Colon image enhancement method and system based on deep learning
The deep learning-based colon image enhancement system solves the problems of poor image quality and difficulty in highlighting lesion features in colon image acquisition technology, and realizes intelligent enhancement of high-quality images and improvement of diagnostic accuracy.
Patent Information
- Application Number
- CN202511241058.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-02
- Publication Date
- 2025-11-28
AI Technical Summary
Existing colon imaging technologies suffer from poor image quality, and traditional methods struggle to effectively highlight lesion features, impacting the accuracy of doctors' diagnoses.
A deep learning-based colon image enhancement system is adopted, including an acquisition module, an enhancement execution module, and a feedback adjustment module. Through preprocessing, feature extraction by a deep learning model, and adjustment of enhancement coefficients, the image quality and feature prominence are improved.
It improves the quality and diagnostic accuracy of colon images, enhances the intelligence and automation of images, makes them more adaptable, better highlights the characteristics of lesion areas, and improves the efficiency of medical diagnosis.
Smart Images
Figure CN121032835A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of image enhancement, in particular to a colon image enhancement method and system based on deep learning. BACKGROUND
[0002] In the process of medical diagnosis, the quality of colon images plays a crucial role in accurately detecting lesions and judging the severity of the disease. However, existing colon image acquisition techniques have many problems, resulting in poor image quality. For example, during the colon image acquisition process, due to the complex environment in the intestinal tract, there are interference factors such as gas and mucus, which makes the image prone to appear blurred, low contrast, and high noise, which seriously affects the observation and analysis of the image by doctors.
[0003] Traditional image enhancement methods are mainly based on handcrafted features and fixed algorithms, such as histogram equalization and filtering. Although these methods can improve the visual effect of the image to some extent, they lack the ability to understand the content of the image and adaptively adjust. For complex colon images, traditional methods often fail to achieve ideal enhancement results, and cannot effectively highlight the features of the lesion area, which is not conducive to accurate diagnosis by doctors.
[0004] Therefore, it is necessary to design a colon image enhancement method and system based on deep learning to solve the problems existing in the current technology. SUMMARY
[0005] In view of this, the present application proposes a colon image enhancement method and system based on deep learning, aiming to solve the problem that traditional methods often fail to achieve ideal enhancement results, cannot effectively highlight the features of the lesion area, and are not conducive to accurate diagnosis by doctors.
[0006] In one aspect, the present application proposes a colon image enhancement system based on deep learning, comprising: The acquisition module is configured to acquire a colon image to be enhanced, and pre-process the colon image to be enhanced to obtain a pre-processed colon image; and is further configured to acquire image feature information of the pre-processed colon image, and determine an initial enhancement coefficient of the pre-processed colon image according to the image feature information; The enhancement execution module is configured to perform enhancement operation on the pre-processed colon image according to the initial enhancement coefficient to obtain an enhanced colon image; a feedback adjustment module configured to obtain an evaluation image quality indicator in the enhanced colon image, and determine whether to adjust the initial enhancement coefficient according to the evaluation image quality indicator; if yes, extract structural features and interference features of the enhanced colon image based on a deep learning model, and adjust the initial enhancement coefficient according to the structural features and the interference features to obtain a final enhancement coefficient.
[0007] Further, when the to-be-enhanced colon image is preprocessed to obtain a preprocessed colon image, the method comprises: The to-be-enhanced colon image is sequentially subjected to noise suppression processing, histogram equalization processing and image sharpening processing to obtain the preprocessed colon image.
[0008] Further, when the initial enhancement coefficient of the preprocessed colon image is determined according to the image feature information, the method comprises: The image feature information is analyzed to obtain image feature types and image feature values corresponding to each of the image feature types; An image feature group is constructed according to the image feature types and the image feature values corresponding to each of the image feature types; The image feature group is compared with a historical enhancement group, and the initial enhancement coefficient of the preprocessed colon image is determined according to a comparison result; The initial enhancement coefficient comprises an initial feature enhancement coefficient corresponding to each of the image feature types.
[0009] Further, when the initial enhancement coefficient of the preprocessed colon image is determined according to the comparison result, the method comprises: When there is a historical image feature group identical to the image feature group in the historical enhancement group, a historical enhancement coefficient corresponding to the historical image feature group is taken as the initial feature enhancement coefficient corresponding to the image feature type; When there is no historical image feature group identical to the image feature group in the historical enhancement group, the initial feature enhancement coefficient corresponding to the image feature type is determined according to the image feature value.
[0010] Further, when the initial feature enhancement coefficient corresponding to the image feature type is determined according to the image feature value, the method comprises: The image feature value is compared with a first image feature value and a second image feature value, and the initial feature enhancement coefficient corresponding to the image feature type is determined according to a comparison result; wherein the first image feature value is smaller than the second image feature value; When the image feature value is smaller than or equal to the first image feature value, the initial feature enhancement coefficient corresponding to the image feature type is determined as a first feature enhancement coefficient; determining that the initial feature enhancement coefficient corresponding to the image feature type is a second feature enhancement coefficient when the image feature value is greater than the first image feature value and less than or equal to the second image feature value; determining that the initial feature enhancement coefficient corresponding to the image feature type is a third feature enhancement coefficient when the image feature value is greater than the second image feature value.
[0011] Further, when the evaluation image quality index in the enhanced colon image is obtained, the method comprises: performing feature extraction on the enhanced colon image to obtain a plurality of enhanced image feature values; obtaining an enhanced image standard value corresponding to each of the enhanced image feature values; calculating the evaluation image quality index according to the enhanced image feature value and the corresponding enhanced image standard value.
[0012] Further, when it is determined whether to adjust the initial enhancement coefficient according to the evaluation image quality index, the method comprises: comparing the evaluation image quality index with a preset quality index threshold, and determining whether to adjust the initial enhancement coefficient according to the comparison result; if the evaluation image quality index is greater than or equal to the preset quality index threshold, it is determined that the initial enhancement coefficient is not adjusted; if the evaluation image quality index is less than the preset quality index threshold, it is determined that the initial enhancement coefficient is adjusted.
[0013] Further, when the structural feature and the interference feature of the enhanced colon image are extracted based on a deep learning model, the method comprises: inputting the enhanced colon image into a pre-trained deep learning model; performing feature decomposition on the enhanced colon image through an encoding layer of the deep learning model to extract the structural feature of the colon image; performing feature separation on the enhanced colon image through a decoding layer of the deep learning model to extract the interference feature of the colon image; determining an enhancement influence factor based on the structural feature and the interference feature; adjusting the initial enhancement coefficient according to the enhancement influence factor to obtain the final enhancement coefficient.
[0014] Further, when the final enhancement coefficient is obtained by adjusting the initial enhancement coefficient according to the enhancement influence factor, the method comprises: comparing the enhancement influence factor with a preset adjustment coefficient mapping table to determine the adjustment coefficient of the initial enhancement coefficient according to the comparison result. adjust the initial enhancement coefficient according to the adjustment coefficient to obtain the final enhancement coefficient.
[0015] Compared with the prior art, the colon image enhancement system based on deep learning has the advantages that the colon image quality can be effectively improved. The acquisition module is preprocessed by noise suppression, histogram equalization and image sharpening, which reduces noise, enhances contrast and clarity, and can also determine the initial enhancement coefficient, so that the enhancement execution module can enhance the preprocessed image in a targeted manner. The feedback adjustment module is very important, which evaluates the image quality indicators. If the preset standard is not reached, the initial enhancement coefficient is adjusted by using the deep learning model to extract features, so as to ensure that a high-quality enhanced image is obtained. This mechanism improves the accuracy and adaptability of image enhancement. Compared with the traditional method, the system and method are more intelligent and automatic, the deep learning model can better process complex features, and effective enhancement is realized. The system process is rigorous, and each module cooperates to form a complete system, which can provide clear and accurate images for medical diagnosis, improve the accuracy and efficiency of diagnosis, and has important practical application value.
[0016] In another aspect, the application further provides a colon image enhancement method based on deep learning, comprising the following steps: acquiring a colon image to be enhanced, and preprocessing the colon image to be enhanced to obtain a preprocessed colon image; acquiring image feature information of the preprocessed colon image, and determining an initial enhancement coefficient of the preprocessed colon image according to the image feature information; performing an enhancement operation on the preprocessed colon image according to the initial enhancement coefficient to obtain an enhanced colon image; obtaining an evaluation image quality indicator in the enhanced colon image, and determining whether to adjust the initial enhancement coefficient according to the evaluation image quality indicator; if yes, extracting structural features and interference features of the enhanced colon image based on a deep learning model, and adjusting the initial enhancement coefficient according to the structural features and interference features to obtain a final enhancement coefficient.
[0017] It can be understood that the colon image enhancement method and system based on deep learning have the same advantages, which will not be described here. BRIEF DESCRIPTION OF DRAWINGS
[0018] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are for purposes of illustration only and are not considered a limitation of the present application. Moreover, like reference numerals are used to designate identical components throughout the specification. In the drawings: Figure 1A structural block diagram of a colon image enhancement system based on deep learning provided by an embodiment of the present application is shown in the figure. Figure 2 A flow chart of a colon image enhancement method based on deep learning provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION
[0019] Exemplary embodiments of the present disclosure will be described in detail with reference to the drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be accurately conveyed to those skilled in the art. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the drawings and in conjunction with the embodiments.
[0020] Reference is made to Figure 1 As shown in the figure, in some embodiments of the present application, the present embodiment provides a colon image enhancement system based on deep learning, comprising: The acquisition module is configured to acquire a colon image to be enhanced, and pre-process the colon image to be enhanced to obtain a pre-processed colon image; and is further configured to acquire image feature information of the pre-processed colon image, and determine an initial enhancement coefficient of the pre-processed colon image according to the image feature information; The enhancement execution module is configured to perform an enhancement operation on the pre-processed colon image according to the initial enhancement coefficient to obtain an enhanced colon image; The feedback adjustment module is configured to obtain an evaluation image quality indicator in the enhanced colon image, and determine whether to adjust the initial enhancement coefficient according to the evaluation image quality indicator; if so, extract structural features and interference features of the enhanced colon image based on a deep learning model, and adjust the initial enhancement coefficient according to the structural features and interference features to obtain a final enhancement coefficient.
[0021] It can be understood that the deep learning-based colon image enhancement system of the embodiment can effectively improve the quality of the colon image. The acquisition module is preprocessed through noise suppression, histogram equalization and image sharpening, etc., to reduce noise, enhance contrast and clarity, and also determine the initial enhancement coefficient, so that the enhancement execution module can enhance the preprocessed image in a targeted manner. The feedback adjustment module is very important, which evaluates the image quality indicators, and if the preset standard is not met, the deep learning model is used to extract features and adjust the initial enhancement coefficient to ensure that a high-quality enhanced image is obtained. This mechanism improves the accuracy and adaptability of image enhancement. Compared with traditional methods, the system and method are more intelligent and automated, the deep learning model can better process complex features, and effective enhancement is achieved. The system process is rigorous, and each module cooperates to form a complete system, which can provide clear and accurate images for medical diagnosis, improve the accuracy and efficiency of diagnosis, and has important practical application value.
[0022] Specifically, when the to-be-enhanced colon image is preprocessed to obtain the preprocessed colon image, the following steps are included: The to-be-enhanced colon image is sequentially subjected to noise suppression processing, histogram equalization processing and image sharpening processing to obtain the preprocessed colon image.
[0023] It can be understood that noise suppression processing can effectively reduce random noise in the image, making the image smoother and avoiding interference of noise on subsequent processing and analysis. Histogram equalization processing can adjust the gray scale distribution of the image, enhance the overall contrast of the image, and make the details in the image clearer and more distinguishable. Image sharpening processing further highlights the edges and contours of the image, improves the clarity of the image, and makes the features of the colon image more obvious. After a series of preprocessing operations, the quality of the preprocessed colon image is significantly improved, laying a good foundation for subsequent enhancement operations.
[0024] Specifically, when the initial enhancement coefficient of the preprocessed colon image is determined according to the image feature information, the following steps are included: The image feature information is analyzed to obtain image feature types and image feature values corresponding to each image feature type; An image feature group is constructed according to the image feature types and the image feature values corresponding to each image feature type; The image feature group is compared with a historical enhancement group, and the initial enhancement coefficient of the preprocessed colon image is determined according to the comparison result; The initial enhancement coefficient includes an initial feature enhancement coefficient corresponding to each image feature type.
[0025] In the embodiment, the image feature types include contrast, brightness, noise level, image color, etc. Different image feature types reflect different aspects of the characteristics of the colon image. The contrast feature embodies the difference between the bright and dark parts in the image, and appropriate contrast can make the tissue structure in the colon image more clear and distinguishable; the brightness feature determines the overall brightness of the image, and appropriate brightness helps doctors observe the image details more accurately; the noise level feature reflects the strength of random interference signals in the image, and lower noise level can ensure the purity of the image and avoid interference in diagnosis; the image color feature covers the color information of the image, and accurate color restoration is of great significance for identifying the pathological conditions of the colon tissue.
[0026] Specifically, when determining the initial enhancement coefficient of the preprocessed colon image according to the comparison result, the method comprises: when there is a historical image feature group in the historical enhancement group that is the same as the image feature group, taking the historical enhancement coefficient corresponding to the historical image feature group as the initial feature enhancement coefficient corresponding to the image feature type; when there is no historical image feature group in the historical enhancement group that is the same as the image feature group, determining the initial feature enhancement coefficient corresponding to the image feature type according to the image feature value.
[0027] It can be understood that after the image feature group is constructed, comparing it with the historical enhancement group is a key step for determining the initial enhancement coefficient. The historical enhancement group is a series of image feature groups and their corresponding successful enhancement coefficients accumulated by the system in the past processing process. By comparison, the enhancement experience of similar image features in history can be referred to to determine a more appropriate initial enhancement coefficient for the current preprocessed colon image. For example, if the contrast feature value in the current image feature group is similar to the contrast feature value of a certain group in the historical enhancement group, and the initial contrast feature enhancement coefficient corresponding to the historical group has achieved good results after enhancement, then the coefficient can be taken as the initial feature enhancement coefficient of the contrast of the current image.
[0028] Specifically, when determining the initial feature enhancement coefficient corresponding to the image feature type according to the image feature value, the method comprises: comparing the image feature value with a first image feature value and a second image feature value, and determining the initial feature enhancement coefficient corresponding to the image feature type according to the comparison result; wherein the first image feature value is less than the second image feature value; when the image feature value is less than or equal to the first image feature value, determining the initial feature enhancement coefficient corresponding to the image feature type as the first feature enhancement coefficient; determining the initial feature enhancement coefficient corresponding to the image feature type as the second feature enhancement coefficient when the image feature value is greater than the first image feature value and less than or equal to the second image feature value; determining the initial feature enhancement coefficient corresponding to the image feature type as the third feature enhancement coefficient when the image feature value is greater than the second image feature value.
[0029] It can be understood that by comparing the image feature value with the first image feature value and the second image feature value, the appropriate initial feature enhancement coefficient can be determined for different image feature types according to different intervals of the image feature value. This way of determining the initial feature enhancement coefficient in different intervals can more carefully consider the influence of the change of the image feature value on the enhancement effect. For example, for the image feature type of contrast, if the image feature value is less than or equal to the first image feature value, it indicates that the contrast of the image is low, and the first feature enhancement coefficient is determined at this time, which is relatively large to enhance the contrast to a large extent, so that the tissue structure in the image is clearer. If the image feature value of the contrast is greater than the first image feature value and less than or equal to the second image feature value, it indicates that the contrast of the image is at a medium level, and the second feature enhancement coefficient is used, which is moderate to moderately enhance the contrast. When the image feature value of the contrast is greater than the second image feature value, it indicates that the contrast of the image is already high, and the third feature enhancement coefficient is used, which is small to avoid excessive enhancement of the contrast and cause image distortion. This method of determining the initial feature enhancement coefficient according to the image feature value in different intervals can perform enhancement operations according to different feature conditions of the colon image, improve the accuracy and adaptability of the colon image enhancement, and further ensure the clarity and accuracy of the image in subsequent medical diagnosis, providing more reliable diagnostic basis for doctors. At the same time, combined with the further adjustment of the initial enhancement coefficient by the feedback adjustment module, a more perfect and intelligent colon image enhancement system can be formed to better meet the actual needs of medical diagnosis.
[0030] Specifically, when obtaining the evaluation image quality index in the enhanced colon image, the method comprises: performing feature extraction on the enhanced colon image to obtain a plurality of enhanced image feature values; obtaining an enhanced image standard value corresponding to each of the enhanced image feature values; calculating the evaluation image quality index according to the enhanced image feature values and the corresponding enhanced image standard values.
[0031] In this embodiment, the evaluation image quality index is obtained by the following formula: ; wherein, EI represents the i-th enhanced image feature value; ωi represents the weight coefficient corresponding to the i-th enhanced image feature value; Fi represents the i-th enhanced image feature value; Ti represents the enhanced image standard value corresponding to the i-th enhanced image feature value; Fi,max and Fi,min represent the normalized upper limit and lower limit of the i-th enhanced image feature value, respectively.
[0032] In this embodiment, the enhanced image feature value is preferably an enhanced contrast, brightness, noise level, image color, etc. These feature values correspond to the image feature types involved in determining the initial enhancement coefficient.
[0033] Specifically, when determining whether to adjust the initial enhancement coefficient according to the evaluation image quality indicator, the method comprises: comparing the evaluation image quality indicator with a preset quality indicator threshold, and determining whether to adjust the initial enhancement coefficient according to the comparison result; if the evaluation image quality indicator is greater than or equal to the preset quality indicator threshold, it is determined that the initial enhancement coefficient is not adjusted; if the evaluation image quality indicator is less than the preset quality indicator threshold, it is determined that the initial enhancement coefficient is adjusted.
[0034] It can be understood that comparing the evaluation image quality indicator with the preset quality indicator threshold is an important basis for determining whether to adjust the initial enhancement coefficient. The preset quality indicator threshold is an empirical value determined according to a large number of experiments and clinical practice, which represents the minimum standard of image quality that meets the needs of medical diagnosis. When the evaluation image quality indicator is greater than or equal to the preset quality indicator threshold, it means that the quality of the enhanced colon image has reached or exceeded the expected standard, and the initial enhancement coefficient can make the image achieve good enhancement effect. At this time, there is no need to adjust the initial enhancement coefficient, which avoids the decline of image quality caused by excessive operation. When the evaluation image quality indicator is less than the preset quality indicator threshold, it means that the quality of the enhanced colon image does not meet the requirements, and the initial enhancement coefficient needs to be adjusted to further improve the image quality. This judgment method based on quantitative indicators avoids the interference of subjective factors, making the image enhancement process more scientific and accurate.
[0035] Specifically, when extracting the structural features and interference features of the enhanced colon image based on the deep learning model, the method comprises: inputting the enhanced colon image into a pre-trained deep learning model; performing feature decomposition on the enhanced colon image through the encoding layer of the deep learning model to extract the structural features of the colon image; perform feature separation on the enhanced colon image through a decoding layer of the deep learning model to extract interference features of the colon image; determine an enhancement influence factor based on the structural features and the interference features; adjust the initial enhancement coefficient according to the enhancement influence factor to obtain the final enhancement coefficient.
[0036] It can be understood that the pre-trained deep learning model has strong ability in extracting structural features and interference features. The encoding layer can deeply analyze the enhanced colon image, decompose its complex information, and mine features representing the essential structure of the image, such as the outline of the colon and the morphology of the internal organization. These structural features are key information that needs to be retained and strengthened in the image. The decoding layer focuses on separating interference features such as random noise and artifacts from the image. These interference features will affect the quality of the image and the accuracy of the diagnosis, and need to be processed. By combining the structural features and the interference features to determine the enhancement influence factor, the various aspects of the image can be considered comprehensively. The enhancement influence factor reflects the specific influence degree of the structural features and the interference features on the image enhancement. For example, if there are many interference features and the influence is large, the enhancement influence factor will cause the initial enhancement coefficient to adjust in the direction of reducing interference. If the structural features are prominent but need to be further strengthened, the enhancement influence factor will promote the initial enhancement coefficient to play a greater role in enhancing the structural features.
[0037] In this embodiment, the enhancement influence factor is obtained by the following formula: ; wherein IF represents the enhancement influence factor, a and β are weight coefficients of the structural features and the interference features respectively, SF represents the quantized value of the structural features, and DF represents the quantized value of the interference features. The values of a and β are adjusted according to different application scenarios and requirements to balance the influence of the structural features and the interference features on the adjustment of the initial enhancement coefficient.
[0038] In this embodiment, when quantifying the structural features, various methods can be adopted. A common way is to convert the structural features into numerical representation, for example, by calculating the area, perimeter, shape complexity and other geometric parameters of the structural features to quantify. For the contour structure of the colon, its perimeter and area can be calculated, and a longer perimeter and a suitable area can indicate that the contour features are clearer and more complete, and these numerical values can be used as the basis for quantifying the structural features. For the internal tissue morphology features, texture analysis methods can be used to quantify the characteristics of the tissue morphology by calculating the contrast, correlation and entropy of the texture. For the quantification of interference features, various strategies can also be adopted. For random noise interference features, the variance of the noise can be calculated, and the larger the variance, the more obvious the noise, and the variance value can be used as the quantification value of the interference features. For artifact interference features, the area, number and contrast with normal image regions can be detected to quantify the artifacts.
[0039] Specifically, when adjusting the initial enhancement coefficient according to the enhancement influence factor to obtain the final enhancement coefficient, the method comprises: comparing the enhancement influence factor with a preset adjustment coefficient mapping table, and determining an adjustment coefficient of the initial enhancement coefficient according to a comparison result; adjusting the initial enhancement coefficient according to the adjustment coefficient to obtain the final enhancement coefficient.
[0040] It can be understood that the preset adjustment coefficient mapping table is established by the system according to a large number of experiments and historical data, which records the initial enhancement coefficient adjustment coefficients corresponding to different enhancement influence factors. By comparing the currently calculated enhancement influence factor with the preset adjustment coefficient mapping table, the appropriate adjustment coefficient can be quickly and accurately found. For example, when the enhancement influence factor is in a certain specific interval, the corresponding adjustment coefficient will be explicitly given in the mapping table. This adjustment coefficient is determined according to past experience and experimental results, which reflects the direction and amplitude of the initial enhancement coefficient that needs to be adjusted under the enhancement influence factor. After the adjustment coefficient is determined, the initial enhancement coefficient can be adjusted. When the feature type is brightness, contrast and image color, the final enhancement coefficient is the product of the initial enhancement coefficient and the adjustment coefficient, i.e. the final enhancement coefficient = initial enhancement coefficient x adjustment coefficient. This multiplication calculation method can reasonably enlarge or reduce the initial enhancement coefficient according to the size of the adjustment coefficient, so as to achieve the purpose of adjusting the corresponding features of the image. For example, if the adjustment coefficient is greater than 1, the final enhancement coefficient will be greater than the initial enhancement coefficient, which means that the feature needs to be further enhanced; if the adjustment coefficient is less than 1, the final enhancement coefficient is less than the initial enhancement coefficient, that is, the enhancement degree of the feature needs to be weakened. When the feature type is noise level, the final enhancement coefficient is the difference between the initial enhancement coefficient and the adjustment coefficient, i.e. the final enhancement coefficient = initial enhancement coefficient - adjustment coefficient. Because the noise level is a feature that needs to be reduced, through this difference calculation method, the final enhancement coefficient can be adjusted in the direction of reducing noise. For example, if the adjustment coefficient is positive, the final enhancement coefficient will be less than the initial enhancement coefficient, so as to reduce the noise in the image and improve the purity of the image. By using different adjustment methods for different feature types, different features of the colon image can be enhanced and optimized more accurately. For brightness, contrast and image color, which need to be enhanced or adjusted appropriately, multiplication adjustment can flexibly control the amplitude of enhancement; for noise level, which needs to be suppressed, subtraction adjustment can directly and effectively reduce the noise effect.
[0041] Referring to Figure 2 In some embodiments of the present application, the present embodiment provides a deep learning-based colon image enhancement method, comprising the following steps: S100: Collecting a colon image to be enhanced, and preprocessing the colon image to be enhanced to obtain a preprocessed colon image; S200: Collecting image feature information of the preprocessed colon image, and determining an initial enhancement coefficient of the preprocessed colon image according to the image feature information; S300: Performing an enhancement operation on the preprocessed colon image according to the initial enhancement coefficient to obtain an enhanced colon image; S400: Obtain an evaluation image quality indicator in the enhanced colon image, and determine whether to adjust the initial enhancement coefficient according to the evaluation image quality indicator; if yes, extract a structure feature and an interference feature of the enhanced colon image based on a deep learning model, and adjust the initial enhancement coefficient according to the structure feature and the interference feature to obtain a final enhancement coefficient.
[0042] Those skilled in the art will understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. In addition, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0043] The present application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems) and computer program products of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of flows and / or blocks in the flowcharts and / or block diagrams. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more flows or blocks.
[0044] These computer program instructions can also be stored in a computer-readable memory that can guide the computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including instruction devices, which implement the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more flows or blocks.
[0045] These computer program instructions can also be loaded into a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more flows or blocks.
[0046] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application but not to limit it. Although the present application has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that the specific embodiments of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or equivalent replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.
Claims
1. A colon image enhancement system based on deep learning, characterized in that, include: The acquisition module is configured to acquire images of the colon to be enhanced and to preprocess the images of the colon to be enhanced to obtain preprocessed images of the colon. It is also configured to acquire image feature information of the preprocessed colon image and determine the initial enhancement coefficient of the preprocessed colon image based on the image feature information; An enhancement execution module is configured to perform an enhancement operation on the preprocessed colon image based on the initial enhancement coefficient to obtain an enhanced colon image; The feedback adjustment module is configured to obtain an evaluation image quality index in the enhanced colon image and determine whether to adjust the initial enhancement coefficient based on the evaluation image quality index. If so, the structural features and interference features of the enhanced colon image are extracted based on the deep learning model, and the initial enhancement coefficient is adjusted according to the structural features and interference features to obtain the final enhancement coefficient.
2. The colon image enhancement system based on deep learning according to claim 1, characterized in that, Preprocessing the colon image to be enhanced to obtain a preprocessed colon image includes: The colon image to be enhanced is subjected to noise suppression, histogram equalization, and image sharpening processes in sequence to obtain the preprocessed colon image.
3. The deep learning-based colon image enhancement system according to claim 2, characterized in that, When determining the initial enhancement coefficient of the preprocessed colon image based on the image feature information, the following steps are included: The image feature information is parsed to obtain the image feature type and the image feature value corresponding to each image feature type; Construct an image feature group based on the image feature type and the image feature value corresponding to each image feature type; The image feature group is compared with the historical enhancement group, and the initial enhancement coefficient of the preprocessed colon image is determined based on the comparison result. The initial enhancement coefficients include the initial feature enhancement coefficients corresponding to each of the image feature types.
4. The deep learning-based colon image enhancement system according to claim 3, characterized in that, When determining the initial enhancement coefficient of the preprocessed colon image based on the comparison results, the following steps are included: When there is a historical image feature group in the historical enhancement group that is the same as the image feature group, the historical enhancement coefficient corresponding to the historical image feature group is used as the initial feature enhancement coefficient corresponding to the image feature type; When there is no historical image feature group in the historical enhancement group that is the same as the image feature group, the initial feature enhancement coefficient corresponding to the image feature type is determined according to the image feature value.
5. The deep learning-based colon image enhancement system according to claim 4, characterized in that, When determining the initial feature enhancement coefficient corresponding to the image feature type based on the image feature values, the following are included: The image feature value is compared with the first image feature value and the second image feature value, and the initial feature enhancement coefficient corresponding to the image feature type is determined based on the comparison result; wherein, the first image feature value is less than the second image feature value; When the image feature value is less than or equal to the first image feature value, the initial feature enhancement coefficient corresponding to the image feature type is determined as the first feature enhancement coefficient; When the image feature value is greater than the first image feature value and less than or equal to the second image feature value, the initial feature enhancement coefficient corresponding to the image feature type is determined as the second feature enhancement coefficient; When the image feature value is greater than the second image feature value, the initial feature enhancement coefficient corresponding to the image feature type is determined as the third feature enhancement coefficient.
6. The deep learning-based colon image enhancement system according to claim 5, characterized in that, Obtaining the evaluation image quality metrics in the enhanced colon image includes: Feature extraction is performed on the enhanced colon image to obtain several enhanced image feature values; Obtain the standard value of the enhanced image corresponding to each of the enhanced image feature values; The evaluation image quality index is calculated based on the enhanced image feature values and their corresponding enhanced image standard values.
7. The deep learning-based colon image enhancement system according to claim 6, characterized in that, When determining whether to adjust the initial enhancement coefficient based on the evaluated image quality index, the following steps are included: The evaluated image quality index is compared with a preset quality index threshold, and the initial enhancement coefficient is adjusted based on the comparison result. If the evaluated image quality index is greater than or equal to the preset quality index threshold, it is determined that the initial enhancement coefficient will not be adjusted. If the evaluated image quality index is less than the preset quality index threshold, it is determined that the initial enhancement coefficient should be adjusted.
8. The colon image enhancement system based on deep learning according to claim 7, characterized in that, When extracting structural and interference features from the enhanced colon image using a deep learning model, the following are included: The enhanced colon image is then input into a pre-trained deep learning model; The enhanced colon image is decomposed using the encoding layer of the deep learning model to extract its structural features. The enhanced colon image is subjected to feature separation through the decoding layer of the deep learning model to extract interference features of the colon image; The enhancement influence factor is determined based on the structural features and the interference features; The initial enhancement coefficient is adjusted according to the enhancement influence factor to obtain the final enhancement coefficient.
9. The deep learning-based colon image enhancement system according to claim 8, characterized in that, When adjusting the initial enhancement coefficient based on the enhancement impact factor to obtain the final enhancement coefficient, the following steps are included: The enhancement factor is compared with the preset adjustment coefficient mapping table, and the adjustment coefficient of the initial enhancement coefficient is determined based on the comparison result. The initial enhancement coefficient is adjusted according to the adjustment coefficient to obtain the final enhancement coefficient.
10. A deep learning-based colon image enhancement method, applied to the deep learning-based colon image enhancement system as described in any one of claims 1-9, characterized in that, include: Acquire images of the colon to be enhanced, and preprocess the images to obtain preprocessed colon images; The image feature information of the preprocessed colon image is acquired, and the initial enhancement coefficient of the preprocessed colon image is determined based on the image feature information; An enhancement operation is performed on the preprocessed colon image based on the initial enhancement coefficient to obtain an enhanced colon image; Obtain the evaluation image quality index in the enhanced colon image, and determine whether to adjust the initial enhancement coefficient based on the evaluation image quality index; If so, the structural features and interference features of the enhanced colon image are extracted based on the deep learning model, and the initial enhancement coefficient is adjusted according to the structural features and interference features to obtain the final enhancement coefficient.