An endoscope-based lesion detection auxiliary exposure method and an auxiliary exposure device

Abstract

The application discloses an endoscope-based lesion detection auxiliary exposure method and auxiliary exposure device, and the method comprises the following steps: globally exposing a target position to obtain an exposure image; judging the definition of the exposure image; when the definition of the exposure image is less than a definition threshold, acquiring an infrared thermal image of the target position; demarcating a lesion feature range in the target position from the infrared thermal image of the target position; partition-exposing the lesion feature range in the target position to obtain a lesion feature image; and detecting the lesion feature image obtained through the partition exposure to identify the lesion in the image. The application can effectively overcome the interference of liquid mixtures in the environment in the body of a subject on endoscope imaging through infrared thermal imaging means; meanwhile, a regional endoscope exposure method is also used, which can effectively reduce the problem of hair burning of the tip of the endoscope, and optimizes the use experience of the endoscope and the lesion detection experience of the subject.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of endoscopy, in particular to a lesion detection auxiliary exposure method and device based on endoscopy. BACKGROUND

[0002] An endoscope is a medical instrument with the purpose of observation, which is an auxiliary device for examination, diagnosis or treatment by extending into the natural or surgically opened orifices of a subject (human or animal body). One of the functions of the endoscope is to help medical personnel identify whether there is a tumor, nodule or other lesion in the environment of the subject's body. However, due to the presence of various liquid mixtures in the environment of the subject's body and the peristalsis of the intestines and other cavities, the collected environment images may not meet the clarity requirements, which may interfere with the identification of the lesion. At this time, the exposure of the endoscope is generally increased to penetrate these liquid mixtures.

[0003] However, the existing endoscope adjusts the exposure globally. When the lesion is located in a position where the light is insufficient or the endoscope view is blocked by various liquid mixtures, in order to observe the local tissue more clearly, the global exposure amount has to be increased, which increases the heat generation of the tip head of the endoscope. The high temperature may affect the performance of the endoscope and cause discomfort to the subject. SUMMARY

[0004] Therefore, the present application provides a lesion detection auxiliary exposure method and device based on endoscopy.

[0005] The first aspect of the present application provides a lesion detection auxiliary exposure method based on endoscopy, comprising the following steps:

[0006] using a visible light acquisition module to perform global exposure on the target position to obtain an exposure image;

[0007] judging the clarity of the exposure image;

[0008] when the clarity of the exposure image is less than a clarity threshold, performing the following steps:

[0009] using an infrared thermal image module to obtain an infrared thermal image of the target position; the infrared thermal image of the target position contains a plurality of thermal pixels;

[0010] defining a lesion feature range in the target position from the infrared thermal image of the target position;

[0011] using a visible light image acquisition module to perform partition exposure on the lesion feature range in the target position to obtain a lesion feature image;

[0012] detecting the lesion feature image obtained by the partition exposure to identify the lesion in the image.

[0013] Further, the step of demarcating the lesion feature range in the target position from the target position image comprises the following steps:

[0014] Obtaining the coordinate value of the hot pixel point in the target position infrared thermal image;

[0015] Marking the coordinate value of the hot pixel point;

[0016] Combining the hot pixel points with adjacent coordinate values to form a lesion feature area;

[0017] Obtaining the coordinate value of the edge hot pixel point of the lesion feature area to obtain the lesion feature range in the target position.

[0018] Further, after the step of demarcating the lesion feature range in the target position from the target position infrared thermal image, the following steps are further included:

[0019] Judging whether the target position infrared thermal image has shifted, and if so, correcting the lesion feature range in the target position.

[0020] Further, the step of correcting the lesion feature range in the target position comprises the following steps:

[0021] Obtaining a plurality of target position infrared thermal images collected within a preset time period;

[0022] Marking the lesion feature range in the target position in the plurality of target position infrared thermal images, respectively;

[0023] Performing image superposition processing on the plurality of target position infrared thermal images to obtain an incomplete superposition lesion feature area coordinate region;

[0024] Taking the edge contour of the incomplete superposition lesion feature area coordinate region as the corrected lesion feature area coordinate region.

[0025] Further, the step of partitioning exposure of the lesion feature range in the target position comprises the following steps:

[0026] Performing coordinate transformation on the lesion feature range in the target position based on coordinate transformation coefficients;

[0027] Controlling the visible light image acquisition module to perform partitioned exposure on the lesion feature range in the target position.

[0028] Further, the coordinate transformation coefficients are obtained by the following steps:

[0029] Taking a calibration plane, the calibration plane having at least three calibration points;

[0030] applying a heat source on the calibration point;

[0031] acquiring the calibration point coordinates on the calibration plane using a visible light image acquisition module to obtain a first calibration point coordinate set;

[0032] acquiring the calibration point coordinates on the calibration plane using an infrared thermal image module to obtain a second calibration point coordinate set;

[0033] determining the coordinate transformation coefficient according to the coordinate conversion relationship of the first calibration point coordinate set and the second calibration point coordinate set.

[0034] Further, the control visible light image acquisition module partitions the lesion feature range in the target position for exposure, specifically including the following steps:

[0035] dividing the photosensitive area of the visible light image acquisition module into multiple sub-regions;

[0036] setting the exposure weight value of each sub-region;

[0037] controlling the visible light image acquisition module for partitioned exposure according to the weight value of each sub-region.

[0038] Further, when the exposure image clarity is greater than or equal to a preset clarity threshold, the following steps are performed:

[0039] detecting the exposure image to identify the lesion in the image.

[0040] Further, the identification of the lesion in the image specifically includes the following steps:

[0041] setting a training set composed of various types of lesion feature images, and marking these lesion feature images with respect to the lesion type;

[0042] inputting the training set into a neural network model for training to obtain a trained neural network model;

[0043] inputting the lesion feature image or exposure image into the trained neural network model to obtain the confidence of the lesion feature image or exposure image belonging to various types of lesions;

[0044] selecting the lesion type with the highest confidence as the identification result of the lesion in the lesion feature image or exposure image.

[0045] The second aspect of the present application discloses a lesion detection auxiliary exposure device installed in an endoscope, comprising the following modules:

[0046] a visible light image acquisition module, which is used for visible light exposure;

[0047] A global exposure module is configured to control the visible light acquisition module to perform global exposure on the target position to obtain an exposure image.

[0048] A definition module is configured to define the target position according to the exposure image.

[0049] An infrared thermal image module is configured to acquire an infrared thermal image of the target position when the definition of the exposure image is less than a definition threshold; the infrared thermal image of the target position includes a plurality of thermal image points.

[0050] A feature range definition module is configured to define a lesion feature range in the target position from the infrared thermal image of the target position.

[0051] A partition exposure module is configured to control the visible light image acquisition module to perform partition exposure on the lesion feature range in the target position to obtain a lesion feature image.

[0052] A lesion identification module is configured to detect the lesion feature image obtained by the partition exposure and identify a lesion in the image.

[0053] The embodiment of the present application has the following beneficial effects: the lesion detection auxiliary exposure method and the lesion detection auxiliary exposure device based on an endoscope can effectively overcome the interference of a liquid mixture in the body of a subject on endoscope imaging when the exposure image is not clear, and obtain an accurate lesion area; meanwhile, the regional endoscope exposure method is used, and after the lesion area coordinates are determined, the lesion area is subjected to partition exposure to further identify the lesion, which can effectively avoid the temperature of the tip of the endoscope being too high, and optimizes the use experience of the endoscope and the lesion detection experience of the subject.

[0054] Additional aspects and advantages of the present application will be described in the following description part, some of which will become apparent from the following description, or will be understood through practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0056] Figure 1 is the main step flow chart of the lesion detection auxiliary exposure method and the lesion detection auxiliary exposure device based on an endoscope of the present application;

[0057] Figure 2 is a main structure schematic diagram of a lesion detection auxiliary exposure method and a lesion detection auxiliary exposure device based on an endoscope. DETAILED DESCRIPTION

[0058] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0059] In view of the problem that the lesion recognition may not be clear due to the liquid mixture blocking the field of view when using an endoscope to identify the lesion, the present embodiment proposes a lesion detection auxiliary exposure method based on an endoscope, which overcomes the interference of the liquid mixture by means of partition exposure of the lesion area.

[0060] Embodiment 1

[0061] The lesion detection auxiliary exposure method based on an endoscope of the present embodiment, as shown in Figure 1 , mainly includes the following steps:

[0062] S1. using a visible light acquisition module to perform global exposure on the target position to obtain an exposure image;

[0063] S2. judging the clarity of the exposure image;

[0064] S3. when the clarity of the exposure image is less than the clarity threshold, using an infrared thermal image module to obtain an infrared thermal image of the target position; the infrared thermal image of the target position contains a plurality of thermal image points;

[0065] S4. demarcating a lesion feature range in the target position from the infrared thermal image of the target position;

[0066] S5. using a visible light image acquisition module to perform partition exposure on the lesion feature range in the target position to obtain a lesion feature image;

[0067] S6. detecting the lesion feature image obtained by partition exposure to identify the lesion in the image.

[0068] The lesion detection auxiliary exposure method based on an endoscope of the present embodiment is generally used in an endoscope system, that is, the steps of the present application are implemented through an endoscope and its upper computer. In some alternative embodiments, the present application can also be implemented in the form of Internet of Things or sub-area chain, that is, the infrared thermal image is obtained by operating the endoscope in the form of remote control or cloud control, and the steps of the present application are implemented.

[0069] The steps of the lesion detection auxiliary exposure method based on an endoscope of the present embodiment are described in detail as follows.

[0070] S1. Collecting a global exposure image of the target position using a visible light collecting module. In some embodiments, the lesion feature image collected by the endoscope may not be blocked by the liquid mixture, i.e., the lesion feature image is clear enough and does not need to be exposed in sections. Therefore, this embodiment first determines whether the lesion feature image meets the clarity requirement by performing global exposure on the target position.

[0071] S2. Determining the clarity of the exposure image. In this embodiment, if the clarity of the exposure image is less than the clarity threshold, the subsequent steps S3-S4 need to be performed; if the clarity of the exposure image is greater than or equal to the clarity threshold, step S6 is directly performed for lesion recognition. Specifically, in addition to setting a clarity threshold and determining whether to perform steps S3-S4 by judging whether the clarity of the endoscope image is greater than the clarity threshold, it can also be determined whether the endoscope image is clear enough to recognize the lesion by manual recognition; or the process of determining the clarity of the exposure image can be completed by using a neural learning network.

[0072] S3. Acquiring an infrared thermal image of the target position using an infrared thermal image module; the infrared thermal image of the target position includes a plurality of thermal pixels. The infrared thermal image module is a precision instrument for detecting the infrared radiation of the cavity organ, which displays the changes in temperature distribution, the positions and degrees of change in the interface of the infrared thermal image module in the form of a thermal image. The infrared thermal image refers to an image obtained by converting the infrared radiation based on infrared light into temperature. Since the position near the lesion is usually different in temperature from the rest of the normal position in the image due to changes in blood flow or metabolism, the liquid mixture usually does not produce temperature differences. Therefore, the lesion position is collected by the infrared thermal image, and the thermal radiation difference of the lesion area is different from other areas, i.e., a plurality of thermal pixels are formed in the lesion area. In this embodiment, the thermal pixel represents a pixel with a higher color temperature than other areas in the infrared thermal image, which can be achieved by setting a color temperature threshold. That is, in the infrared thermal image, the pixel with a color temperature higher than the preset color temperature threshold is identified as a thermal pixel, and the pixel with a color temperature lower than the preset color temperature threshold is identified as a normal pixel.

[0073] S4. Defining the lesion feature range in the target position from the infrared thermal image of the target position. Step S2 defines the lesion feature range in the target position from the target position image, which includes the following steps:

[0074] S4-1. Acquiring the coordinate value of the thermal pixel in the infrared thermal image of the target position.

[0075] In this embodiment, a coordinate system can be established on the infrared thermal image, and the positions of the thermal pixel points in the infrared thermal image can be converted into digital signals in the form of coordinates for subsequent formation of the lesion feature range in the target position. In other embodiments, the thermal radiation points in the image can also be identified in the form of a neural model, and coordinate values representing the positions of the thermal radiation points can be output.

[0076] S4-2 Marking the coordinate values of the thermal pixel points.

[0077] S4-3. Combining the thermal pixel points with coordinate values adjacent to each other to form a lesion feature area.

[0078] In this embodiment, the coordinate values of the thermal pixel points can be marked manually or automatically by an algorithm. Since the purpose of step S4 is to obtain the lesion feature range in the target position, after the thermal pixel points are marked, this embodiment combines the thermal pixel points with coordinate values adjacent to each other to expand the connection area of the thermal pixel points, and finally forms a lesion feature area. In this embodiment, the lesion feature area refers to the area in the infrared thermal image where the lesion feature phenomenon appears, and in the coordinate system, it refers to the coordinate area connected by a plurality of thermal pixel points adjacent to each other.

[0079] This embodiment can obtain the lesion feature range by the coordinate system marking method, and can also collect the lesion feature area by the image segmentation model, that is, by identifying the lesion feature area in the infrared thermal image through a trained CNN (Convolutional Neural Network) model. Specifically, by inputting multiple images marked with lesion areas and images without lesions to the CNN model, the CNN model is trained so that the CNN model has the ability to identify the lesion feature area in the infrared thermal image.

[0080] S4-4. Obtaining the coordinate values of the edge thermal pixel points of the lesion feature area to obtain the lesion feature range in the target position.

[0081] In this embodiment, after the lesion feature area is determined, in order to convert the lesion feature area into a digital signal, the aforementioned coordinate system is used to represent the lesion feature area in the form of a coordinate area. Specifically, the thermal pixel points at the edge positions in the lesion feature area can be collected, the lesion feature range in the corresponding target position can be formed by the edge position thermal pixel points, and the lesion feature area can be represented by the coordinate area. For some independently distributed thermal pixel points, since there is no lesion area with a size of only one pixel point in existing medical practice, this embodiment usually ignores these independent thermal pixel points, that is, does not identify them as lesion areas; or submits the independent thermal pixel points to manual processing to determine whether there is a lesion at the marked position of the independent thermal pixel points.

[0082] In some embodiments, due to objective reasons such as peristalsis of the cavity organ or shaking of the endoscope during manual operation, the lesion feature range in the target position collected by the endoscope may not be clear. Therefore, the present embodiment further comprises the following steps after the step S4 of delimiting the lesion feature range in the target position from the infrared thermal image of the target position:

[0083] S4-5. Determine whether the infrared thermal image of the target position is shifted. If the infrared thermal image of the target position is shifted, correct the lesion feature range in the target position.

[0084] The step S4-5 specifically comprises:

[0085] S4-5-1. Obtain a plurality of infrared thermal images collected in a preset time period;

[0086] S4-5-2. Mark the lesion feature range in the target position in each of the plurality of infrared thermal images;

[0087] S4-5-3. Perform image superposition processing on the plurality of infrared thermal images to obtain an incomplete superposition lesion feature region coordinate region;

[0088] S4-5-4. Take the edge profile of the incomplete superposition lesion feature region coordinate region as a corrected lesion feature region coordinate region.

[0089] In the present embodiment, when the infrared thermal image of the target position is shifted, by collecting a plurality of infrared thermal images before and after the shift, superimposing the plurality of infrared thermal images, and taking the maximum lesion feature region coordinate region as a corrected lesion feature region coordinate region, the problem of unclear lesion feature range in the target position caused by the shift of the endoscope can be solved.

[0090] S5. Perform zoned exposure on the lesion feature range in the target position using the visible light image acquisition module to obtain a lesion feature image. The step S5 of performing zoned exposure on the lesion feature range in the target position using the visible light image acquisition module specifically comprises the following steps:

[0091] S5-1. Perform coordinate transformation on the lesion feature range in the target position based on the coordinate transformation coefficient;

[0092] S5-2. Control the visible light image acquisition module to perform zoned exposure on the lesion feature range in the target position.

[0093] The partition exposure of step S5 is mainly realized by the visible light image acquisition module. The visible light image acquisition module is different from the infrared thermal image module, and is an acquisition module that can acquire images in the visible light range of human eyes. Due to differences in installation positions or focal lengths, the coordinate region in the infrared thermal image cannot be directly applied to the visible light image acquisition module, and a coordinate transformation operation needs to be performed first to obtain a coordinate transformation coefficient, and then step S5-1 is executed to perform coordinate transformation on the lesion feature range in the target position based on the coordinate transformation coefficient.

[0094] In step S5-1, the coordinate transformation coefficient is obtained through the following steps:

[0095] S5-1-1. Take a calibration plane, and the calibration plane has at least three calibration points;

[0096] S5-1-2. Apply a heat source to the calibration points;

[0097] S5-1-3. Acquire the coordinates of the calibration points on the calibration plane using the visible light image acquisition module to obtain a first calibration point coordinate set;

[0098] S5-1-4. Acquire the coordinates of the calibration points on the calibration plane using the infrared thermal image module to obtain a second calibration point coordinate set;

[0099] S5-1-5. Determine the coordinate transformation coefficient according to the coordinate conversion relationship of the first calibration point coordinate set and the second calibration point coordinate set.

[0100] In this embodiment, the calibration plane is used to set the coordinate transformation coefficient. According to mathematical logic, three points can determine a plane, so at least three calibration points are set on the calibration plane in this embodiment, so that the visible light image acquisition module can acquire the coordinates of the three calibration points, i.e., the first calibration point coordinate set. Since the infrared thermal image module cannot recognize the calibration points on the calibration plane, a heat source is applied to the calibration points in this embodiment so that the calibration points can be acquired by the infrared thermal image module to obtain the second calibration point coordinate set. Finally, the coordinate conversion relationship between the infrared thermal image module coordinates and the visible light image acquisition module coordinates is established through the first calibration point coordinate set and the second calibration point coordinate set to obtain the coordinate transformation coefficient.

[0101] After obtaining the coordinate transformation coefficient, the lesion feature range in the target position is transformed into the lesion feature range in the target position, and the visible light image acquisition module is controlled to perform partition exposure on the lesion feature range in the target position.

[0102] Step S5-2 controls the visible light image acquisition module to perform partition exposure on the lesion feature range in the target position, which specifically includes the following steps:

[0103] S5-2-1. Divide the photosensitive region of the visible light image acquisition module into multiple sub-regions;

[0104] S5-2-2. Set the exposure weight value of each sub-region;

[0105] S5-2-3. Control the visible light image acquisition module to perform partition exposure according to the weight value of each sub-region.

[0106] In this embodiment, the exposure is not performed on the whole region of the image collected by the endoscope, but only on the lesion characteristic region in the image collected by the endoscope. Since the lesion characteristic image of the in-vivo environment of the subject is collected by the visible light image acquisition module, the photosensitive region of the visible light image acquisition module is divided into multiple sub-regions, and the exposure weight value of each sub-region is set. Specifically, the sub-partition exposure weight value involved in the lesion characteristic range in the target position is set to be high, and the exposure weight value of the remaining sub-regions is set to be low, that is, the partition exposure of the lesion characteristic range in the target position can be realized. The more the photosensitive region of the visible light image acquisition module is divided in this embodiment, the better the partition exposure effect of the lesion characteristic range in the target position is obtained, but the more calculation resources are consumed by the division of the region. Therefore, the division of the photosensitive region of the visible light image acquisition module is determined according to actual needs. In addition to setting the exposure weight value, the exposure effect of the partition exposure can be improved by means such as light compensation, image gain, wide dynamic range, or the partition exposure can be replaced, and the clarity of the lesion characteristic region coordinate region is directly improved.

[0107] S6. Detect the lesion characteristic image obtained by the partition exposure, and identify the lesion in the image.

[0108] Step S6 mainly consists of the following steps:

[0109] S6-1. Set a training set composed of various types of lesion characteristic images, and mark these lesion characteristic images with respect to the lesion type;

[0110] S6-2. Input the training set into the neural network model for training to obtain a trained neural network model;

[0111] S6-3. Input the lesion characteristic image or exposure image into the trained neural network model to obtain the confidence degree of the lesion characteristic image or exposure image belonging to various types of lesions;

[0112] S6-4. Select the lesion type with the highest confidence degree as the identification result of the lesion in the lesion characteristic image or exposure image.

[0113] The recognition of the lesion in step S6 of the embodiment is mainly completed by a neural network model. Specifically, a training set composed of various types of lesion feature images is set, and the lesion feature images are labeled with respect to the lesion types; the training set is input into the neural network model for training, so that the neural network model has the ability to recognize the input image. In the embodiment, the neural network model can be implemented by a LeNet, AlexNet, VGG, GoogLeNet, ResNet or the like. Taking the Resnet model as an example, the trained Resnet model can recognize the input lesion feature image and output the confidence of the lesion feature image belonging to various lesion types, and the lesion type with the highest confidence is selected as the recognition result of the input lesion feature image.

[0114] Embodiment 2

[0115] The embodiment discloses a lesion detection auxiliary exposure device installed on an endoscope, mainly comprising the following modules:

[0116] The visible light image acquisition module is used for visible light exposure.

[0117] The global exposure module is used for controlling the visible light acquisition module to perform global exposure on the target position to obtain an exposure image.

[0118] The definition judgment module is used for judging the definition of the exposure image.

[0119] The infrared thermal image module is used for acquiring an infrared thermal image of the target position when the definition of the exposure image is less than a definition threshold; the infrared thermal image of the target position contains a plurality of thermal pixels.

[0120] The feature range demarcation module is used for demarcating a lesion feature range in the target position from the infrared thermal image of the target position.

[0121] The partition exposure module is used for controlling the visible light image acquisition module to perform partition exposure on the lesion feature range in the target position to obtain a lesion feature image.

[0122] The lesion recognition module is used for detecting the lesion feature image obtained by the partition exposure and recognizing the lesion in the image.

[0123] In the embodiment, the infrared thermal image module and the visible light image acquisition module are installed at the tip of the endoscope by default, and the global exposure module, the definition judgment module, the feature range demarcation module, the partition exposure module and the lesion recognition module are generally installed in the host computer of the endoscope.

[0124] The embodiments of the present application further disclose a computer program product or computer program, which comprises computer instructions stored in a computer readable storage medium. A processor of a computer device can read the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to enable the computer device to perform the method shown in the above. Figure 1

[0125] In some alternative embodiments, the functions / operations mentioned in the block diagrams can not occur in the order mentioned in the operation diagrams. For example, two blocks shown in succession can actually be executed substantially concurrently with each other, or the blocks can sometimes be executed in reverse order, depending on the functionality / operations involved. Furthermore, the embodiments presented and described in the flow diagrams of the present application are only examples. As such, the disclosed methods are not limited to the order of operations and logic flow presented in the present application. Alternative embodiments are possible where the order of operations is changed and where sub operations described as part of a larger operation are executed in a different order, or are executed concurrently with each other.

[0126] Furthermore, although the present application is described in the context of functional modules, it is to be understood that one or more of the functions and / or features described can be integrated in a single physical device and / or software module, or one or more functions and / or features can be implemented in separate physical devices or software modules. It is also to be understood that detailed discussion of the actual implementation of each module is unnecessary to an understanding of the present application. Rather, the actual implementation is to be understood in the context of the attributes, functions and internal relationships discussed in the context of the various functional modules disclosed herein. Accordingly, the skilled artisan will appreciate the implementation of the present application in light of the disclosure and descriptions herein without undue experimentation with the claims appended hereto. It is also to be understood that the particular conceptualization disclosed is illustrative only and not limiting of the scope of the present application, which is to be determined by the appended claims and their equivalents.

[0127] In the description of the present application, reference has been made to descriptive terms such as "one embodiment", "some embodiments", "an example", "a specific example" or "some examples" etc. Such terminology means that a particular feature, structure, material or characteristic being described is included in at least one embodiment or example of the application. The illustrative examples given are not necessarily the only way to implement the application. At the same time, the illustrative examples are not necessarily mutually exclusive, although every specific example is uniquely constituted. Furthermore, the particular features, structures, materials or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0128] ​While the embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary and that changes in form and detail can be made without departing from the principles and spirit of the application. The scope of the application is defined by the claims and their equivalents.

[0129] The above is a specific description of the preferred embodiment of the present application, but the present application is not limited to the described embodiment, and those skilled in the art can make various equivalent modifications or replacements without departing from the spirit of the present application, and these equivalent modifications or replacements are all included in the scope defined by the claims of the present application.

Claims

1. An auxiliary exposure device for lesion detection, installed on an endoscope, characterized in that, Includes the following modules: Visible light image acquisition module, the visible light image acquisition module being used for visible light exposure; A global exposure module is used to control the visible light acquisition module to perform global exposure on the target position to obtain an exposed image. A sharpness determination module is used to determine the sharpness of the exposed image; An infrared thermal imaging module is used to acquire an infrared thermal image of a target location when the sharpness of the exposed image is less than a sharpness threshold; the infrared thermal image of the target location contains a number of thermal pixels. A feature range delineation module is used to delineate the feature range of lesions in the target location from the infrared thermogram of the target location; A partitioned exposure module is used to control the visible light image acquisition module to perform partitioned exposure on the lesion feature range in the target location to obtain a lesion feature image; Specifically, the step of dividing the lesion feature range in the target location for exposure includes the following steps: The coordinates of the lesion feature range in the target location are transformed based on the coordinate transformation coefficients. The visible light image acquisition module is controlled to perform zoned exposure of the lesion feature range in the target location; The coordinate transformation coefficients are obtained through the following steps: Choose a calibration plane, which has at least three calibration points; Apply a heat source to the calibration point; The coordinates of the calibration points on the calibration plane are acquired using a visible light image acquisition module to obtain the first set of calibration point coordinates. The coordinates of the calibration points on the calibration plane are acquired using an infrared thermal imaging module to obtain the second set of calibration point coordinates. Based on the coordinate conversion relationship between the first and second calibration point coordinate sets, determine the coordinate transformation coefficients; The lesion identification module is used to detect lesion feature images obtained by zoned exposure and identify lesions in the images. When the sharpness of the exposed image is greater than or equal to a preset sharpness threshold, the lesion identification module performs the following steps: The exposed image obtained from the global exposure is detected, and lesions in the image are identified.

2. The lesion detection auxiliary exposure device according to claim 1, characterized in that, The step of delineating the lesion feature range in the target location from the infrared thermal image of the target location specifically includes the following steps: Obtain the coordinate values ​​of the thermal pixels in the infrared thermal image of the target location; Mark the coordinate values ​​of the hot pixels; By combining adjacent hot pixels with similar coordinate values, a lesion feature region is formed. Obtain the coordinates of the hot pixels at the edge of the lesion feature region to get the lesion feature range at the target location.

3. The lesion detection auxiliary exposure device according to claim 1, characterized in that, After the step of delineating the lesion feature range in the target location from the infrared thermal image of the target location, the following steps are also included: Determine whether the infrared thermal image of the target location has shifted. If it has shifted, then correct the range of lesion features in the target location.

4. The lesion detection auxiliary exposure device according to claim 3, characterized in that, The correction of the lesion feature range at the target location specifically includes the following steps: Acquire several infrared thermal images of the target location within a preset time period; Mark the lesion feature range in the target location in several infrared thermal images of the target location; Several infrared thermograms of the target location are overlaid to obtain incompletely overlaid coordinate regions of lesion feature areas. The edge contour of the incompletely superimposed lesion feature region coordinate area is taken as the corrected lesion feature region coordinate area.

5. The lesion detection auxiliary exposure device according to claim 1, characterized in that, The control of the visible light image acquisition module to perform zoned exposure of the lesion feature range in the target location specifically includes the following steps: The photosensitive area of ​​the visible light image acquisition module is divided into multiple sub-regions; Set the exposure weight value for each sub-region; The visible light image acquisition module is controlled to perform zoned exposure based on the weight value of each sub-region.

6. The lesion detection auxiliary exposure device according to claim 1, characterized in that, The identification of lesions in the image specifically includes the following steps: Set up a training set consisting of feature images of various types of lesions, and label these feature images with respect to the lesion type; The training set is input into the neural network model for training, resulting in a trained neural network model; The lesion feature image or exposure image is input into a trained neural network model to obtain the confidence level of the lesion feature image or exposure image belonging to various types of lesions; The lesion type with the highest confidence level is selected as the lesion identification result in the lesion feature image or exposure image.

Images