Endoscope image processing system and method
Through the endoscopic image processing system, personalized processing is carried out based on the basic information and physiological information of the target personnel, highlighting key diagnostic areas, solving the problem of insufficient image clarity and contrast in the prior art, and improving the accuracy and efficiency of medical diagnosis .
Patent Information
- Application Number
- CN202411362696.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-27
- Publication Date
- 2024-12-27
AI Technical Summary
The existing endoscopic image processing technology has problems such as insufficient image clarity, insufficient color restoration and poor contrast, which makes it difficult for doctors to clearly observe the details of the lesions during the diagnosis process, and lacks personalized processing, which affects the recognition of the lesions. efficiency.
An endoscopic image processing system was designed, through the image processing module, personalized processing based on the target personnel's basic information and physiological information, highlighting key diagnostic areas, and combining image fusion and recognition modules to achieve high-precision image display and lesion recognition. .
It significantly improves the accuracy and efficiency of medical diagnosis, provides an intuitive and meticulous foundation for lesion observation, realizes personalized treatment for target personnel, and improves the efficiency of doctors to identify lesions and the accuracy of diagnosis.
Smart Images
Figure CN119207729A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the technical field of endoscopic image processing, and in particular to an endoscopic image processing system and method. Background Art
[0002] With the continuous advancement of medical technology, the requirements for endoscopic image quality are increasing. Traditional endoscopic imaging often has problems such as insufficient image clarity, inaccurate color reproduction, and poor contrast. These defects make it difficult for doctors to clearly observe the details of the lesions during the diagnosis process, which may cause misdiagnosis or missed diagnosis. In addition, different patients have different areas of focus. The current endoscopic image processing technology is relatively popular and lacks personalized processing. Therefore, the existing endoscopic image processing technology lacks personalized processing, which affects the efficiency of identifying lesions. Summary of the invention
[0003] The embodiments of the present disclosure provide an endoscopic image processing system and method to solve the problem of lack of personalized processing.
[0004] In a first aspect, an embodiment of the present disclosure provides an endoscopic image processing system, comprising: An endoscope, used for acquiring original images, wherein the original images include an original endoscope image of a target person; An image processing module is connected to the endoscope and is used to process the original image according to target information to obtain a target image. The target information includes basic information and physiological information of the target person. The image display module is connected to the image processing module and is used to display the target image.
[0005] In an exemplary embodiment of the present disclosure, the image processing module includes a basic processing module, a special processing module and a target determination module; Both the basic processing module and the special processing module are connected to the endoscope; A basic processing module, used for processing the original image according to the first information to obtain the first image, wherein the first information includes the basic information of the target person; A special processing module, used for processing the original image according to the second information to obtain the second image, wherein the second information includes the physiological information of the target person; The target determination module is connected to the basic processing module and the special processing module respectively, and is used to determine the target image based on the first image and the second image.
[0006] In an exemplary embodiment of the present disclosure, the first information includes a plurality of basic information; The basic processing module is specifically used for: Determine a first area of the target person based on multiple basic information, where the first area is a first focus area of the endoscope image of the target person; The first region is processed to obtain a first image.
[0007] In an exemplary embodiment of the present disclosure, the second information includes a plurality of physiological information; Special processing modules are specifically used for: Determine a second area of the target person based on multiple physiological information, where the second area is a second focus area of the endoscopic image of the target person; The second region is processed to obtain a second image.
[0008] In an exemplary embodiment of the present disclosure, the target determination module further includes an image fusion unit; The image fusion unit is used to fuse the first image and the second image.
[0009] In an exemplary embodiment of the present disclosure, the target determination module further includes an image selection unit; The image selection unit is used to select the first image or the second image as the target image according to the judgment rule.
[0010] In an exemplary embodiment of the present disclosure, an endoscope image processing system further includes an image recognition module; The image recognition module is connected to the image processing module and the image display module respectively, and is used to recognize the target image and determine whether the target person has a lesion.
[0011] In an exemplary embodiment of the present disclosure, an endoscope image processing system further includes a comparison database, the comparison database including a first normal image set and a first abnormal image set; Recognize the target image, including: determining a first feature of the target image; determining a second feature of the first normal image set, and determining a third feature of the first abnormal image set; Calculate the similarity between the first feature and the second feature to obtain a first similarity; The similarity between the first feature and the third feature is calculated to obtain a second similarity.
[0012] In an exemplary embodiment of the present disclosure, determining whether a target person has a lesion includes: If the first similarity is greater than or equal to the first threshold, and the second similarity is less than the second threshold, the target person does not have a lesion; If the first similarity is less than the first threshold and the second similarity is greater than or equal to the second threshold, the target person has a lesion.
[0013] In a second aspect, the present disclosure provides an endoscopic image processing method, comprising: Acquire an original image, wherein the original image includes an endoscope original image of the target person; Processing the original image according to the target information to obtain the target image, the target information includes the basic information and physiological information of the target person; Display the target image.
[0014] The beneficial effects of an endoscopic image processing system and method provided by the embodiments of the present disclosure are: The disclosed embodiments greatly improve the accuracy and efficiency of medical diagnosis. The disclosed embodiments capture clear original images through a high-precision endoscope, providing doctors with an intuitive and detailed basis for lesion observation. Combined with the image processing module, the image quality is intelligently optimized according to the specific basic information and physiological information of the target person, and the key diagnostic areas are highlighted to achieve personalized processing. The image display module presents the processed target image in an intuitive manner, which is convenient for doctors to make diagnostic decisions quickly and accurately. Therefore, the disclosed embodiments achieve personalized processing for target personnel and improve the efficiency of doctors in identifying lesions. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, 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 disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0016] Figure 1 is a structural schematic diagram of an endoscope image processing system provided by an embodiment of the present disclosure; Figure 2 is a structural schematic diagram of another endoscope image processing system provided by an embodiment of the present disclosure; Figure 3 is a structural schematic diagram of another endoscopic image processing system provided by an embodiment of the present disclosure; Figure 4 It is a flowchart of an endoscopic image processing method provided in an embodiment of the present disclosure. DETAILED DESCRIPTION
[0017] In order to enable people in the technical field to better understand the present solution, the technical solution in the embodiment of the present solution will be clearly described below in conjunction with the drawings in the embodiment of the present solution. Obviously, the described embodiment is an embodiment of a part of the present solution, not all of the embodiments. Based on the embodiments in the present solution, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present solution.
[0018] The term "including" and any other variations in the specification and claims of this solution and the above drawings mean "including but not limited to", and is intended to cover non-exclusive inclusions and is not limited to the examples listed in the text. In addition, the terms "first" and "second" are used to distinguish different objects, not to describe a specific order.
[0019] The following is a detailed description of the implementation of the present disclosure in conjunction with the specific drawings: Figure 1 This is a schematic diagram of the structure of an endoscope image processing system provided by an embodiment of the present disclosure. Figure 1 , the endoscope image processing system comprises: An endoscope, used for acquiring original images, wherein the original images include an original endoscope image of a target person; An image processing module is connected to the endoscope and is used to process the original image according to target information to obtain a target image. The target information includes basic information and physiological information of the target person. The image display module is connected to the image processing module and is used to display the target image.
[0020] In this embodiment, the endoscope, as a front-end image acquisition module of the endoscope image processing system, can penetrate into the human body (such as the stomach, respiratory tract, joint cavity, etc.) and directly transmit optical images to obtain original images. For example, in a gastrointestinal endoscopic examination, the endoscope can clearly present details such as the state of the mucosa inside the stomach and intestines and whether there are lesions.
[0021] Among them, the original image includes the original endoscope image of the target person, which can reflect the real situation of the specific part of the target person. The original image can contain various complex information, such as the color, texture, shape and possible abnormal structure of the tissue. Different inspection parts and inspection purposes will lead to different characteristics of the original image. For example, when performing respiratory endoscopy, the original image can highlight the patency of the airway, the inflammation of the mucosa and other related features.
[0022] The image processing module is generally connected to the endoscope through a data transmission line (such as a high-definition data line, a fiber optic transmission line, etc.) to ensure that the original image data collected by the endoscope can be received quickly and stably. The above connection must ensure the real-time and accuracy of data transmission to avoid loss or distortion of the original image data.
[0023] The basic information of the target person may include age, gender, height and weight. The original image may be processed according to the basic information of the target person.
[0024] When the age is greater than or equal to the first value, the focus is on processing area A, which may be an area with age-related diseases; when the age is less than the first value, the focus is on processing area B, which may be an area related to development. The first value may be dynamically adjusted according to actual conditions.
[0025] When the gender is the first gender or the second gender, the focus is on processing the C region, and the C region may be a region with different physiological structures.
[0026] When the height is greater than or equal to the second value, the processing is focused on area D; when the height is less than the second value, the processing is focused on area E. The second value can be dynamically adjusted according to actual conditions.
[0027] When the body weight is greater than or equal to the third value, the treatment is focused on the F region, which may be a fat accumulation region; when the body weight is less than the third value, the treatment is focused on the G region. The third value may be dynamically adjusted according to actual conditions.
[0028] For example, for target persons of different ages, the degree of development and physiological characteristics of human organs are different, and the image processing module can adaptively adjust the original image according to the age information. For the original image of children, more attention can be paid to the characteristics related to growth and development, while for adults, more emphasis can be placed on detecting common disease characteristics related to age. According to gender information, some differences in physiological structures can also be reflected in image processing, such as the pelvic structure of men and women requires different image emphases in some endoscopic examinations.
[0029] The target person's physiological information may include blood pressure, heart rate and body temperature, which are closely related to the health status of the human body.
[0030] If the blood pressure value is greater than the first value or less than the second value, the H area is processed intensively; If the heart rate is greater than the third value or less than the fourth value, the J region is processed emphatically, and the J region may be a region related to the heart; If the body temperature is greater than the fifth value or less than the sixth value, the K area is treated with emphasis.
[0031] Among them, the A region, the B region, the C region, the D region, the E region, the F region, the G region, the H region, the I region, the J region and the K region may be intersection regions.
[0032] For example, if the target person has an abnormal heart rate, the image processing module can focus on analyzing the original image related to the heart. When the target person's body temperature rises, it indicates that there is an inflammatory response in the body. The image processing module can enhance the image of the parts where inflammation may occur (such as the respiratory tract, digestive tract, etc.) to more clearly observe the scope and degree of inflammation. By comprehensively considering the above physiological information, the image processing module can process the original image in a targeted manner, highlight the image area related to abnormal physiological information, and improve the detection rate of lesions.
[0033] The processed target image reflects the key information related to the health status of the target person more clearly and intuitively. The target image can help doctors make more accurate diagnoses and reduce the risk of misdiagnosis caused by blurred images or unclear information.
[0034] The image display module is connected to the image processing module to ensure that the processed target image data can be received with high quality and converted into a visual target image for display. Doctors can directly observe the target image on the image display module and understand the situation inside the target user's body in real time. For example, during the diagnosis process, the image display module can simultaneously display multiple target images (such as images from different angles and at different time points), which is convenient for doctors to conduct comparative analysis and make more accurate diagnoses.
[0035] It can be concluded from the above that this embodiment greatly improves the accuracy and efficiency of medical diagnosis. This embodiment captures clear original images through a high-precision endoscope, providing doctors with an intuitive and detailed basis for lesion observation. Combined with the image processing module, the image quality is intelligently optimized according to the specific basic information and physiological information of the target person, the key diagnostic area is highlighted, and personalized processing is achieved. The image display module presents the processed target image in an intuitive manner, which is convenient for doctors to make diagnostic decisions quickly and accurately. Therefore, this embodiment realizes personalized processing for target personnel and improves the efficiency of doctors in identifying lesions.
[0036] In one embodiment of the present disclosure, reference Figure 2 ,The image processing module includes a basic processing module, a special processing module and a target determination module; Both the basic processing module and the special processing module are connected to the endoscope; A basic processing module, used for processing the original image according to the first information to obtain the first image, wherein the first information includes the basic information of the target person; A special processing module, used for processing the original image according to the second information to obtain the second image, wherein the second information includes the physiological information of the target person; The target determination module is connected to the basic processing module and the special processing module respectively, and is used to determine the target image based on the first image and the second image.
[0037] In this embodiment, the basic processing module can receive the original image of the endoscope.
[0038] The first information contains the basic information of the target person (such as age, gender, height, weight). When processing the original image, the above basic information plays an important reference role.
[0039] If the basic information is based on age, the first adjustment strategy is selected to focus on image processing of area A or area B; If the basic information is based on gender, the second adjustment strategy is selected to focus on image processing of area C; If the basic information is based on height, the third adjustment strategy is selected to focus on image processing of area D or area E; If the basic information is based on body weight, the fourth adjustment strategy is selected to focus on image processing of the F area or the G area; The first adjustment strategy, the second adjustment strategy, the third adjustment strategy and the fourth adjustment strategy may be the same or different. The first adjustment strategy may be scaling down or up; the second adjustment strategy may be contrast adjustment; the third adjustment strategy may be geometric transformation; and the fourth adjustment strategy may be clarity enhancement.
[0040] For example, for target persons of different age groups, the size, proportion and degree of development of human organs are different. The basic processing module will adjust the scaling ratio of the original image according to the age information. For children, the original image can be enlarged to show the organ details more clearly. According to the gender information, since there are differences in certain physiological structures between men and women (such as pelvic structure), the basic processing module can perform different degrees of sharpening or contrast adjustment on the image of the corresponding part to highlight the gender-related features. The height and weight information can assist in determining the relative position and size of the organs and perform corresponding geometric transformations on the original image.
[0041] A special processing module can obtain the original image of the endoscope.
[0042] The second information includes the target person's physiological information (such as blood pressure, heart rate and body temperature). The above physiological information is of special significance to image processing.
[0043] When the physiological information is based on blood pressure, the fifth adjustment strategy is selected to focus on image processing of the H area; When the physiological information is based on the heart rate, the sixth adjustment strategy is selected to focus on image processing of the J area; When the physiological information is based on body temperature, the seventh adjustment strategy is selected to focus on image processing of the K area; The fifth adjustment strategy, the sixth adjustment strategy and the seventh adjustment strategy may be the same or different. The fifth adjustment strategy may be filtering; the sixth adjustment strategy may be improving color saturation or brightness; and the seventh adjustment strategy may be enhancing contrast.
[0044] For example, when the target person has an abnormal heart rate, the special processing module can focus on analyzing the original image related to the heart, and highlight the blood flow of the heart tissue and changes in the heart structure by changing the color saturation, brightness and other parameters of the image; when the body temperature rises, it indicates that there is inflammation in the body. The special processing module can enhance the original image of the part where inflammation may occur (such as the respiratory tract, digestive tract, etc.), such as increasing the contrast of the image, using a filtering algorithm to highlight the edge of the inflammation area, etc., so as to more clearly observe the scope and degree of inflammation. Based on the processing of the above physiological information, the second image obtained can more specifically reflect the image features related to the current physiological condition of the target person.
[0045] The target determination module is connected to the basic processing module and the special processing module respectively, so that the module can quickly obtain data of the first image and the second image. The target determination module performs fusion processing on the first image and the second image or selects based on a standard to determine the target image.
[0046] It can be concluded from the above that the subdivision design of the image processing module significantly enhances the performance of the endoscope image processing system. The basic processing module optimizes the image according to the basic information of the patient, and then the special processing module further refines the image in combination with the physiological information, providing personalized processing, ensuring that the target image not only conforms to the individual characteristics but also highlights the pathological details. The target determination module combines the advantages of both, accurately locks the target image, provides doctors with a more accurate and comprehensive basis for diagnosis, and greatly improves the accuracy and efficiency of diagnosis.
[0047] In one embodiment of the present disclosure, reference Figure 2 , the first information includes multiple basic information; The basic processing module is specifically used for: Determine a first area of the target person based on multiple basic information, where the first area is a first focus area of the endoscope image of the target person; The first region is processed to obtain a first image.
[0048] In this embodiment, the first information includes a plurality of basic information, and a first region of the target person can be determined based on the plurality of basic information. The first region is a first focus region of the endoscopic image of the target person.
[0049] There is a mapping relationship between each basic information and the first area of the target person; The sample data of the historical target personnel is mapped with the corresponding basic information, and different mapping relationships correspond to a first key focus area.
[0050] Specifically, the first area can be determined based on factors related to age, gender, height and weight.
[0051] First, people of different ages have different degrees of organ development. For example, for children, who are in the growth and development stage, their bones have not yet been fully calcified. In bone-related endoscopic images, the epiphysis is the first area, because this area is crucial for the growth and development of children's bones. For the elderly, due to organ aging and possible degenerative lesions, such as in gastrointestinal endoscopy, areas with thinning intestinal mucosa and shallower vascular texture will become key areas of concern.
[0052] Second, there are differences in physiological structure between men and women. Taking the urinary system as an example, the prostate of men is prone to hyperplasia after middle age, so in the corresponding target image, the prostate area can be determined as the first area. The uterus and appendages of women are more prone to gynecological diseases during their childbearing years, so in gynecological endoscopy, the uterus and its appendages will become the focus of attention.
[0053] Third, the distribution space of organs in tall people is relatively large. For people with heavy weight, the distribution of body fat will affect the position and display effect of organs. For example, in laparoscopic examination, if the target person is overweight, the liver and other organs will be wrapped in fat. Then, in the target image, the area with more fat accumulation that affects the normal morphology and function of the liver will be determined as the first area.
[0054] Specifically, the first region may be processed to obtain the first image as follows: First, image enhancement processing. After determining the first area, image enhancement processing will be performed to make the features of the area more obvious. For example, for the epiphyseal area of children mentioned above, the contrast and brightness of the image in this area can be improved to make the details of the epiphysis clearer, which helps doctors observe whether the development of the epiphysis is normal. For areas where the intestinal mucosa of the elderly is thinning, the display of vascular texture can be highlighted by enhancing color saturation and other methods, so as to detect vascular lesions in time.
[0055] Second, image magnification processing. For some small areas of focus, such as small nodules in the prostate or small cysts in the uterine appendages, the basic processing module will magnify the first area to a level where the details of the lesion can be clearly observed, thereby obtaining a first image that is more conducive to diagnosis.
[0056] It can be concluded from the above that this embodiment significantly improves the accuracy and efficiency of diagnosis and realizes personalized processing. This embodiment achieves efficient positioning of lesions or key parts by accurately identifying the first area of the target person as the focus of attention, reducing the time for doctors to review non-critical areas. At the same time, the first area is specially processed to optimize the quality of the generated first image, making the details clearer, which helps doctors to more accurately judge the condition and formulate treatment plans in a timely manner.
[0057] In one embodiment of the present disclosure, reference Figure 2 , the second information includes a plurality of physiological information; Special processing modules are specifically used for: Determine a second area of the target person based on multiple physiological information, where the second area is a second focus area of the endoscopic image of the target person; The second region is processed to obtain a second image.
[0058] In this embodiment, the second information includes a plurality of physiological information, and the second area can be determined based on the plurality of physiological information, and the second area is a second focus area of the endoscopic image of the target person.
[0059] There is a mapping relationship between each physiological information and the second area of the target person; The sample data of the historical target person is mapped with the corresponding physiological information, and different mapping relationships correspond to a second key focus area.
[0060] Specifically, the plurality of physiological information may include blood pressure, heart rate, and body temperature.
[0061] First, blood pressure information.
[0062] If the target person's blood pressure is higher than the normal range for a long time, during the cardiac endoscopy, the left ventricular wall area of the heart will be determined as the second area because high blood pressure will affect the heart, such as left ventricular hypertrophy. Because long-term high blood pressure will increase the afterload of the heart, the left ventricle needs to overcome greater resistance to eject blood, which will lead to thickening of the left ventricular wall.
[0063] During renal endoscopy, hypertension can lead to lesions in the renal arterioles, which in turn affects the blood circulation of the kidneys. At this time, the afferent arterioles and glomerular areas of the kidneys will be identified as the second area of focus, because these areas are essential for maintaining the filtration function of the kidneys.
[0064] Second, heart rate information.
[0065] When the target person's heart rate is higher or lower than the normal range, during the cardiac endoscopy, the sinoatrial node area and the area where the cardiac conduction system (such as the atrioventricular node, His bundle, etc.) is located will be determined as the second area. This is because abnormal heart rate is often related to abnormal function of the heart's pacing and conduction system. The sinoatrial node is the normal pacemaker of the heart, and lesions in the conduction system can cause changes in heart rate.
[0066] Third, body temperature information.
[0067] When the target person's body temperature is higher than normal, it indicates infection or inflammation. If it is a respiratory infection, during bronchoscopy, the area of bronchial mucosal congestion and edema will be identified as the second area. This is because the inflammatory response usually causes vasodilation and increased permeability of local tissues, which in turn causes congestion and edema of the mucosa.
[0068] When a patient has an unexplained low fever and is suspected of having an intra-abdominal infection, during laparoscopic examination, organ areas prone to infection, such as the appendix and gallbladder, will be identified as the second area. This is because the above organs will have an inflammatory response when infected, manifested as local redness, swelling, exudation, etc., and a rise in body temperature is an external manifestation of the inflammatory response.
[0069] Specifically, the second region is processed to obtain the second image in the following manner: First, image color adjustment.
[0070] For the left ventricular wall area (second area) determined by abnormal blood pressure, the image color of this area can be adjusted to warm tones (such as red, orange) to highlight the degree of myocardial hypertrophy. By increasing the saturation of red, the myocardial tissue can be made more eye-catching in the image, making it easier to observe the morphology and arrangement of myocardial cells.
[0071] For the area of respiratory bronchial mucosal congestion caused by increased body temperature (the second area), the contrast between red and blue can be enhanced to more clearly distinguish the congested blood vessels (in red) from the surrounding tissues (in blue), thereby highlighting the scope and severity of the inflammatory area.
[0072] Second, image clarity is enhanced.
[0073] For the cardiac conduction system region (second region) determined to have abnormal heart rate, the resolution and sharpness of the image are improved to make the subtle structures such as the sinoatrial node and the atrioventricular node more clearly identifiable. In this embodiment, a filtering algorithm can be used to remove noise in the image and enhance the signal strength, so that the image of the key area is clearer.
[0074] From the above, it can be concluded that this embodiment combines multiple physiological information to determine the second focus area of the endoscopic image, which significantly enhances the personalization and accuracy of the diagnosis. By comprehensively considering the physiological state of the target person, the special processing module can accurately lock the second area that is highly related to the disease, making the diagnosis more focused. Further detailed processing of this area, the second image obtained not only retains key information, but also improves the readability of the image, providing doctors with a more comprehensive and accurate basis for diagnosis, and promoting the scientificity and effectiveness of medical decision-making.
[0075] In one embodiment of the present disclosure, reference Figure 3 ,The target determination module also includes an image fusion unit; The image fusion unit is used to fuse the first image and the second image.
[0076] In this embodiment, the method for fusing the first image and the second image may include: pixel-based fusion and feature-based fusion.
[0077] Specifically, pixel-based fusion is as follows: First, the simple weighted average method. For each corresponding pixel in the first image and the second image, they are added according to a certain weight. At this time, the first image corresponds to the first weight; the second image corresponds to the second weight. If the second image needs to be displayed more prominently, the second weight can be set to be greater than the first weight.
[0078] Second, principal component analysis (PCA). The first image and the second image are regarded as a two-dimensional data matrix and converted into a new feature space through PCA transformation. In the new feature space, the main eigenvectors are selected to reconstruct the image and achieve fusion. The above method can extract the main information in the image, reduce redundancy, and highlight the key features of the image.
[0079] Specifically, the feature-based fusion is as follows: First, feature extraction. This embodiment extracts key features from the first image and the second image, such as organ contours, geometric shapes and other features determined based on the basic information of the target person in the first image, and texture features of lesion areas and abnormal tissues determined based on physiological information in the second image. For example, in a lung endoscopy image, the first image can highlight the overall contour of the lungs and the general structure of the bronchi (basic information such as age, height, and weight), and the second image can focus on local shadows and texture changes caused by inflammation or other physiological abnormalities (physiological information such as body temperature).
[0080] Second, feature matching and fusion. This embodiment matches the extracted features and then fuses the matched features. For example, for the same lesion area, only a fuzzy contour change is shown in the first image, while the texture and color changes of the lesion area can be clearly seen in the second image. Through feature fusion, the two aspects of information can be combined to obtain a more comprehensive and accurate image of the lesion area.
[0081] It can be concluded from the above that this embodiment can provide more comprehensive information. The fused target image can comprehensively reflect the basic information and physiological information of the target person. Through fusion, the doctor can simultaneously observe the structural characteristics of the heart and information related to the physiological function state in one image, so as to have a more comprehensive understanding of the health status of the heart. This embodiment can also improve the accuracy of diagnosis. Since multiple aspects of information are fused, the limitations of single image information are reduced. After the image is fused, the doctor can more accurately judge the location, range and relationship of the lesion with the surrounding normal tissue, avoiding misdiagnosis or missed diagnosis due to insufficient information.
[0082] In one embodiment of the present disclosure, reference Figure 3 ,The target determination module also includes an image selection unit; The image selection unit is used to select the first image or the second image as the target image according to the judgment rule.
[0083] In this embodiment, the judgment rule can be selected according to the reliability of the information, the type or severity of the lesion.
[0084] First, based on the reliability of information.
[0085] If the basic information of the target person is very accurate and highly stable, while the physiological information is easily affected by some temporary factors (such as emotional fluctuations before the examination, having just taken certain medications, etc.) and is not very reliable, then the image selection unit is more inclined to select the first image obtained based on the basic information processing as the target image.
[0086] For example, in an endoscopic examination of a healthy young person who has no obvious symptoms of illness but is just undergoing a routine physical examination, basic information (such as age, gender, height, and weight) can reflect the characteristics of his normal physiological structure, while the physiological information at this time (such as heart rate, blood pressure, etc.) fluctuates slightly due to the tension during the physical examination. In this case, the first image can better represent his true internal body condition.
[0087] On the contrary, if the physiological information has been measured multiple times and verified to be accurate, but the basic information may be somewhat vague (such as the target person cannot remember some of his or her basic information clearly), then the second image obtained based on the processing of the physiological information will be preferred.
[0088] For example, when performing an endoscopic examination on an amnesia patient, basic information such as age and weight is difficult to obtain accurately, but reliable physiological information such as blood pressure and blood biochemical indicators can be obtained through detailed physiological examinations. At this time, the second image is more valuable for diagnosis.
[0089] Second, based on the type and severity of the lesion.
[0090] When the lesion type is mainly related to abnormal changes in physiological functions, the image selection unit will give priority to the second image. For example, in the endoscopic examination of the cardiovascular system, if the main problem of the target person is vascular endothelial damage and atherosclerosis caused by long-term unstable blood pressure, then the second image obtained based on physiological information such as blood pressure can more clearly show the characteristics of the lesion area, such as thickening of the vascular wall, formation of atherosclerotic plaques, etc., which is more helpful for doctors to evaluate and diagnose the lesions.
[0091] However, if the lesion is mainly related to the basic structural characteristics of the body, such as congenital organ malformation, in this case the first image can better reflect the congenital structural abnormality of the organ, and the image selection unit will select the first image as the target image.
[0092] The image selection unit can realize automatic selection through preset algorithms and programs, such as deep learning algorithms. Designed according to the above judgment rules, after inputting the basic information and physiological information of the target person, the algorithm will automatically analyze the reliability of this information, the type of lesions and other factors, and then automatically make a decision on whether to select the first image or the second image.
[0093] It can be concluded from the above that this embodiment, by introducing the image selection unit, realizes the function of intelligently selecting the first image or the second image as the target image based on the judgment rule. This embodiment helps doctors improve the accuracy and efficiency of diagnosis and provide patients with more accurate and personalized medical services.
[0094] In one embodiment of the present disclosure, reference Figure 3 ,An endoscopic image processing system also includes an image recognition module; The image recognition module is connected to the image processing module and the image display module respectively, and is used to recognize the target image and determine whether the target person has a lesion.
[0095] In this embodiment, the image recognition module can obtain a processed target image containing rich information. For example, the image processing module optimizes the original image based on the basic information and physiological information of the target person, highlights the key areas of concern, and the processed target image can provide more valuable data for the image recognition module, enabling it to perform lesion identification more efficiently.
[0096] The process of image recognition module judging lesions can be as follows: First, feature extraction.
[0097] Extract various lesion-related features from the target image. The above features include but are not limited to texture features of the image (such as the difference between the texture of lesion tissue and normal tissue. The texture of normal tissue is usually uniform, while the texture of lesion tissue is disordered), color features (such as some lesions can cause the color of tissue to become darker or lighter, such as inflammation can cause tissue to become red, and ischemia can cause tissue to become white, etc.), shape features (for example, tumors can present irregular shapes, while normal organs usually have relatively regular geometric shapes), etc.
[0098] Second, pattern recognition.
[0099] The extracted features are compared with known lesion patterns. A large number of lesion sample patterns are stored in the image recognition module, which are obtained by analyzing and summarizing a large number of historical case images. For example, when the texture feature of a certain area in the target image is extracted to be rough and irregular, the color feature is dark red, and the shape feature is circular with blurred boundaries, it is matched with the stored cancer lesion pattern. If the similarity is higher than the first similarity, it is determined that there is a cancer lesion in the area.
[0100] Use machine learning algorithms for identification. By training a large amount of image data with lesion information annotated, the machine learning model (such as the neural network model) in the image recognition module can learn the characteristic patterns of lesions and make lesion judgments on new target images based on these learned patterns. For example, a neural network model trained with a large amount of data can automatically identify lesion features in the target image that are similar to those in the training data and give a probability estimate of the lesion.
[0101] It can be concluded from the above that this embodiment realizes automatic recognition of target images and lesion judgment by introducing an image recognition module and closely cooperating with an image processing module and an image display module. Therefore, this embodiment not only significantly improves the intelligent level of diagnosis and provides doctors with timely and accurate diagnosis basis, but also enhances the objectivity of diagnosis, reduces the interference of human factors, and further improves the accuracy and reliability of medical diagnosis.
[0102] In one embodiment of the present disclosure, reference Figure 3 , an endoscope image processing system further includes a comparison database, the comparison database including a first normal image set and a first abnormal image set; Recognize the target image, including: determining a first feature of the target image; determining a second feature of the first normal image set, and determining a third feature of the first abnormal image set; Calculate the similarity between the first feature and the second feature to obtain a first similarity; The similarity between the first feature and the third feature is calculated to obtain a second similarity.
[0103] In this embodiment, the first feature of the target image may include a color feature, a texture feature, and a shape feature.
[0104] For example, in terms of color features, different lesions will cause the color of a specific area in the target image to change. For example, liver lesions will cause the liver area to appear dark yellow or grayish white in the target image, which is different from the normal liver. In terms of texture features, the texture of normal tissue often has a certain regularity and continuity, while the texture of diseased tissue will become disordered, rough or interrupted. For example, in a lung endoscopy image, the texture of normal alveolar tissue is clear and uniform, but when suffering from diseases such as pneumonia or pulmonary fibrosis, the texture of the alveolar area will become blurred and messy. In terms of shape features, the appearance of normal organs is usually regular and symmetrical, and lesions will cause the shape of the organs to change. For example, the presence of a tumor may cause the organ to bulge or deform locally.
[0105] The first normal image set is composed of endoscopic images of a large number of healthy individuals. For the second feature in this image set, the color feature reflects the typical color range of normal tissues, such as normal gastric mucosa usually appears uniform pink in endoscopic images. The texture feature reflects the texture pattern formed by the normal structural arrangement of tissues under normal physiological conditions. For example, the normal vascular texture is clear, smooth and regularly branched in the image. The shape feature is manifested as the standard geometric shape of each organ under normal physiological conditions, such as a normal heart valve has a specific and regular valve shape under endoscopy.
[0106] The first abnormal image set contains endoscopic images of various known lesion types. The third feature reflects the abnormal color changes caused by the lesion in terms of color features. For example, hemorrhagic lesions will make the corresponding area in the image appear dark red. In terms of texture features, since the lesions destroy the normal tissue structure, there will be thickening, distortion of the texture of the lesion area, or the emergence of new abnormal texture patterns. In terms of shape features, lesions may cause significant changes in the overall or local shape of the organ. For example, tumors may cause irregular lumps and protrusions on the organ.
[0107] By calculating the first similarity between the first feature and the second feature, the closeness of the target image to the normal image can be evaluated. If the first similarity is higher than the first threshold, it means that the tissue in the target image is similar to the normal tissue in terms of color, texture and shape, and the possibility of lesions is relatively low. On the contrary, calculating the second similarity between the first feature and the third feature can measure the similarity between the target image and the known abnormal image. If the second similarity is higher than the first threshold, then there is a lesion in the tissue in the target image, and the type of lesion can be inferred based on the similarity type with the abnormal image set.
[0108] When calculating similarity, a variety of methods can be used. For example, for color features, the distance measurement method of the color histogram, such as Bhattacharyya distance, histogram intersection distance, etc., can be used to measure the similarity of the color distribution between the target image and the normal / abnormal image. For texture features, the similarity can be calculated using methods based on texture statistics, such as the eigenvalues (contrast, correlation, energy, entropy, etc.) of the gray-level co-occurrence matrix. The above eigenvalues can reflect the uniformity, directionality and other characteristics of the texture. For shape features, shape descriptors, such as Fourier descriptors, moment invariants, etc., can be used to describe the geometric characteristics of the shape and calculate the similarity between the shapes. By combining the similarity calculations of the above different features, the similarity between the target image and the normal / abnormal image set can be evaluated more comprehensively and accurately.
[0109] From the above, it can be concluded that the comparison database and recognition process integrated in the endoscopic image processing system significantly improve the accuracy and efficiency of lesion detection. By comparing the feature similarity between the target image and the normal and abnormal image sets, the system can automatically distinguish the physiological state and detect potential lesions in time. The recognition method based on big data and feature matching reduces the subjectivity of human judgment and improves the objectivity and accuracy of diagnosis. At the same time, fast similarity calculation accelerates the diagnosis process.
[0110] In one embodiment of the present disclosure, reference Figure 3 , to determine whether the target person has lesions, including: If the first similarity is greater than or equal to the first threshold, and the second similarity is less than the second threshold, the target person does not have a lesion; If the first similarity is less than the first threshold and the second similarity is greater than or equal to the second threshold, the target person has a lesion.
[0111] In this embodiment, the first similarity is the similarity calculated between the first feature of the target image and the second feature of the first normal image set. When the first similarity is greater than or equal to the first threshold, it indicates that the target image is very similar to the image in the normal image set in terms of multiple features such as color, texture, and shape. For example, in a gastric endoscopy, if the color (uniform pink), texture (with regular wrinkled texture), and shape (regular overall contour of the stomach) of the gastric mucosa in the target image highly match the gastric image features in the normal image set and exceed the set first threshold, it indicates that the gastric tissue of the target person meets normal physiological characteristics in the above aspects.
[0112] The second similarity is the similarity calculated between the first feature of the target image and the third feature of the first abnormal image set. When the second similarity is less than the second threshold, it means that the target image has a large difference in features from the known abnormal image set. For example, in the judgment of lung endoscopy images, if the features of the lung tissue in the target image and the features of the lesion images such as pneumonia and pulmonary fibrosis contained in the first abnormal image set (such as color abnormality, texture disorder, shape change, etc.) are very low in similarity and lower than the second threshold, then from the perspective of comparing these features, it is unlikely that there is a lesion.
[0113] The first threshold and the second threshold change dynamically.
[0114] Combining the above two conditions, when the first similarity satisfies a value greater than or equal to the first threshold value, and the second similarity satisfies a value less than the second threshold value, it can be relatively reliably determined that the target person does not have a lesion. This is based on the comprehensive consideration that the target image has a high similarity with the normal image, but a low similarity with the abnormal image.
[0115] When the first similarity is less than the first threshold and the second similarity is greater than or equal to the second threshold, it can be determined that the target person has a lesion based on the feature comparison results with the normal image and the abnormal image. The above judgment rule is based on the situation that the target image has a large difference from the normal image and a large similarity with the abnormal image.
[0116] It can be concluded from the above that this embodiment determines whether the target person has a lesion by setting a clear similarity threshold. This strategy greatly enhances the diagnostic capability of the endoscopic image processing system and improves the sensitivity and specificity of the diagnosis.
[0117] An endoscope image processing system corresponding to the above embodiment, Figure 4The following is a flowchart of an endoscopic image processing method provided by an embodiment of the present disclosure. For ease of explanation, only the part related to the embodiment of the present disclosure is shown. Figure 4 , the endoscope image processing method comprises: Acquire an original image, wherein the original image includes an endoscope original image of the target person; Processing the original image according to the target information to obtain the target image, the target information includes the basic information and physiological information of the target person; Display the target image.
[0118] In this embodiment, the original image is obtained by an endoscope. An endoscope usually uses an optical system (such as a lens, an optical fiber, etc.) to penetrate into the cavity or organ inside the human body. At the target site, the light is reflected and transmitted to an image sensor (such as a CCD or CMOS sensor) through the optical system. For example, in a gastrointestinal endoscopy, the endoscope is inserted into the gastrointestinal tract, and the internal optical structure collects the light in the gastrointestinal tract and then converts it into an electrical signal to form an original image signal of the gastrointestinal tract.
[0119] The clarity and resolution of the original image are affected by factors such as the optical quality of the endoscope and the performance of the image sensor. Environmental factors can also affect the original image. For example, liquid, gas, and impurities in the gastrointestinal tract can interfere with the reflection and propagation of light, causing the image to be blurred, shadowed, or distorted.
[0120] Specifically, the process of processing the original image according to the target information to obtain the target image can be as follows: First, the target image is processed using basic information.
[0121] The basic information of the target person includes age, gender, height, and weight. The above information can provide important references in the image preprocessing stage. For example, for people of different ages, the degree of development and physiological characteristics of human organs are different. When processing images, different degrees of enhancement or adjustment can be made according to age. For target images of children, because their organs are small and immature, the images will be enlarged and some areas related to growth and development will be highlighted; for adults, according to the common disease risks corresponding to their age, the areas prone to lesions will be focused on and processed.
[0122] Gender differences also affect the processing of target images. For example, there are differences in the physiological structure of certain organs between men and women. When processing target images related to the reproductive system, targeted target image optimization can be performed on specific organ structures based on gender information, so that the target image can more clearly display the characteristic areas related to gender.
[0123] Second, the target image is processed using physiological information.
[0124] Physiological information such as blood pressure, heart rate, and body temperature can assist in the processing of target images. For example, if the target person's heart rate is abnormal, it indicates that there is a problem with the heart function. When processing endoscopic images of the heart, the image can be filtered to enhance the details of the heart tissue, so as to better observe whether there are lesions in the heart structure.
[0125] When the body temperature rises, it indicates that there is an inflammatory response in the body. When processing images of areas where inflammation may occur (such as the respiratory tract, digestive tract, etc.), the contrast of the image can be increased to make the inflamed area easier to identify.
[0126] Image display usually uses a high-resolution medical display, which can accurately present information such as image color, brightness, and contrast, providing doctors with a clear visual experience.
[0127] During the diagnosis process, clear target images can help doctors more accurately determine the location, size, shape and other characteristics of the lesion. For example, during a lung endoscopy, doctors can observe the color changes of lung tissue, the presence of nodules, the edge characteristics of nodules, etc. through the clear images on the monitor, thereby making an accurate diagnosis of lung diseases.
[0128] At the same time, displaying the target image can also realize multi-image comparative analysis. For example, target images acquired at different time points or different angles can be displayed simultaneously, making it easier for doctors to observe the development trend of the lesion or evaluate the lesion from different angles.
[0129] It can be concluded from the above that this embodiment greatly improves the accuracy and efficiency of medical diagnosis. This embodiment captures clear original images through a high-precision endoscope, providing doctors with an intuitive and detailed basis for lesion observation. According to the specific basic information and physiological information of the target person, the image quality is intelligently optimized, the key diagnostic area is highlighted, and personalized processing is achieved. Finally, this embodiment presents the processed target image in an intuitive manner, which is convenient for doctors to make diagnostic decisions quickly and accurately. Therefore, this embodiment realizes personalized processing for target personnel and improves the efficiency of doctors in identifying lesions.
[0130] The above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than to limit them. Although the present disclosure has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present disclosure.
Claims
1. An endoscope image processing system, characterized in that: include: An endoscope, used to obtain an original image, wherein the original image includes an endoscope original image of a target person; An image processing module, connected to the endoscope, for processing the original image according to target information to obtain a target image, wherein the target information includes basic information and physiological information of the target person; An image display module is connected to the image processing module and is used to display the target image.
2. An endoscopic image processing system as claimed in claim 1, characterized in that: The image processing module includes a basic processing module, a special processing module and a target determination module; The basic processing module and the special processing module are both connected to the endoscope; The basic processing module is used to process the original image according to the first information to obtain the first image, wherein the first information includes the basic information of the target person; The special processing module is used to process the original image according to the second information to obtain the second image, wherein the second information includes the physiological information of the target person; The target determination module is connected to the basic processing module and the special processing module respectively, and is used to determine the target image based on the first image and the second image.
3. An endoscopic image processing system as claimed in claim 2, characterized in that: The first information includes a plurality of basic information; The basic processing module is specifically used for: Determine a first area of the target person based on the multiple basic information, where the first area is a first focus area of the endoscopic image of the target person; The first region is processed to obtain a first image.
4. An endoscopic image processing system as claimed in claim 3, characterized in that: The second information includes a plurality of physiological information; The special processing module is specifically used for: Determine a second area of the target person based on the multiple physiological information, where the second area is a second focus area of the endoscopic image of the target person; The second area is processed to obtain a second image.
5. An endoscopic image processing system as claimed in claim 4, characterized in that: The target determination module also includes an image fusion unit; The image fusion unit is used to fuse the first image and the second image.
6. An endoscopic image processing system as claimed in claim 4, characterized in that: The target determination module also includes an image selection unit; The image selection unit is used to select the first image or the second image as the target image according to a judgment rule.
7. An endoscopic image processing system as claimed in claim 1, characterized in that: Also includes an image recognition module; The image recognition module is connected to the image processing module and the image display module respectively, and is used to recognize the target image and determine whether the target person has a lesion.
8. An endoscopic image processing system as claimed in claim 7, characterized in that: Also included is a comparison database, the comparison database including a first normal image set and a first abnormal image set; The identifying the target image includes: determining a first feature of the target image; determining a second feature of the first normal image set, and determining a third feature of the first abnormal image set; Calculating the similarity between the first feature and the second feature to obtain a first similarity; The similarity between the first feature and the third feature is calculated to obtain a second similarity.
9. An endoscopic image processing system as claimed in claim 8, characterized in that: The step of judging whether the target person has a lesion comprises: If the first similarity is greater than or equal to a first threshold, and the second similarity is less than a second threshold, the target person does not have a lesion; If the first similarity is less than a first threshold and the second similarity is greater than or equal to a second threshold, the target person has a lesion.
10. An endoscopic image processing method, characterized in that: include: Acquire an original image, wherein the original image includes an endoscopic original image of a target person; Processing the original image according to target information to obtain a target image, wherein the target information includes basic information and physiological information of the target person; The target image is displayed.