An intelligent analysis system for internal medicine medical imaging data

Through the intelligent analysis system, the detection accuracy problems under the influence of contrast agent flow and esophageal peristalsis in barium meal examination were solved, and the detection accuracy and diagnostic auxiliary effect of the lesion area were improved.

CN118587220BActive Publication Date: 2024-11-05DALIAN QINGDONG TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411073746.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-07
Publication Date
2024-11-05
Estimated Expiration
2044-08-07

Smart Images

  • Figure CN118587220B_ABST
    Figure CN118587220B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image analysis technology, and in particular to an intelligent analysis system for medical imaging data in internal medicine, which system acquires a number of esophageal images; obtains suspected lesion areas and suspected normal areas based on the esophageal images; obtains abnormal features and normal features based on the suspected lesion areas and suspected normal areas, and obtains the possibility of lesions based on the abnormal features and normal features; obtains a reference area based on the possibility of lesions; obtains the degree of abnormality based on the normal features of the possibility of lesions and suspected normal areas; obtains a lesion area based on the degree of abnormality; obtains the degree of lesion based on the lesion area and its degree of abnormality; and obtains a marked lesion esophageal image based on the lesion area and its degree of lesion. The present invention improves the accuracy of lesion area detection by acquiring the lesion area and the degree of lesion and visually marking them on the esophageal image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to an intelligent analysis system for internal medicine image data. Background Art

[0002] In the field of internal medicine, barium meal radiography is an important means of clinically diagnosing esophageal cancer, and it can more intuitively display the pathological conditions of the esophageal cavity. The barium meal examination requires oral administration of an appropriate amount of contrast agent, and the lesion area and degree of the lesion are analyzed based on the esophageal imaging after taking the contrast agent. Since it takes a certain amount of time for the contrast agent to pass through the esophagus, during the barium meal examination, multiple X-rays are usually taken of the same location. The contour, shape, and peristalsis of the esophagus need to be analyzed based on the obtained X-rays to determine the lesion area and degree of the lesion.

[0003] Since it takes a certain amount of time for the contrast medium to pass through the esophagus, during the barium meal examination, the esophageal images are usually taken multiple times at the same location to observe the contour, shape and peristalsis of the esophagus. Due to the flow of the contrast medium in the esophagus and the influence of the peristalsis of the esophagus itself, the target image may be disconnected or blurred, which may interfere with the detection results of the lesion area. Summary of the invention

[0004] The present invention provides an intelligent analysis system for internal medicine medical imaging data to solve the existing problems.

[0005] The intelligent analysis system of internal medicine imaging data of the present invention adopts the following technical scheme:

[0006] An embodiment of the present invention provides an intelligent analysis system for internal medicine medical imaging data, the system comprising the following modules:

[0007] Medical image acquisition module: used to sequentially acquire a plurality of esophageal images of the patient to be detected, wherein the plurality of esophageal images are esophageal images of the patient at multiple times after taking contrast agent; the esophageal images include the esophageal area;

[0008] Suspected lesion differentiation module: used to obtain suspected lesion areas and suspected normal areas in the esophageal image based on the width of the esophageal area in the esophageal image and the distribution of the esophageal area between different esophageal images;

[0009] Reference area screening module: used to obtain abnormal features of the suspected lesion area and normal features of the suspected normal area according to the grayscale distribution features and texture features in the suspected lesion area and the suspected normal area, and obtain the possibility of the lesion in the suspected lesion area according to the abnormal features and the normal features combined with the morphological conditions of the suspected lesion area; and screen several reference areas according to the possibility of the lesion in the suspected lesion area;

[0010] Lesion analysis module: used to obtain the abnormality of the reference area according to the difference in the possibility of lesions between the reference areas in the esophageal image and the normal characteristics of the suspected normal area; to screen the lesion area according to the abnormality of the reference area; to obtain the lesion degree of the lesion area according to the distribution of the lesion area in the esophageal image and the abnormality of the lesion area;

[0011] Lesion area marking module: used to mark the lesion area in the esophagus image according to the morphological distribution of the lesion area in the esophagus image and the degree of the lesion in the lesion area, so as to obtain a marked lesion esophagus image.

[0012] Furthermore, the suspected lesion area and the suspected normal area in the esophageal image are obtained based on the width of the esophageal area in the esophageal image and the distribution of the esophageal area between different esophageal images, including:

[0013] The average esophageal width of all esophageal regions in the esophageal image is obtained, and the esophageal region with an esophageal width greater than the average esophageal width is recorded as a characteristic stenosis region;

[0014] According to the position distribution of characteristic narrow areas between esophageal images, several suspected lesion areas in each esophageal image are obtained;

[0015] Each esophageal region except the independent stenosis region in the target esophageal image is recorded as an independent non-stenosis region; an area consisting of several rows of continuous independent non-stenosis regions along the esophageal region distribution direction in the target esophageal image is recorded as a suspected normal area.

[0016] Furthermore, the plurality of suspected lesion areas in each esophagus image obtained according to the position distribution of the characteristic stenosis areas between the esophagus images include:

[0017] Any esophageal image is recorded as a target esophageal image, and all characteristic stenosis areas in all esophageal images except the target esophageal image are marked as related stenosis areas in the target esophageal image; all characteristic stenosis areas and all related stenosis areas in the target esophageal image are collectively referred to as independent stenosis areas; an area consisting of several rows of independent stenosis areas in the target esophageal image that are continuous along the esophageal area distribution direction is recorded as a suspected lesion area.

[0018] Furthermore, the method for acquiring abnormal features of the suspected lesion area includes:

[0019] According to the edge smoothness of the suspected lesion area and the disorder of the grayscale texture in the suspected lesion area, the abnormal characteristics of the suspected lesion area are obtained.

[0020] Furthermore, the method for acquiring the normal features of the suspected normal area includes:

[0021] According to the difference in esophageal width between esophageal regions in the suspected normal region, the edge smoothness of the suspected normal region, and the grayscale uniformity of the suspected normal region, the normal features of the suspected normal region are obtained.

[0022] Furthermore, the method for obtaining the lesion possibility of the suspected lesion area includes:

[0023] For any suspected lesion area, one side of the suspected lesion area is selected as the target side, and the lesion bending vector and the normal bending vector of the suspected lesion area are obtained according to the position distribution of the pixel points at the junction of the suspected lesion area and the suspected normal area adjacent to the suspected lesion area in the target side direction;

[0024] According to the difference between the lesion bending vector and the normal bending vector of the suspected lesion area, the abnormal characteristics, the difference in esophageal width, and the normal characteristics of the suspected normal area adjacent to the suspected lesion area, the initial lesion possibility on the target side of the suspected lesion area is obtained; the average of the initial lesion possibilities on both sides of the suspected lesion area is recorded as the lesion possibility of the suspected lesion area.

[0025] Furthermore, the method for obtaining the pathological bending vector and the normal bending vector includes:

[0026] For adjacent suspected lesion areas and suspected normal areas, a neighborhood range of the boundary edge pixels between the two areas is obtained; the edges corresponding to the target side of the neighborhood range are the sub-edges of the suspected lesion area and the sub-edges of the suspected normal area. Within the neighborhood range, the sum vector of the vectors between the two adjacent edge pixels of the sub-edge of the suspected lesion area from the boundary pixel along the direction away from the suspected normal area is used as the normal bending vector; within the neighborhood range, the sum vector of the vectors between the two adjacent edge pixels of the sub-edge of the suspected normal area from the boundary pixel along the direction away from the suspected lesion area is used as the lesion bending vector.

[0027] Furthermore, the method for obtaining the abnormality degree of the reference area includes:

[0028] According to the difference in the possibility of lesions between the reference areas and the normal characteristics of the suspected normal areas adjacent to the reference areas, the abnormality index of the reference areas is obtained;

[0029] According to the difference of normal features of adjacent suspected normal areas at the same position of the reference area between the esophageal images, an adjacent change index of the reference area is obtained;

[0030] The abnormality degree of the reference area is obtained according to the difference of the abnormality index of the reference area at the same position between the esophageal images and the adjacent change index of the reference area.

[0031] Furthermore, the method for obtaining the lesion degree of the lesion area includes:

[0032] For any esophageal image, the ratio of the lesion area to the esophageal length of the esophageal image where the lesion area is located is recorded as the length lesion value of the lesion area;

[0033] The ratio of the average value of the esophageal width of all esophageal regions in the lesion region to the average value of the esophageal width of all esophageal regions in the esophageal image where the lesion region is located is recorded as the esophageal lesion value of the lesion region;

[0034] The normalized result of the product of the abnormality degree of the lesion area, the length lesion value and the esophageal lesion value is recorded as the lesion degree of the lesion area.

[0035] Furthermore, the method for acquiring the esophageal image with marked lesions includes:

[0036] For any lesion area, the marking color value of the lesion area is obtained according to the lesion degree of the lesion area; the minimum circumscribed rectangle of the lesion area is obtained, and the pixel points on the minimum circumscribed rectangle of the lesion area are marked using the marking color value of the lesion area to obtain a marked lesion esophageal image.

[0037] The beneficial effects of the technical solution of the present invention are as follows: the present invention preliminarily screens the esophageal area through the feature that the esophageal width of the esophageal lesion area is significantly reduced relative to the esophageal normal area, obtains the suspected lesion area and the suspected normal area, and narrows the judgment range of the lesion area; obtains the abnormal features of the suspected lesion area and the normal features of the suspected normal area through the features that the edge of the esophageal lesion area is rough and not smooth, the grayscale distribution of the esophageal lesion area is uneven, and the texture is messy, and the abnormal features of the suspected lesion area and the normal features of the suspected normal area are comprehensively judged. The lesion possibility of the suspected lesion area is further screened, and the reliability of the lesion area detection result is improved; the esophageal lesion will affect the peristalsis of the esophagus, so the abnormal degree of the reference area is obtained through the change of the lesion possibility of the reference area in different esophageal images and the change of the normal features of the suspected normal area adjacent to the reference area, and the lesion area is screened according to the abnormal degree, and the lesion degree of the lesion area is obtained. So far, the present invention visually marks the lesion area and its lesion degree on the esophageal image to improve the accuracy of lesion area detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 This is a module flow chart of an intelligent analysis system for internal medicine imaging data of the present invention;

[0040] Figure 2 An esophagus image provided in this embodiment. DETAILED DESCRIPTION

[0041] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation, structure, features and effects of an intelligent analysis system for medical imaging data of internal medicine proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0042] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0043] The specific scheme of the intelligent analysis system for internal medicine image data provided by the present invention is described in detail below with reference to the accompanying drawings.

[0044] See also Figure 1 , which shows a module flow chart of an internal medicine medical image data intelligent analysis system provided by an embodiment of the present invention, the system includes the following modules:

[0045] Module 101: A medical image acquisition module, used to sequentially acquire a number of esophagus images of a patient to be examined.

[0046] It should be noted that the purpose of this embodiment is to analyze the suspected lesion area and the suspected normal area of ​​the esophagus region in the esophagus image by combining multiple esophageal images according to the symptom characteristics of medullary esophageal cancer, obtain the lesion area and the degree of lesion, provide auxiliary diagnosis for doctors in the diagnosis process, and improve the accuracy of the diagnosis results. Therefore, it is necessary to collect esophageal images first.

[0047] Specifically, according to a plurality of preset acquisition moments, the esophageal image of the patient who has taken contrast agent is collected at each acquisition moment in turn using a fluoroscopic device to obtain a plurality of esophageal images, wherein the plurality of preset acquisition moments are 1 minute, 10 minutes, 20 minutes and 30 minutes after taking the contrast agent, respectively. This embodiment is described using this as an example.

[0048] Module 102: A suspected lesion distinguishing module, which is used to obtain a suspected lesion area and a suspected normal area in the esophagus image according to the width of the esophagus area in the esophagus image and the distribution of the esophagus area between different esophagus images.

[0049] It should be noted that when analyzing the symptom characteristics of medullary esophageal cancer based on esophageal images, it is necessary to distinguish between the suspected lesion area and the suspected normal area in the esophageal image. Since the cancer tissue grows invasively in the esophageal wall and involves the entire circumference or most of the esophagus, the width of the esophageal lesion area is significantly reduced relative to the normal esophageal area, so the suspected lesion area and the suspected normal area in the esophageal image are distinguished accordingly. Therefore, based on the width of the esophageal area in the esophageal image and the distribution of the esophageal area between different esophageal images, the suspected lesion area and the suspected normal area in the esophageal image are judged.

[0050] In one embodiment of the present invention, Figure 2 The esophageal image shown is taken from the front of the human body, so in the embodiment of the present invention, the method for obtaining the esophageal area includes: obtaining a number of edge pixel points using an edge detection algorithm for the esophageal image; for any row of horizontally arranged pixels in the esophageal image, calculating the Euclidean distance between the edge pixel points in the row, selecting the area between the two edge pixel points with the largest Euclidean distance as an esophageal area, and recording the Euclidean distance between the edge pixel points with the farthest Euclidean distance in the row as the esophageal width of the esophageal area; wherein the edge detection algorithm is a well-known technology, and the specific method is not introduced here.

[0051] In other embodiments of the present invention, other methods such as semantic segmentation, image segmentation, edge extraction, etc. may also be used to obtain the esophagus region, which is not limited here.

[0052] Preferably, the average esophageal width of all esophageal regions in the esophageal image is obtained, and the esophageal region with an esophageal width greater than the average esophageal width is recorded as a characteristic stenosis region;

[0053] It should be noted that, since the width of the esophagus in the esophageal lesion area is significantly smaller than that in the normal esophageal area, a preliminary screening of the esophageal area was performed based on the esophageal width.

[0054] It should be further explained that after the patient orally ingests the contrast agent, the contrast agent will flow in the patient's digestive tract. Depending on the time of acquisition of the esophageal image, the obviousness of the lesion features in different esophageal images is different. In order to more accurately determine the location of the lesion area in each esophageal image, multiple esophageal images are combined to comprehensively determine the suspected lesion area. That is, according to the location distribution of the characteristic narrow areas between the esophageal images, several suspected lesion areas in each esophageal image are obtained.

[0055] Furthermore, each esophageal region except the independent stenosis region in the target esophageal image is recorded as an independent non-stenosis region; and a region consisting of several rows of continuous independent non-stenosis regions along the esophageal region distribution direction in the target esophageal image is recorded as a suspected normal region.

[0056] Preferably, a plurality of suspected lesion areas in each esophagus image are obtained according to the position distribution of characteristic stenosis areas between esophagus images;

[0057] Specifically, the method for obtaining the suspected lesion area is as follows: any esophageal image is recorded as a target esophageal image, and all characteristic stenosis areas in all esophageal images except the target esophageal image are marked as related stenosis areas corresponding to all esophageal areas in the target esophageal image; all characteristic stenosis areas and all related stenosis areas in the target esophageal image are collectively referred to as independent stenosis areas; an area consisting of independent stenosis areas that are continuous in a certain row of positions along the esophageal area distribution direction in the target esophageal image is recorded as a suspected lesion area;

[0058] Module 103: Reference area screening module: used to obtain abnormal features of suspected lesion areas and normal features of suspected normal areas based on the grayscale distribution features and texture features in the suspected lesion areas and the suspected normal areas, and obtain the possibility of lesions in the suspected lesion areas based on the abnormal features and the normal features in combination with the morphological conditions of the suspected lesion areas; and screen out several reference areas based on the possibility of lesions in the suspected lesion areas.

[0059] It should be noted that the suspected lesion area and suspected normal area obtained based on the difference between the esophageal width of each esophageal area and the average esophageal width only perform a preliminary screening of the esophageal area. The suspected lesion area and the suspected normal area need to be further analyzed and the suspected lesion area is further screened. Due to the invasive growth of cancer tissue, the mucosa in the esophageal lesion area disappears or is destroyed, and the mucosal texture becomes coarse and disordered, resulting in a rough and uneven esophageal edge, and there is a dense soft tissue shadow in the esophagus. The esophageal lesion area shows uneven grayscale distribution and messy texture. Therefore, the suspected lesion area is further screened based on this. First, based on the abnormal features of the suspected lesion area and the normal features of the suspected normal area in the esophageal image, combined with the morphological conditions of the suspected lesion area, the possibility of lesions in the suspected lesion area is obtained, and then the suspected lesion area is further screened.

[0060] Preferably, the abnormal features of the suspected lesion area are obtained according to the smoothness of the edge of the suspected lesion area and the disorder of the grayscale texture in the suspected lesion area.

[0061] As an example, the abnormal characteristics of the suspected lesion area are expressed by the formula:

[0062]

[0063] In the formula, For the abnormal features of the suspected lesion area; For the The variance of the edge chain code value of the suspected lesion area. The method for obtaining the edge chain code value is a well-known technology, and the specific method is not introduced here; For the The information entropy of the grayscale values ​​of all pixels in the suspected lesion area; For the The number of pixels in the suspected lesion area; For the In the suspected lesion area The maximum gray value of all pixels within the eight neighborhoods of a pixel; For the In the suspected lesion area The minimum grayscale value of all pixels within the eight neighborhoods of a pixel.

[0064] It should be noted that The larger the value, the The less smooth the edge of the suspected lesion area is; The larger the value, the The more chaotic the texture distribution of the suspected lesion area is; The larger the value, the The more uneven the grayscale distribution is in the suspected lesion area, the The more suspected lesion areas are, the more likely they are to have lesions.

[0065] It should be noted that the normal esophagus has uniform grayscale distribution and smooth edges. Since the esophagus is squeezed by surrounding organs, it will produce physiological stenosis. There will be relative bulging of the esophagus between the two narrow areas. The normal characteristics of the suspected normal area are analyzed based on this.

[0066] Preferably, the normal features of the suspected normal area are obtained according to the difference in esophageal width between esophageal areas in the suspected normal area, the edge smoothness of the suspected normal area, and the grayscale uniformity of the suspected normal area.

[0067] As an example, the normal characteristics of a suspected normal area are expressed as:

[0068]

[0069] In the formula, For the Normal features of suspected normal areas; For the The maximum value of the esophageal width of all esophageal areas in the suspected normal area; For the The minimum value of the esophageal width among all esophageal areas in the suspected normal area; For the The variance of the grayscale values ​​of all pixels in the suspected normal area; For the The maximum value of the number of consecutive identical chain code values ​​in the edge chain code values ​​of the suspected normal area. The method for obtaining the edge chain code value is a well-known technology, and the specific method is not introduced here; For the The number of pixels in the suspected normal area; is a linear normalization function, and the normalized object is all suspected normal areas .

[0070] It should be noted that The smaller the value of The more uniform the grayscale distribution of the suspected normal area is; The larger the value, the The greater the bulging of the suspected normal area; The larger the value, the The smoother the edge of the suspected normal area, the The larger the normal features of a suspected normal area.

[0071] It should be further explained that, since medullary esophageal cancer can cause the esophageal area adjacent to the upper side of the lesion area to expand, and the stenosis of the lesion area is irregular, the expansion degree of the upper esophagus is analyzed based on the angle between the lesion area and the upper esophagus. The smaller the angle, the greater the expansion degree of the upper esophagus, and the greater the influence of the suspected area on the upper esophagus. Therefore, the greater the normal characteristics of the suspected normal area adjacent to the upper side of the suspected lesion area, the less the suspected normal area is affected by the suspected lesion area, and the less likely the suspected lesion area is to belong to the lesion area.

[0072] Preferably, for any suspected lesion area, one side of the suspected lesion area is selected as the target side, and the lesion bending vector and the normal bending vector of the suspected lesion area are obtained according to the position distribution of pixel points at the junction of the suspected normal area adjacent to the suspected lesion area in the direction of the target side.

[0073] Specifically, the method for obtaining the lesion bending vector and the normal bending vector is as follows: for adjacent suspected lesion areas and suspected normal areas, a neighborhood range of the boundary edge pixel points between the two areas is obtained; the edge corresponding to the target side of the neighborhood range is the suspected lesion area sub-edge and the suspected normal area sub-edge, and the suspected lesion area sub-edge and the suspected normal area sub-edge respectively include edge pixels; among them, is the preset reference quantity, =5, this embodiment is described as an example; within the neighborhood range, the sum vector of the vectors between the two adjacent edge pixels from the boundary pixel point along the direction back to the suspected normal area of ​​the sub-edge of the suspected lesion area is used as the normal bending vector, and within the neighborhood range, the sum vector of the vectors between the two adjacent edge pixels from the boundary pixel point along the direction back to the suspected lesion area of ​​the sub-edge of the suspected normal area is used as the lesion bending vector; it should be noted that if the edge pixel points of the suspected lesion area sub-edge or the suspected normal area sub-edge within the neighborhood range are not enough When the normal bending vector and the lesion bending vector are obtained, only the edge pixel points included in the suspected lesion area sub-edge and the suspected normal area sub-edge within the neighborhood range are used.

[0074] Furthermore, for any suspected lesion area, the initial lesion possibility on the target side of the suspected lesion area is obtained based on the difference between the lesion bending vector and the normal bending vector of the suspected lesion area, the abnormal characteristics, the difference in esophageal width, and the normal characteristics of the suspected normal area adjacent to the suspected lesion area; the average of the initial lesion possibilities on both sides of the suspected lesion area is recorded as the lesion possibility of the suspected lesion area.

[0075] As an example, the initial lesion likelihood is formulated as:

[0076]

[0077] In the formula, For the The initial lesion probability of the target side of the suspected lesion area; For the abnormal features of the suspected lesion area; For the The variance of esophageal width in all esophageal regions within the suspected lesion area; For the The minimum angle between the lesion bending vector and the normal bending vector of the suspected lesion area; For the The mean esophageal width of all esophageal areas within the suspected lesion area; For the The normal features of the suspected normal area adjacent to the suspected lesion area; is an exponential function with a natural constant as the base; 1 is to prevent is a negative number; is a linear normalization function, and the normalized object is all suspected lesion areas .

[0078] It should be noted that The larger the value, the The greater the degree of dilation of the upper esophagus of the suspected lesion area; The larger the value, the closer it is to the The greater the degree of interference of the suspected normal area adjacent to the upper side of the suspected lesion area by the suspected lesion area, the greater the possibility that the suspected lesion area is the lesion area; The smaller the value, the The more suspected lesion areas are, the greater the possibility of lesions; The larger the The more discrete the esophageal width of all esophageal areas in the suspected lesion area, the The more suspected lesion areas are, the greater the possibility of lesions; The larger the The more suspected lesion areas there are, the greater the possibility of lesions.

[0079] Furthermore, the lesion probability is greater than the preset probability threshold. The suspected lesion area is recorded as the reference area; among them, the preset possibility threshold This embodiment is described using this as an example.

[0080] Module 104: A lesion analysis module, which is used to obtain the abnormality of the reference area according to the difference in the possibility of lesions between the reference areas in the esophageal image and the normal characteristics of the suspected normal area; to screen the lesion area according to the abnormality of the reference area; and to obtain the lesion degree of the lesion area according to the distribution of the lesion area in the esophageal image and the abnormality of the lesion area.

[0081] It should be noted that since different esophageal images are collected at different times, the degree of obviousness of the lesions is also different, so it is necessary to combine multiple esophageal images to further screen the reference area. Under normal circumstances, the esophagus itself will perform regular peristalsis, and the lesion area will become stiff and the peristalsis will be weakened due to cancer. The degree of esophageal peristalsis is mainly reflected in the change of the possibility of lesions in the reference area in different esophageal images and the change of normal characteristics of the suspected normal area adjacent to the upper side of the reference area, so the degree of lesions in the lesion area is judged accordingly. First, according to the difference in the possibility of lesions between the reference areas and the normal characteristics of the suspected normal area adjacent to the upper side of the reference area, the abnormal index of the reference area is obtained, and then according to the difference in normal characteristics of the suspected normal area adjacent to the upper side of the reference area at the same position between the esophageal images, the adjacent change index of the reference area is obtained, and the abnormal index of the reference area and the adjacent change index are combined for judgment to obtain the degree of lesions in the lesion area.

[0082] Preferably, the abnormality index of the reference area is obtained according to the difference in the possibility of lesions between the reference areas and the normal characteristics of the suspected normal area adjacent to the reference area.

[0083] Specifically, the method for obtaining the abnormal index is: record any reference area as the target reference area, calculate the difference between the lesion possibility of the target reference area and the minimum value of the lesion possibility of all reference areas in the esophageal image where the target reference area is located, and record the ratio of the difference to the normal characteristics of the suspected normal area adjacent to the upper side of the target reference area as the abnormal index of the target reference area.

[0084] It should be noted that the greater the difference between the lesion probability of the target reference area and the minimum value of the lesion probability of all reference areas in the esophageal image, the less the target reference area conforms to the characteristics of a normal esophagus, and the greater the abnormal index of the target reference area.

[0085] Preferably, the adjacent change index of the reference area is obtained according to the difference in normal features between adjacent suspected normal areas of the reference area at the same position between the esophagus images.

[0086] As an example, the method for obtaining the adjacent change index of the benchmark area is:

[0087] For any reference region, the suspected lesion region with the same position as the reference region in each esophageal image is recorded as a characteristic region of the reference region, and the suspected normal region adjacent to the upper side of each characteristic region of the reference region is recorded as a characteristic reference region of the reference region. The calculation formula of the adjacent change index of the reference region is:

[0088]

[0089] In the formula, For the neighboring change indicators of the benchmark area; is the number of esophageal images; For the The benchmark area is Normal features of the characteristic reference area in the esophageal image; For the The benchmark area is Normal features of the characteristic reference area in the esophageal image; is the absolute value function.

[0090] Preferably, the abnormality degree of the reference area is obtained according to the difference of abnormality indicators of the reference area at the same position between the esophagus images and the adjacent change indicators of the reference area.

[0091] Specifically, the method for obtaining the abnormality degree is as follows: for any reference area, the difference between the maximum value and the minimum value of the abnormality index of the characteristic area of ​​the reference area in all esophageal images is calculated, and the linear normalization result of the ratio of the difference to the mean value of the adjacent change index of the characteristic reference area of ​​the reference area in all esophageal images is recorded as the abnormality degree of the reference area;

[0092] Specifically, the abnormality level is greater than the preset abnormality threshold. The reference area is recorded as the lesion area; the abnormal threshold is preset This embodiment is described using this as an example.

[0093] It should be noted that when judging the degree of lesion in the lesion area, it is necessary to analyze based on the distribution of the lesion area in the esophageal image and the degree of abnormality of the lesion area. The larger the proportion of the lesion area in all esophageal areas, the greater the degree of abnormality of the lesion area, and the more abnormal the esophageal width of the lesion area, the greater the degree of lesion in the lesion area.

[0094] Preferably, the lesion degree of the lesion area is obtained according to the distribution of the lesion area in the esophagus image and the abnormality degree of the lesion area.

[0095] Specifically, the method for obtaining the degree of lesions is as follows: for any esophageal image, the number of rows between two esophageal regions with the largest number of rows in all esophageal regions in the esophageal image is recorded as the esophageal length of the esophageal image; for any lesion region, the number of pixel rows occupied by the lesion region is recorded as the region length of the lesion region; the ratio of the lesion region to the esophageal length of the esophageal image where the lesion region is located is recorded as the length lesion value of the lesion region;

[0096] The ratio of the average value of the esophageal width of all esophageal regions in the lesion region to the average value of the esophageal width of all esophageal regions in the esophageal image where the lesion region is located is recorded as the esophageal lesion value of the lesion region;

[0097] The linear normalization result of the product of the abnormality degree of the lesion area, the length lesion value and the esophageal lesion value is recorded as the lesion degree of the lesion area.

[0098] As an example, the degree of lesion is expressed as:

[0099]

[0100] In the formula, For the The degree of lesion in each lesion area; For the The degree of abnormality in the lesion area; For the The length of the lesion area; For the The length of the esophagus in the esophageal image where the lesion area is located; For the The mean esophageal width of all esophageal regions in the lesion area; For the The mean value of the esophageal width of all esophageal areas in the esophageal image where the lesion area is located; is a linear normalization function, and the normalized object is all lesion areas .

[0101] Module 105: a lesion region marking module, which is used to mark the lesion region in the esophagus image according to the morphological distribution of the lesion region in the esophagus image and the lesion degree of the lesion region, so as to obtain a marked lesion esophagus image.

[0102] It should be noted that after obtaining the lesion area and the degree of the lesion, the lesion area and the degree of the lesion need to be visualized on the esophageal image to assist doctors in diagnosing medullary esophageal cancer.

[0103] Preferably, for any lesion area, a marking color value of the lesion area is obtained according to the lesion degree of the lesion area; the minimum circumscribed rectangle of the lesion area is obtained, and the pixel points on the minimum circumscribed rectangle of the lesion area are marked using the marking color value of the lesion area to obtain a marked lesion esophageal image.

[0104] Specifically, the method for obtaining the marking color value is as follows: for any lesion area, the product of the lesion degree of the lesion area and 255 is recorded as the marking color value of the lesion area; wherein 255 is the maximum value of the R channel in the RGB image. The method for obtaining the marked lesion esophageal image is as follows: for any lesion area in any esophageal image, the esophageal image is converted into an RGB image, and for any pixel point at the corresponding position of the minimum circumscribed rectangle of the lesion area in the RGB image, the marking color value of the lesion area is used as the R channel value of the RGB of the pixel point, and 0 is used as the G channel value and B channel value of the pixel point, thereby obtaining the marked lesion esophageal image.

[0105] This embodiment adopts Model to present inverse proportional relationship and normalization processing, As the input of the model, the implementer can set the inverse proportional function and normalization function according to the actual situation.

[0106] So far, the present invention is completed.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent analysis system for internal medicine medical imaging data, characterized in that: The system includes the following modules: Medical image acquisition module: used to sequentially acquire a plurality of esophageal images of the patient to be detected, wherein the plurality of esophageal images are esophageal images of the patient at multiple times after taking contrast agent; the esophageal images include the esophageal area; Suspected lesion differentiation module: used to obtain suspected lesion areas and suspected normal areas in the esophageal image based on the width of the esophageal area in the esophageal image and the distribution of the esophageal area between different esophageal images; The reference area screening module is used to obtain the abnormal features of the suspected lesion area and the normal features of the suspected normal area according to the grayscale distribution features and texture features in the suspected lesion area and the suspected normal area, and obtain the lesion possibility of the suspected lesion area according to the abnormal features and the normal features in combination with the morphological conditions of the suspected lesion area; according to the lesion possibility of the suspected lesion area, a number of reference areas are screened; wherein the method for obtaining the reference area is: the suspected lesion area whose lesion possibility is greater than a preset possibility threshold F is recorded as the reference area; Lesion analysis module: used to obtain the abnormality of the reference area according to the difference in the possibility of lesions between the reference areas in the esophageal image and the normal characteristics of the suspected normal area; to screen the lesion area according to the abnormality of the reference area; to obtain the lesion degree of the lesion area according to the distribution of the lesion area in the esophageal image and the abnormality of the lesion area; Lesion area marking module: used to mark the lesion area in the esophagus image according to the morphological distribution of the lesion area in the esophagus image and the lesion degree of the lesion area, so as to obtain a marked lesion esophagus image; The method for obtaining the lesion possibility of a suspected lesion area includes: for any suspected lesion area, one side of the suspected lesion area is selected as the target side, and the lesion bending vector and the normal bending vector of the suspected lesion area are obtained according to the position distribution of the pixel points at the junction of the suspected normal area adjacent to the suspected lesion area in the target side direction; according to the difference between the lesion bending vector and the normal bending vector of the suspected lesion area, the abnormal characteristics, the difference in esophageal width, and the normal characteristics of the suspected normal area adjacent to the suspected lesion area, the initial lesion possibility of the target side of the suspected lesion area is obtained; the average of the initial lesion possibilities on both sides of the suspected lesion area is recorded as the lesion possibility of the suspected lesion area; Among them, the method for obtaining the lesion bending vector and the normal bending vector includes: for adjacent suspected lesion areas and suspected normal areas, obtaining the neighborhood range of the boundary edge pixel points between the two areas; the edges corresponding to the target side of the neighborhood range are the suspected lesion area sub-edges and the suspected normal area sub-edges, and within the neighborhood range, the sum vector of the vectors between the two adjacent edge pixel points of the suspected lesion area sub-edge from the boundary pixel point along the direction away from the suspected normal area is used as the normal bending vector, and within the neighborhood range, the sum vector of the vectors between the two adjacent edge pixel points of the suspected normal area sub-edge from the boundary pixel point along the direction away from the suspected lesion area is used as the lesion bending vector.

2. According to claim 1, an intelligent analysis system for internal medicine medical imaging data is characterized in that: The suspected lesion area and the suspected normal area in the esophageal image are obtained based on the width of the esophageal area in the esophageal image and the distribution of the esophageal area between different esophageal images, including: The average esophageal width of all esophageal regions in the esophageal image is obtained, and the esophageal region with an esophageal width smaller than the average esophageal width is recorded as a characteristic stenosis region; According to the position distribution of characteristic narrow areas between esophageal images, several suspected lesion areas in each esophageal image are obtained; Each esophageal region except the independent stenosis region in the target esophageal image is recorded as an independent non-stenosis region; a region consisting of several rows of continuous independent non-stenosis regions along the esophageal region distribution direction in the target esophageal image is recorded as a suspected normal region; The method of obtaining several suspected lesion areas in each esophageal image according to the position distribution of characteristic stenosis areas between esophageal images includes: recording any esophageal image as a target esophageal image, marking all esophageal areas corresponding to all characteristic stenosis areas in all esophageal images except the target esophageal image as related stenosis areas in the target esophageal image; collectively referring to all characteristic stenosis areas and all related stenosis areas in the target esophageal image as independent stenosis areas; and recording an area consisting of independent stenosis areas whose positions in several rows in the target esophageal image are continuous along the esophageal area distribution direction as a suspected lesion area.

3. According to claim 1, the intelligent analysis system for internal medicine medical imaging data is characterized in that: The method for acquiring the abnormal features of the suspected lesion area includes: According to the edge smoothness of the suspected lesion area and the disorder of the grayscale texture in the suspected lesion area, the abnormal characteristics of the suspected lesion area are obtained.

4. According to claim 1, the intelligent analysis system for internal medicine imaging data is characterized in that: The method for acquiring the normal features of the suspected normal area includes: According to the difference in esophageal width between esophageal regions in the suspected normal region, the edge smoothness of the suspected normal region, and the grayscale uniformity of the suspected normal region, the normal features of the suspected normal region are obtained.

5. According to claim 1, the intelligent analysis system for internal medicine imaging data is characterized in that: The method for obtaining the abnormality degree of the reference area includes: According to the difference in the possibility of lesions between the reference areas and the normal characteristics of the suspected normal areas adjacent to the reference areas, the abnormality index of the reference areas is obtained; According to the difference of normal features of adjacent suspected normal areas of the reference area at the same position between the esophageal images, an adjacent change index of the reference area is obtained; The abnormality degree of the reference area is obtained according to the difference of abnormality indicators of the reference area at the same position between the esophageal images and the adjacent change indicators of the reference area.

6. According to claim 1, the intelligent analysis system for internal medicine imaging data is characterized in that: The method for obtaining the lesion degree of the lesion area includes: For any esophageal image, the ratio of the lesion area to the esophageal length of the esophageal image where the lesion area is located is recorded as the length lesion value of the lesion area; The ratio of the average value of the esophageal width of all esophageal regions in the lesion region to the average value of the esophageal width of all esophageal regions in the esophageal image where the lesion region is located is recorded as the esophageal lesion value of the lesion region; The normalized result of the product of the abnormality degree of the lesion area, the length lesion value and the esophageal lesion value is recorded as the lesion degree of the lesion area.

7. According to claim 1, the intelligent analysis system for internal medicine imaging data is characterized in that: The method for acquiring the esophageal image with marked lesions comprises: For any lesion area, the marking color value of the lesion area is obtained according to the lesion degree of the lesion area; the minimum circumscribed rectangle of the lesion area is obtained, and the pixel points on the minimum circumscribed rectangle of the lesion area are marked using the marking color value of the lesion area to obtain a marked lesion esophageal image.