular tube color characteristic acquisition module is used for acquiring the color characteristic of the glandular tube in the endoscope image;
the edge color feature acquisition module is used for acquiring color features of an edge zone in the pathological image;
a similarity value calculation module for calculating a similarity value between the color feature of the duct and the color feature of the marginal zone;
and the correlation analysis module is used for analyzing the correlation between the endoscope image and the pathological image according to the calculated similarity value.
2. The apparatus for analyzing correlation of image according to claim 1, wherein the duct color feature obtaining module comprises:
the glandular tube Mask extracting unit is used for extracting the glandular tube Mask in the endoscopic image;
the endoscope image processing unit is used for carrying out image preprocessing on the endoscope image and extracting a saturation space;
the glandular duct part extracting unit is used for extracting the glandular duct part through which the straight line passes in the saturation space according to the calibrated straight line and the extracted glandular duct Mask;
and the saturation calculating unit is used for calculating a point with the highest saturation value in the region where each segment of the glandular part where the straight line passes through is, and the saturation value of the point is used as the color characteristic of the segment of the glandular pipe.
3. The apparatus for analyzing correlation between images according to claim 2, wherein the ductal Mask extracting unit comprises:
the segmentation model creating subunit is used for creating and training a U2Net segmentation model; the U2Net segmentation model comprises an encoder, a decoder and a feature map fusion module connected with the decoder and the last-stage encoder;
and the glandular tube Mask extraction subunit is used for reading the endoscope image through the trained U2Net segmentation model and outputting the glandular tube Mask in the endoscope image.
4. The apparatus for analyzing correlation between images according to claim 2, wherein the endoscopic image processing unit comprises:
the color correction subunit is used for performing color correction on the endoscope image by adopting a gray world white balance algorithm;
and the denoising processing subunit is used for denoising the endoscope image after color correction.
5. The apparatus for analyzing correlation of images according to claim 1, wherein the edge color feature obtaining module includes:
the pathological image processing unit is used for carrying out filtering denoising and graying processing on the pathological image;
an edge band extraction unit for extracting an edge band in the processed pathological image;
and the gray value calculating unit is used for intercepting the edge band at equal distance, taking the point with the lowest gray value in each intercepting block as a sampling point, and taking the difference value between 255 and the gray value of the sampling point as the color characteristic of the sampling point.
6. The apparatus for analyzing correlation of an image according to claim 5, wherein the edge band extracting unit includes:
the thresholding subunit is used for carrying out image thresholding on the processed pathological image to obtain a pathological binary image;
the edge detection subunit is used for carrying out edge detection on the pathology binary image to obtain an edge profile;
and the edge band extraction subunit is used for cutting the edge contour into an upper part and a lower part, calculating the edge average gray value of the two parts, and taking the half part contour with lower edge gray value as the edge band in the pathological image.
7. The apparatus for analyzing correlation of images according to claim 1, wherein the similarity value calculating module comprises:
the list mean value sampling unit is used for carrying out mean value sampling on the list corresponding to the color features of the edge zone to enable the length of the list corresponding to the color features of the edge zone to be consistent with that of the list corresponding to the color features of the duct;
and the cosine similarity value calculation unit is used for calculating the cosine similarity value between the lists corresponding to the two color features with the same length.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110779993.7A CN113436183A (en) | 2021-07-09 | 2021-07-09 | Image correlation analysis device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110779993.7A CN113436183A (en) | 2021-07-09 | 2021-07-09 | Image correlation analysis device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113436183A true CN113436183A (en) | 2021-09-24 |
Family
ID=77759945
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110779993.7A Pending CN113436183A (en) | 2021-07-09 | 2021-07-09 | Image correlation analysis device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113436183A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107767365A (en) * | 2017-09-21 | 2018-03-06 | 华中科技大学鄂州工业技术研究院 | A kind of endoscopic images processing method and system |
CN112419135A (en) * | 2020-11-19 | 2021-02-26 | 广州华多网络科技有限公司 | Watermark recognition online training, sampling and removing method, device, equipment and medium |
CN113012252A (en) * | 2021-03-24 | 2021-06-22 | 苏州深透智能科技有限公司 | SPECT imaging prediction model creation method, device, equipment and storage medium |
US20210201499A1 (en) * | 2019-12-30 | 2021-07-01 | Medo Dx Pte. Ltd | Apparatus and method for image segmentation using a deep convolutional neural network with a nested u-structure |
-
2021
- 2021-07-09 CN CN202110779993.7A patent/CN113436183A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107767365A (en) * | 2017-09-21 | 2018-03-06 | 华中科技大学鄂州工业技术研究院 | A kind of endoscopic images processing method and system |
US20210201499A1 (en) * | 2019-12-30 | 2021-07-01 | Medo Dx Pte. Ltd | Apparatus and method for image segmentation using a deep convolutional neural network with a nested u-structure |
CN112419135A (en) * | 2020-11-19 | 2021-02-26 | 广州华多网络科技有限公司 | Watermark recognition online training, sampling and removing method, device, equipment and medium |
CN113012252A (en) * | 2021-03-24 | 2021-06-22 | 苏州深透智能科技有限公司 | SPECT imaging prediction model creation method, device, equipment and storage medium |
Non-Patent Citations (2)
Title |
---|
SHENGSEN CHEN 等: "The Pink Pattern Visualized in Magnifying Endoscopy with Narrow-Band Imaging is a New Feature of Early Gastric Differentiated Cancer--A Bridge Between Endoscopic Image and Histopathological Change", 《RESEARCH SQUARE》 * |
XUEBIN QIN 等: "U2-Net: Going deeper with nested U-structure for salient object detection", U2-NET: GOING DEEPER WITH NESTED U-STRUCTURE FOR SALIENT OBJECT DETECTION * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111667489B (en) | Cancer hyperspectral image segmentation method and system based on double-branch attention deep learning | |
CN109035263B (en) | Automatic brain tumor image segmentation method based on convolutional neural network | |
CN111407245B (en) | Non-contact heart rate and body temperature measuring method based on camera | |
CN112017191A (en) | Method for establishing and segmenting liver pathology image segmentation model based on attention mechanism | |
CN111429473B (en) | Chest film lung field segmentation model establishment and segmentation method based on multi-scale feature fusion | |
CN109146872B (en) | Heart coronary artery image segmentation and identification method based on deep learning and optical flow method | |
CN111951288B (en) | Skin cancer lesion segmentation method based on deep learning | |
CN111784671A (en) | Pathological image focus region detection method based on multi-scale deep learning | |
CN113034505B (en) | Glandular cell image segmentation method and glandular cell image segmentation device based on edge perception network | |
WO2008044466A1 (en) | Image processing device, image processing method, and image processing program | |
CN113420826B (en) | Liver focus image processing system and image processing method | |
CN113781455B (en) | Cervical cell image anomaly detection method, device, equipment and medium | |
CN113450328A (en) | Medical image key point detection method and system based on improved neural network | |
CN113744271A (en) | Neural network-based automatic optic nerve segmentation and compression degree measurement and calculation method | |
CN115909006A (en) | Mammary tissue image classification method and system based on convolution Transformer | |
CN115908449A (en) | 2.5D medical CT image segmentation method and device based on improved UNet model | |
CN117152433A (en) | Medical image segmentation method based on multi-scale cross-layer attention fusion network | |
CN113298826B (en) | Image segmentation method based on LA-Net network | |
CN114022539A (en) | Microscopic image cell position judgment method | |
CN113436183A (en) | Image correlation analysis device | |
CN115132275B (en) | Method for predicting EGFR gene mutation state based on end-to-end three-dimensional convolutional neural network | |
CN115994999A (en) | Goblet cell semantic segmentation method and system based on boundary gradient attention network | |
CN109377478B (en) | Automatic grading method for osteoarthritis | |
CN112488979A (en) | Endoscope image recognition method | |
CN117392117B (en) | Pressure sore detection method based on OpenCV image processing |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210924 |
|
RJ01 | Rejection of invention patent application after publication |