Intestinal tract imaging recognition method and device, electronic device and endoscope system

Abstract

The invention relates to an intestinal tract imaging recognition method and device, an electronic device and an endoscope system.The intestinal tract imaging recognition method comprises the steps that an intestinal tract area where a detection device is located at the first moment is recognized; when it is determined that the intestinal tract area where the detection device is located at the first moment is a target intestinal tract area, identifying a mark part in the target intestinal tract area; according to the identified mark part, determining a first intestinal segment where the detection device is located at the second moment; the first intestinal segment is one intestinal segment in the target intestinal region, and the second moment is later than the first moment. According to the method, the intestinal region where the detection device is located at one moment is determined by utilizing intestinal region identification with relatively low identification difficulty, and then secondary division of the intestinal region is performed on the basis of an identification part on the premise of identifying the intestinal region, so that more accurate intestinal segments of the detection device at the next moment are determined, and the accuracy of the detection device is improved. Therefore, the identification difficulty of the intestinal segment is reduced, and the identification accuracy of the intestinal segment is improved.

Description

Technical Field

[0001] The present application relates to the field of machine vision technology, and in particular to an intestinal imaging recognition method, device, electronic device and endoscope system. Background Art

[0002] At present, colonoscopy based on machine learning has made great progress. Image processing technology based on machine learning can be used to identify the parts of intestinal images. However, due to the high similarity in structure and morphology of some areas in the intestine, it is difficult to accurately distinguish the various parts of the intestine. In addition, due to the peristalsis of the intestine, the collected intestinal images cannot clearly describe the corresponding parts. Therefore, the accuracy of identifying the intestinal part to which the intestinal image belongs is currently low.

[0003] With regard to the problem in the related art of low accuracy in identifying the intestinal part to which the intestinal image belongs, no effective solution has been proposed yet. Summary of the invention

[0004] In this embodiment, a method, device, electronic device and endoscope system for intestinal imaging recognition are provided to solve the problem of low recognition accuracy of the intestinal part to which the intestinal image belongs in the related art.

[0005] In a first aspect, a method for intestinal imaging recognition is provided in this embodiment, comprising:

[0006] During the inspection of the target intestine, identifying the intestinal region where the detection device is located at a first moment;

[0007] When it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region, identifying a landmark portion in the target intestinal region; the target intestinal region includes at least two different intestinal segments;

[0008] The first intestinal segment where the detection device is located at the second moment is determined according to the identified landmark position; the first intestinal segment is an intestinal segment in the target intestinal region, and the second moment is later than the first moment.

[0009] In some embodiments, the method further comprises:

[0010] According to the first intestinal segment where the detection device is located at the second moment, the result identified as the target intestinal area before the second moment is updated from the target intestinal area to the second intestinal segment in the target intestinal area; the marking site is located between the first intestinal segment and the second intestinal segment.

[0011] In some embodiments, the method further comprises:

[0012] Identifying a low cleanliness area in the target intestine whose cleanliness is lower than a preset cleanliness threshold;

[0013] Cleanliness statistics are performed on all the identified low-cleanliness areas to obtain statistical results of intestinal cleanliness.

[0014] In some embodiments, the cleanliness statistics of all the identified low cleanliness areas are performed to obtain the statistical results of intestinal cleanliness, including:

[0015] Determine a target time period for identifying a target low cleanliness area; the target low cleanliness area is any of the low cleanliness areas;

[0016] Identify whether there is a cleaning event for the target low-cleanliness area within a preset period after the target period;

[0017] The target low cleanliness area where the cleaning event occurs is removed from the low cleanliness area, and cleanliness statistics are performed on the remaining low cleanliness areas to obtain statistical results of intestinal cleanliness.

[0018] In some embodiments, the identifying whether there is a cleaning event for the target low cleanliness area within a preset period after the target period includes:

[0019] If water absorption and / or flushing behavior for the target low cleanliness area is identified within a preset period after the target period, it is determined that a cleaning event for the target low cleanliness area exists within the preset period after the target period.

[0020] In some embodiments, if water absorption behavior and / or water flushing behavior for the target low cleanliness area is identified within a preset period after the target period, determining that a cleaning event for the target low cleanliness area exists within the preset period after the target period includes:

[0021] If water absorption behavior for the target low cleanliness area is identified within a preset time period after the target time period, and the duration of the water absorption behavior exceeds a first duration threshold, it is determined that a cleaning event for the target low cleanliness area exists within the preset time period after the target time period;

[0022] and / or,

[0023] If a flushing action for the target low cleanliness area is identified within a preset time period after the target time period, and the duration of the flushing action exceeds a second duration threshold, it is determined that a cleaning event for the target low cleanliness area occurs within the preset time period after the target time period.

[0024] In some embodiments, the method further comprises:

[0025] For the target low-cleanliness area where the cleaning event occurs, the intestinal area is identified, and the identification result of the intestinal area at the corresponding time in the target time period is updated.

[0026] In some embodiments, the method further comprises:

[0027] Identifying a target intestinal region to which the target low cleanliness region belongs, and an identified landmark site in the target intestinal region;

[0028] Based on the marked position, a third intestinal segment corresponding to the target time period is determined, and the third intestinal segment belongs to the target intestinal area.

[0029] In some embodiments, the method further comprises:

[0030] An abnormal intestinal region in the target intestine is identified.

[0031] In some embodiments, the method further comprises:

[0032] The quality analysis result of the detection by the detection device is generated based on the identification result of each intestinal segment in the target intestine, the statistical result of the intestinal cleanliness and the identification result of the abnormal intestinal area.

[0033] In some of the embodiments, the process of inspecting the target intestine belongs to the stage of withdrawing the scope of the detection device.

[0034] In a second aspect, in this embodiment, a device for intestinal imaging recognition is provided, comprising: a first recognition module, a second recognition module and a determination module; wherein:

[0035] The first identification module is used to identify the intestinal region where the detection device is located at a first moment during the process of inspecting the target intestinal tract;

[0036] The second identification module is used to identify the landmarks in the target intestinal region when it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region; the target intestinal region includes at least two different intestinal segments;

[0037] The determination module is used to determine the first intestinal segment where the detection device is located at a second moment according to the identified landmark position; the first intestinal segment is an intestinal segment in the target intestinal area, and the second moment is later than the first moment.

[0038] In a third aspect, an electronic device is provided in this embodiment, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the intestinal imaging recognition method described in the first aspect is implemented.

[0039] In a fourth aspect, an endoscope system is provided in this embodiment, comprising: a display, a light source device, an endoscope, and the electronic device as described in the first aspect above.

[0040] Compared with the related art, the intestinal imaging recognition method, device, electronic device and endoscope system provided in this embodiment, in the process of inspecting the target intestine, identify the intestinal region where the detection device is located at the first moment; when it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region, identify the landmark position in the target intestinal region; the target intestinal region includes at least two different intestinal segments; according to the identified landmark position, determine the first intestinal segment where the detection device is located at the second moment; the first intestinal segment is a type of intestinal segment in the target intestinal region, and the second moment is later than the first moment. It first uses the intestinal region recognition with relatively low recognition difficulty to determine the intestinal region where the detection device is located at one moment, and then, on the premise of identifying the intestinal region, performs a secondary division of the intestinal region based on the landmark position, thereby determining a more accurate intestinal segment of the detection device at the latter moment, thereby reducing the difficulty of intestinal segment recognition and improving the accuracy of intestinal segment recognition.

[0041] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0043] Figure 1 is a hardware structure block diagram of a terminal of the intestinal imaging recognition method of this embodiment;

[0044] Figure 2 is a flow chart of the intestinal imaging recognition method of this embodiment;

[0045] Figure 3 is a schematic diagram of the intestinal structure using the intestinal imaging recognition method of this embodiment;

[0046] Figure 4 is a flow chart of an intestinal imaging recognition method according to an embodiment of the present application;

[0047] Figure 5 4 is a structural block diagram of the intestinal imaging recognition device of this embodiment. DETAILED DESCRIPTION

[0048] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0049] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the", "these" and the like in this application do not represent quantitative restrictions, and they can be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether directly or indirectly. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. Usually, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0050] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar electronic device with certain computing capabilities. For example, running on a terminal, Figure 1 FIG. 4 is a hardware structure diagram of a terminal of the intestinal imaging recognition method of this embodiment. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown in the figure) processor 102 and memory 104 for storing data, wherein processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. The above terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is for illustration only and does not limit the structure of the above terminal. Figure 1More or fewer components as shown, or with Figure 1 Different configurations shown.

[0051] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the intestinal imaging recognition method in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0052] The transmission device 106 is used to receive or send data via a network. The above network includes a wireless network provided by the communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0053] In this embodiment, a method for intestinal imaging recognition is provided. Figure 2 is a flow chart of the intestinal imaging recognition method of this embodiment. Figure 2 As shown, the process includes the following steps:

[0054] Step S210, during the inspection of the target intestine, identifying the intestinal region where the detection device is located at the first moment.

[0055] The detection device may be an insertion portion of an endoscope used for intestinal examination, such as an insertion portion of a colonoscope. As the detection device moves in the target intestine, multiple frames of image information in the target intestine are acquired, and the target intestine is examined based on the image information in the target intestine. It should be noted that the above-mentioned first moment may be a moment when the target intestine is examined in real time, or may be an imaging moment corresponding to a certain frame of the intestinal image when reviewing several frames of intestinal images in historical data. Figure 3 FIG. 1 is a schematic diagram of the intestinal structure to which the intestinal imaging recognition method of this embodiment is applied. Figure 3As shown, there is a right flexure 302 of the colon between the ascending colon 301 and the transverse colon 303; there is a left flexure 304 of the colon between the transverse colon 303 and the descending colon-sigmoid colon 305; in addition, the intestine also includes the appendix, cecum, ileum, rectum and anal canal (not shown in the figure). In the process of inspecting the target intestine, the intestinal area where the detection device is located can be determined by identifying the image of the position of the detection device in the target intestine.

[0056] The above-mentioned intestinal region can be a tissue region composed of multiple intestinal segments that is pre-divided according to the intestinal structure in the target intestine, or it can be a region that contains a certain intestinal segment alone. Specifically, based on the structural similarity of the intestinal segments, the structural positional relationship, the similarity in image characteristics, or the difficulty of image recognition, multiple intestinal segments can be divided into one intestinal region, and the intestinal region containing multiple intestinal segments can be used as the target intestinal region. For example, the ascending colon and the transverse colon themselves have a high degree of structural similarity, which makes recognition more difficult and has a low accuracy. This embodiment takes into account that the ascending colon and the transverse colon are separated by the right flexure of the colon and are close in position, so the area from the ascending colon to the transverse colon is divided into a target intestinal region, and the target intestinal region includes two intestinal segments, the ascending colon and the transverse colon, and the right flexure of the colon between the ascending colon and the transverse colon.

[0057] Specifically, after the signal transmitted by the detection device is image processed to obtain the corresponding intestinal images of several frames, the intestinal region where the detection device is located in the target intestine can be identified based on any intestinal region recognition method. For example, traditional image recognition methods such as image segmentation and template matching, as well as machine learning and other methods are used to achieve intestinal region recognition. Optionally, the image data of several frames of the intestine obtained can be analyzed based on a well-trained intestinal recognition model. Among them, the intestinal recognition model includes but is not limited to convolutional neural networks such as residual networks ResNet, VGG16, AlexNet, and 3DResNet. When training the intestinal recognition model, the colonoscopy video used for training can be collected first, and the non-information segments representing intestinal flushing, intestinal water absorption, blurred intestinal images, and intestinal surgery can be eliminated, and the information segments left after the elimination are data labeled. For example, based on the experience and judgment of the annotators, the time starting and ending points of the terminal ileum, cecum, ascending colon-transverse colon, right flexure of the colon, transverse colon, left flexure of the colon, descending colon-sigmoid colon, rectum and anus appearing in the information segment screen are marked, and the corresponding labels are marked as "terminal ileum", "cecum", "ascending colon-transverse colon", "right flexure of the colon", "transverse colon", "left flexure of the colon", "descending colon-sigmoid colon", "rectum" and "anus". Furthermore, after multiple annotators have performed annotations according to the above examples, the intersection of the time periods marked by different annotators can be taken and then submitted to the reviewer for review to improve the quality of the annotations.

[0058] In particular, during the labeling process, for the labeling of a smaller part contained in a certain intestinal region, if the morphological features of the part in the image are more obvious, the corresponding image segment can be labeled as the label of the part, otherwise, the corresponding image segment is labeled as the label of the intestinal region of the part. For example, for the image segment corresponding to the right flexure of the colon, if the morphological features of the right flexure of the colon are more obvious, the label of the corresponding image segment is labeled as "right flexure of the colon", otherwise, the label of the corresponding image segment is labeled as "ascending colon-transverse colon" to which the right flexure of the colon belongs.

[0059] Afterwards, the labeled data is input into the preset model for training, and finally a fully trained intestinal recognition model is obtained. In the process of inspecting the target intestine, the fully trained intestinal recognition model can be used to identify the intestinal region where the detection device is located in the target intestine. For example, when the intestinal recognition model recognizes that there is a label of "terminal ileum" in the intestinal image at a certain moment, it can be confirmed that the intestinal region of the terminal ileum has been recognized. In order to further improve the accuracy of recognition, it can be determined that the detection device is in the intestinal region when the label of a certain intestinal region is recognized for a preset number of consecutive times. For example, if the "terminal ileum" label is recognized twice in succession, it is determined that the detection device has reached the terminal ileum in the target intestine; if the "ascending colon-transverse colon" label is recognized twice in succession, it is determined that the detection device is in the ascending colon-transverse colon region; if the "right flexure of colon" label is recognized twice in succession, or the "transverse colon" label is recognized twice in succession, it is determined that the detection device has reached the transverse colon in the target intestine, and the previously recognized part between the cecum and the right flexure of colon is updated to the ascending colon; if the "left flexure of colon" label is recognized twice in succession, or the "descending colon-sigmoid colon" label is recognized twice in succession, it is determined that the detection device has reached the descending colon-sigmoid colon region in the target intestine. The identification of other intestinal segments or intestinal regions is similar to the above process and will not be repeated here.

[0060] Step S220, when it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region, identifying the landmark position in the target intestinal region; the target intestinal region includes at least two different intestinal segments.

[0061] Wherein, when it is determined that the intestinal region where the detection device is located at the first moment is a target intestinal region containing multiple intestinal segments, the target intestinal region can be more finely identified to determine the specific intestinal segment in the target intestinal region. Wherein, in the target intestinal region, the physiological position of the marker part is located between different intestinal segments. Based on this, the marker part in the target intestinal region can be used as a boundary between different intestinal segments, that is, the marker part can play a demarcation role for the intestinal segments in the target intestinal region. In addition, the physiological characteristics of the marker part itself and the image characteristics reflected in the image are more obvious than those of the intestinal segments, so the recognition difficulty is lower. Therefore, by identifying the marker part that plays a demarcation role, compared with directly identifying the intestinal segments in the target intestinal region, the recognition accuracy can be improved and the recognition difficulty can be reduced, so that it can be specifically and accurately identified that the detection device is in which intestinal segment adjacent to the marker part, thereby realizing the recognition of more detailed intestinal segments in the target intestinal region.

[0062] Therefore, this embodiment identifies the landmark position in the target intestinal region to determine the specific intestinal segment where the detection device is located at a certain moment. For example, the landmark position can be the right flexure of the colon between the ascending colon and the transverse colon. Or it can be the left flexure of the colon between the transverse colon and the descending colon. Continuing to explain with the target intestinal region being the ascending colon-transverse colon region as an example. When it is determined that the intestinal region where the detection device is located at the first moment is the ascending colon-transverse colon region, it is determined whether the right flexure of the colon in the ascending colon-transverse colon region is identified at the first moment and thereafter.

[0063] Step S230, determining the first intestinal segment where the detection device is located at the second moment according to the identified landmark position; the first intestinal segment is an intestinal segment in the target intestinal area, and the second moment is later than the first moment.

[0064] Among them, according to the landmark position of the identified target intestinal area, the target intestinal area identified at different times can be divided into specific intestinal segments. The above-mentioned first intestinal segment is the intestinal segment reached after the detection device passes the landmark position according to the inspection order during the intestinal inspection. Therefore, this embodiment is based on the target intestinal area identified at the first moment, and after the landmark position in the target intestinal area is subsequently identified, the intestinal segment where the detection device is specifically located in the target intestinal area at a second moment later than the first moment can be determined. For example, it is identified that the target intestinal area where the detection device is located at the first moment is the ascending colon-transverse colon area, and the right bend of the colon in the ascending colon-transverse colon area is identified at the first moment or between the first moment and the second moment, and the detection device is still in the ascending colon-transverse colon area at the second moment, then it can be determined that the detection device is in the intestinal segment of the transverse colon at the second moment. In addition, the ascending colon-transverse colon area corresponding to the moment before the right bend of the colon is identified can also be updated to the ascending colon.

[0065] Compared with the related art, it is difficult to identify each intestinal segment directly based on image features because the location features of some intestinal segments in the intestine are not obvious enough, or it is difficult to collect stable images due to peristalsis. This embodiment divides the different intestinal segments in the target intestine, starts with the intestinal area with lower identification difficulty, and after determining that the detection device is in the target intestinal area containing multiple intestinal segments, it identifies the specific intestinal segment in the target intestinal area based on the relatively easy-to-identify landmarks in the target intestinal area, thereby achieving secondary accurate identification of the intestine, reducing the difficulty of identification and improving the accuracy of identification.

[0066] In the above steps S210 to S230, during the inspection of the target intestine, the intestinal region where the detection device is located at the first moment is identified; when it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region, the landmark position in the target intestinal region is identified; the target intestinal region includes at least two different intestinal segments; according to the identified landmark position, the first intestinal segment where the detection device is located at the second moment is determined; the first intestinal segment is a type of intestinal segment in the target intestinal region, and the second moment is later than the first moment. It first uses the intestinal region identification with relatively low identification difficulty to determine the intestinal region where the detection device is located at one moment, and then, on the premise of identifying the intestinal region, a secondary division of the intestinal region is performed based on the landmark position, thereby determining a more accurate intestinal segment of the detection device at the latter moment, thereby reducing the difficulty of intestinal segment identification and improving the accuracy of intestinal segment identification.

[0067] Optionally, in one embodiment, the intestinal imaging recognition method may further include:

[0068] According to the first intestinal segment where the detection device is located at the second moment, the result identified as the target intestinal area before the second moment is updated from the target intestinal area to the second intestinal segment in the target intestinal area; the marking part is located between the first intestinal segment and the second intestinal segment.

[0069] In this embodiment, the recognition result can also be updated to a more accurate and specific intestinal segment when the previous recognition result is the target intestinal area. For example, after determining that the intestinal segment where the detection device is located at the second moment is the transverse colon, the ascending colon-transverse colon area previously identified before the second moment, or before the landmark site is identified, is updated to the ascending colon. Therefore, this embodiment can continuously update the previous intestinal recognition results based on the advancement of the inspection process of the detection device in the intestine, thereby improving the accuracy of the final intestinal recognition results.

[0070] In addition, in one embodiment, the intestinal imaging recognition method may further include:

[0071] Identify low-cleanliness areas in the target intestine whose cleanliness is lower than a preset cleanliness threshold; perform cleanliness statistics on all identified low-cleanliness areas to obtain statistical results of intestinal cleanliness.

[0072] The cleanliness of the intestine will affect the credibility and accuracy of the results of the intestinal inspection. Therefore, in the intestinal imaging recognition, this embodiment also provides an analysis basis for the quality of the intestinal inspection by identifying and counting the cleanliness of the target intestine. Among them, the intestinal image corresponding to the low cleanliness area will be different from the intestinal images corresponding to other intestinal areas in terms of image features. Therefore, by performing image detection on the intestinal image, the cleanliness of the corresponding intestinal area is determined, and the low cleanliness area is obtained. Among them, the low cleanliness area in the target intestine can be identified based on traditional image processing methods such as image feature recognition and threshold segmentation, and the low cleanliness area of ​​the target intestine can also be identified by machine learning.

[0073] Exemplarily, the cleanliness recognition model is used to identify the intestinal area whose cleanliness is lower than a preset cleanliness threshold in the target intestine, and finally form the statistical result of the cleanliness of the target intestine. The statistical result of the cleanliness can be used for colonoscopy quality control. The cleanliness recognition process can be carried out in parallel with the intestinal area recognition process in the above steps S210 to 230, and the cleanliness recognition process and the above intestinal area recognition process can be carried out at the stage of withdrawing the scope during the intestinal examination.

[0074] Specifically, a well-trained cleanliness recognition model or other cleanliness recognition algorithms can be used to recognize intestinal images of a preset number of frames, and determine whether the cleanliness is qualified according to a preset cleanliness threshold. As an embodiment of the present application, the withdrawal speed of the low-cleanliness area with unqualified cleanliness is calculated, so as to calculate the average withdrawal speed of the detection device within the above-mentioned unqualified frame number. Afterwards, based on the above-mentioned number of frames with unqualified cleanliness and the average withdrawal speed, the length of the intestinal cleanliness that is unqualified is calculated. For example, if the intestinal cleanliness is unqualified for t frames, and the average withdrawal speed within the t frames is v, then the length of the intestinal cleanliness that is unqualified is L=t×v. Afterwards, the length of the intestinal cleanliness that is unqualified in different intestinal segments is statistically analyzed to obtain the statistical results of intestinal cleanliness.

[0075] As an embodiment of the present application, the above-mentioned cleanliness recognition model can be implemented based on 3D convolutional neural networks such as 3DResNet. In the training stage, the non-information fragments in the collected colonoscopy video are eliminated. Afterwards, the image screen in the information fragment left is annotated by the annotation personnel for cleanliness. Specifically, based on the actual situation, the cleanliness can be annotated as 0 points, 1 point, 2 points and 3 points, etc., and the time starting point and end point of the above-mentioned 0 points, 1 point, 2 points and 3 points appearing in the picture are marked. The above-mentioned scoring criteria are based on the Boston scoring (Boston Bowel Preparation Scale, referred to as BBPS). Afterwards, data preprocessing such as data desensitization, cutting and standardization are performed based on the annotated data, and then by dividing the data set into a training part and a test part, it is input into a 3D convolutional neural network for training, and finally a well-trained cleanliness recognition model is obtained.

[0076] In one embodiment, cleanliness statistics are performed on all identified low cleanliness areas to obtain intestinal cleanliness statistics, which may specifically include:

[0077] Determine a target time period for identifying a target low cleanliness area; the target low cleanliness area is any low cleanliness area; identify whether there is a cleaning event for the target low cleanliness area within a preset time period after the target time period; remove the target low cleanliness area with a cleaning event from the low cleanliness area, and perform cleanliness statistics on the remaining low cleanliness areas to obtain statistical results of intestinal cleanliness.

[0078] That is, in this embodiment, for the low cleanliness area, it is necessary to determine whether an intestinal cleaning event has occurred after the cleanliness identification of the intestinal area is completed. For example, the low cleanliness area is cleaned by the inspector. Among them, considering that in actual applications, the inspector needs to flush the intestine and suck out excess water when cleaning, so the above-mentioned cleaning event can be identified by the "water absorption" and "flushing" behavior labels in the non-information segment (that is, identifying whether water absorption behavior and flushing behavior have occurred). If a non-information segment follows the low cleanliness area, and the non-information segment contains one of the water absorption behavior and the flushing behavior, or contains both the water absorption behavior and the flushing behavior, it is determined that a cleaning event has occurred in the intestinal area after the cleanliness identification of the intestinal area.

[0079] In one embodiment, identifying whether there is a cleaning event for the target low cleanliness area within a preset period after the target period may specifically include:

[0080] If water absorption and / or flushing behavior for the target low cleanliness area is identified within a preset period after the target period, it is determined that a cleaning event for the target low cleanliness area exists within the preset period after the target period.

[0081] This embodiment takes into account that in actual applications, the above-mentioned cleaning events are more likely to occur in areas with low intestinal cleanliness. Therefore, for the identified target low cleanliness area, in the preset period adjacent to the corresponding target period, based on the recognition of water absorption behavior and flushing behavior, it is determined whether the target low cleanliness area has a cleaning event. Among them, the water absorption behavior and flushing behavior can be identified by performing behavior recognition on several frames of intestinal images of the target intestine. In the process of real-time inspection of the target intestine, water absorption behavior recognition and flushing behavior recognition can be performed after the target low cleanliness area is identified; or when reviewing the historical data of the target intestine, for a low cleanliness area in a certain period, if there is a non-information segment that is removed after the period, it is confirmed from the corresponding non-information segment whether there is an identified water absorption behavior and flushing behavior. It can be understood that the period in which the cleaning event of the target low cleanliness area occurs needs to be later than the period in which the target low cleanliness area is identified.

[0082] As an embodiment of the present application, considering that in actual applications, for other purposes such as cleaning the lens, short-term water absorption or flushing behavior may also occur, but the water absorption or flushing behavior in these cases is often relatively short. Therefore, in order to improve the recognition accuracy of intestinal cleaning events, in an embodiment of the present application, if water absorption behavior for a target low-cleanliness area is identified within a preset period after the target period, and the duration of the water absorption behavior exceeds a first duration threshold, it is determined that there is a cleaning event for the target low-cleanliness area within the preset period after the target period; and / or, if flushing behavior for a target low-cleanliness area is identified within a preset period after the target period, and the duration of the flushing behavior exceeds a second duration threshold, it is determined that there is a cleaning event for the target low-cleanliness area within the preset period after the target period.

[0083] Among them, the duration of the water absorption behavior can correspond to the number of frames of the intestinal image containing the water absorption behavior. Similarly, the duration of the flushing behavior can correspond to the number of frames of the intestinal image containing the flushing behavior. That is, within the preset number of frames after the image frame of the target low cleanliness area is identified, if the number of frames of the intestinal image in which the water absorption behavior is identified exceeds the preset first frame number threshold, and / or the number of frames of the intestinal image in which the flushing behavior is identified exceeds the preset second frame number threshold, it can be determined that a cleaning event has occurred in the target low cleanliness area. The above-mentioned first frame number threshold and the second frame number threshold can be the same or different, and the above-mentioned first duration threshold and the second duration threshold can also be the same or different. The specific setting method can be based on the needs of the actual application scenario.

[0084] The intestinal area where a cleaning event occurred will reduce the credibility of the cleanliness identification of the intestinal area. Based on this, in order to improve the accuracy of the final statistical results of intestinal cleanliness, this embodiment will not include the low cleanliness area where a cleaning event occurred after the cleanliness identification is completed in the statistics of intestinal cleanliness. Therefore, this embodiment removes the target low cleanliness area with a cleaning event from the low cleanliness area, and performs cleanliness statistics on the remaining low cleanliness areas to improve the accuracy of cleanliness statistics.

[0085] In one embodiment, the intestinal imaging recognition method may further include:

[0086] For the target low-cleanliness area where a cleaning event occurs, the intestinal area is identified, and the identification result of the intestinal area at the corresponding time in the target time period is updated.

[0087] This embodiment uses the degree of cleanliness as one of the factors affecting the accuracy of intestinal area recognition. After determining that a cleaning event has occurred in a low-cleanliness area, the recognition of the intestinal area of ​​the image segment of the corresponding time period will be triggered again. Therefore, in this embodiment, after determining that a cleaning event has occurred in the target low-cleanliness area of ​​the target time period, the target intestinal area to which the target low-cleanliness area belongs and the identified landmarks in the target intestinal area can also be identified; based on the landmarks, the third intestinal segment corresponding to the target time period is determined, and the third intestinal segment belongs to the target intestinal area.

[0088] For example, if the time period of the low cleanliness area is identified as time period 1, and a non-information segment containing "water absorption" and "flushing" is immediately following time period 1, the intestinal area is identified for the image segment corresponding to time period 1. Specifically, after determining the target intestinal area corresponding to time period 1, and the marking part therein, the third intestinal segment corresponding to time period 1 is determined based on the marking part. If there is a previous recognition result corresponding to time period 1, and the third intestinal segment identified after the cleaning event occurs is inconsistent with the previous recognition result, for example, the previous recognition result indicates that the corresponding time period 1 is another intestinal segment in the target intestinal area, then the intestinal segment corresponding to time period 1 is updated based on the identified third intestinal segment. Therefore, this embodiment can further improve the accuracy of the intestinal recognition results based on the recognition of cleanliness.

[0089] In addition, in one embodiment, the intestinal imaging recognition method may further include:

[0090] Identify abnormal intestinal areas in the target intestine. Among them, the intestinal area with abnormal morphological structure in the target intestine, the corresponding intestinal image will also be different from the normal intestinal area in image features. Based on this, in the process of inspecting the target intestine, in addition to identifying the intestinal area of ​​the target intestine and the intestinal cleanliness, the intestinal image of the target intestine can also be directly analyzed based on traditional image processing methods or machine learning technology to identify the abnormal intestinal area in the target intestine. Exemplarily, the abnormal intestinal area in the target intestine can be identified based on a well-trained intestinal abnormality recognition model. The intestinal abnormality recognition model can be used to identify hyperplastic polyp areas and adenomatous polyp areas in the intestine, and the above-mentioned abnormal intestinal area can be an intestinal area containing the above-mentioned polyps. Specifically, the above-mentioned intestinal abnormality recognition model can be implemented based on neural networks such as R-CNN, Fast R-CNN, YOLOv3, etc. for target detection.

[0091] In the training phase of the intestinal abnormality recognition model, the collected images of abnormal intestinal areas are used as data sets, and data annotation is performed based on the corresponding specific abnormal conditions. For example, for data such as "single hyperplastic polyp" and "single adenomatous polyp", two annotators can use their subjective judgment to annotate the polyp area in the image with a rectangular frame and label it accordingly, and then one auditor will review the annotation results to improve the annotation quality. For data such as "multiple polyps", two auditors with higher qualifications will annotate each polyp area in the image with a rectangular frame, and label it with "hyperplastic polyp", "adenomatous polyp" and other labels, and then determine whether the annotation results of the two senior doctors are the same. If so, they are included in the data set, thereby improving the annotation quality. After that, data preprocessing such as data desensitization, cutting black frames and data standardization are performed, and the processed data set is divided into a training part and a test part, and the training part is further divided into a training set and a validation set. The above neural network is trained to finally obtain a fully trained intestinal abnormality recognition model.

[0092] It should also be noted that the above-mentioned intestinal abnormality identification can also be performed in parallel with the above-mentioned intestinal cleanliness identification and intestinal area identification, and both are performed during the intestinal examination retraction stage. Therefore, this embodiment can simultaneously identify the intestinal area, intestinal cleanliness, and abnormal intestinal area of ​​the target intestine, thereby speeding up the intestinal examination process, improving the efficiency of intestinal examination, and providing more comprehensive data basis for subsequent intestinal examination quality analysis.

[0093] In one embodiment, the intestinal imaging recognition method may further include:

[0094] The quality analysis results of the detection device detection are generated based on the identification results of each intestinal segment in the target intestine, the statistical results of intestinal cleanliness, and the identification results of abnormal intestinal areas. For example, the intestinal length and corresponding intestinal part of the intestinal cleanliness that are unqualified can be stated in the quality analysis results, and suggestions on whether to re-examine can be given. In addition, the observation degree recognition of intestinal inspection can be combined with the observation degree recognition model that has been trained well to identify the observation degree from the beginning of intestinal inspection to the arrival of the detection device outside the body, and record the duration of qualified observation degree and the corresponding intestinal segment. In addition, when the intestinal cleanliness is low, the inspector can be prompted to flush and absorb water from the target intestine, and monitor whether it is cleaned. When the amount of information on the screen is insufficient, the inspector is prompted to observe the target intestine carefully. When the speed of the picture changes too fast, the inspector is prompted to stabilize the mirror control, and monitor the amount of information and the picture speed to measure whether the inspector optimizes the diagnosis based on the prompts.

[0095] Finally, the quality analysis results of the detection device can be issued and prompts can be given. For example, "Transverse colon, poor intestinal cleanliness, cannot be cleaned, intestinal examination quality is low, polyp and adenoma recognition performance is low, re-examination is recommended"; or "Transverse colon, poor intestinal cleanliness, after cleaning, the cleanliness meets the requirements, and the intestinal examination quality is qualified"; or "Transverse colon, complex intestinal environment, low intestinal examination quality, low polyp and adenoma recognition performance, re-examination is recommended" and so on.

[0096] In one embodiment, based on the above-mentioned intestinal imaging recognition method, the process of inspecting the target intestine belongs to the withdrawal stage of the detection device. Specifically, after confirming that the detection device is in the withdrawal stage, the target intestine is started to be identified as the intestinal segment where the detection device is located, the intestinal cleanliness is identified, and intestinal abnormalities are identified. Among them, after confirming that the detection device has entered the target intestine, it can be determined whether it has entered the withdrawal stage by judging whether the detection device has reached a position where the detection device can be withdrawn; or it can be determined whether the detection device is in the withdrawal stage by performing image movement recognition. In the above-mentioned recognition of the withdrawal position and image movement recognition, as long as one of the recognition results indicates that the detection device is in the withdrawal stage, it can be confirmed that the detection device has reached the withdrawal stage.

[0097] Furthermore, in one embodiment, the intestinal imaging recognition method further includes:

[0098] After confirming that the detection device has been extended into the target intestine, it is determined whether the detection device has reached the preset scope withdrawal mark position in the target intestine; if so, it is confirmed that the detection device is in the scope withdrawal stage. Specifically, the scope withdrawal mark position can be preset. The intestinal images of several frames are subjected to part recognition by any image processing method to determine whether the scope withdrawal mark position is recognized, and if so, it is confirmed that the detection device is in the scope withdrawal stage. As an embodiment of the present application, when it is confirmed that the detection device has reached the intestinal segments such as the ileocecal valve, the appendix orifice, and the terminal ileum based on a well-trained ileocecal recognition model, it can be confirmed that the detection device is in the scope withdrawal stage. The ileocecal recognition model can be specifically implemented by neural networks such as ResNet, VGG16, AlexNet, and 3DResNet. During the training process, the time start and end points of "ileocecal valve", "appendix orifice", and "terminal ileum" appear in the information fragment screen marked by different labelers, and the corresponding labels are marked. Afterwards, the time periods marked by different labelers are intersected, and then reviewed by the reviewer to obtain the labeled data set. Finally, the data is preprocessed and divided into training and testing parts, and finally input into the above neural network model for training to obtain a fully trained ileocecal recognition model.

[0099] In one embodiment, the intestinal imaging recognition method may further include:

[0100] According to the movement information of the detection device after it is extended into the target intestine, it is judged whether the detection device is in the retracting stage. Specifically, the image movement state of several frames of intestinal images is identified based on image processing to determine whether the detection device is in the retracting stage. Exemplarily, based on a well-trained image movement recognition model, information fragments are recognized, and several frames of images are recognized each time, for example, 72 frames of images are recognized each time, to determine whether the detection device is in the advancing, retracting or stopping state, and to obtain a recognition result. Through sliding window analysis of 10 recognition results, if the "retracting" label exceeds the threshold, it is confirmed that the detection device is in the retracting stage.

[0101] Similarly, the above-mentioned picture motion recognition model can be implemented by 3D convolutional neural networks such as 3DResNet. During the training stage, different labelers mark the start and end points of the time when the camera enters, exits, and stops in the information fragment, and label them as "camera enter", "exit", and "stop" respectively. The time periods marked by different labelers are intersected to improve the quality of the annotation, and then the reviewer will review it. Finally, the remaining video part of the information fragment that is not labeled is labeled "other". After that, the labeled data is preprocessed by data desensitization, black frame cropping, and data standardization, and then the data is divided into training part and test part, and the training part is further divided into training set and verification set. Finally, the above-mentioned neural network is trained to obtain a fully trained picture motion recognition model.

[0102] In addition, it is also possible to first perform in vivo and in vitro recognition based on the multi-frame image data transmitted by the detection device, and analyze frame by frame whether the corresponding image is an in vivo image or an in vitro image. If in vivo images are recognized in multiple consecutive frames, and the number of consecutive frames exceeds a preset threshold, it is determined that the detection device has reached the body. After determining that the detection device has reached the body, a well-trained non-information frame screening model can be used to screen out non-information frames in multiple frames of intestinal images. If the number of consecutive non-information frames that are identified exceeds the preset threshold, the consecutive non-information frames are treated as non-information segments, and the screened multiple frames are treated as information segments. The screened information segments are used as processing objects for subsequent intestinal imaging recognition, while the screened non-information segments are not used as objects for subsequent intestinal imaging recognition processing.

[0103] Among them, the recognition of non-information frames can be performed by image processing to perform behavior recognition, image blur recognition, etc. on the acquired intestinal images, and intestinal images identified as blurred images, intestinal images containing surgical behaviors, intestinal images containing water absorption behaviors, and intestinal images containing flushing behaviors are screened out, and the above-screened intestinal images are regarded as non-information frames. Optionally, the non-information frame screening model can be implemented by two-dimensional convolutional neural networks such as ResNet, VGG16, and AlexNet. The training process requires manual labeling of the corresponding image frames of blur, surgery, water absorption, flushing, and information frames, and the image frames containing the labels of "blur", "surgery", "water absorption", and "flushing" are all non-information frames. After that, the above-mentioned convolutional neural network is trained with the labeled data set, and the training method is similar to the training method of other models described previously, which will not be repeated here.

[0104] In addition, after completing the above-mentioned intestinal identification, intestinal cleanliness identification and intestinal abnormality identification, the image transmitted by the detection device can also be analyzed frame by frame through in-vivo and out-vivo identification to determine whether it is an in-vivo image or an out-vivo image. When multiple consecutive frames identify an out-vivo image, it is determined that the detection device has reached the outside of the body. Based on the above-mentioned identification results of each intestinal segment, the statistical results of intestinal cleanliness and the identification results of abnormal intestinal areas, the quality analysis results of the above-mentioned intestinal inspection are generated.

[0105] Figure 4 FIG. 1 is a flow chart of an intestinal imaging recognition method according to an embodiment of the present application. Figure 4 As shown, the intestinal imaging recognition method includes the following steps:

[0106] Step S401, obtaining a colonoscopy video; wherein the colonoscopy video is obtained by performing image processing on a signal transmitted by a detection device;

[0107] Step S402, analyzing the colonoscopy video frame by frame to confirm whether the detection device has reached the body; if so, executing step S403;

[0108] Step S403, using a well-trained non-information frame screening model to identify information segments and non-information segments in the colonoscopy video;

[0109] Step S404, performing a de-focusing stage identification on the information fragment, if it is determined that the detection device is in the de-focusing stage, then executing steps S405, S406 and S407 in parallel;

[0110] Step S405, identifying the intestinal segment where the detection device is located;

[0111] Wherein, if the identification detection device is in a target intestinal region including multiple intestinal segments at a first moment, a landmark portion in the target intestinal region is identified; based on the identified landmark portion, a first intestinal segment where the detection device is located at a second moment is determined; the first intestinal segment belongs to the target intestinal region, and the second moment is later than the first moment; then, based on the first intestinal segment where the detection device is located at the second moment, the result identified before the second moment is updated from the target intestinal region to the second intestinal segment in the target intestinal region; the landmark portion is located between the first intestinal segment and the second intestinal segment;

[0112] Step S406, performing intestinal cleanliness identification based on the well-trained cleanliness identification model;

[0113] Among them, after identifying the low cleanliness area with unqualified cleanliness, determine the target time period corresponding to the low cleanliness area, and judge whether there is a clear event for the low cleanliness area only in the preset time period immediately adjacent to the target time period; if there is, the cleanliness recognition result of the low cleanliness area in the target time period is not included in the cleanliness statistics; in addition, for the target time period corresponding to the low cleanliness area where the cleaning event occurred, re-recognize the intestinal segment of the image in the target time period;

[0114] Step S407, identifying abnormal intestinal areas based on the well-trained intestinal abnormality recognition model;

[0115] Step S408, based on the in-vivo and out-vivo identification, determine whether the detection device has reached the outside of the body; if so, execute step S409;

[0116] Step S409: issuing a quality analysis result detected by the detection device.

[0117] In this embodiment, an intestinal imaging recognition device is also provided, which is used to implement the above-mentioned embodiments and preferred embodiments, and will not be repeated hereafter. The terms "module", "unit", "subunit", etc. used below can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0118] Figure 5 is a structural block diagram of the intestinal imaging recognition device 50 of this embodiment, as shown in Figure 5As shown, the intestinal imaging recognition device 50 includes: a first recognition module 52, a second recognition module 54 and a determination module 56; wherein: the first recognition module 52 is used to identify the intestinal region where the detection device is located at the first moment during the inspection of the target intestine; the second recognition module 54 is used to identify the landmark position in the target intestinal region when it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region; the target intestinal region includes at least two different intestinal segments; the determination module 56 is used to determine the first intestinal segment where the detection device is located at the second moment according to the identified landmark position; the first intestinal segment is a type of intestinal segment in the target intestinal region, and the second moment is later than the first moment.

[0119] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0120] In this embodiment, an electronic device is also provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any of the above method embodiments. In some embodiments, the electronic device may be an image processing device, an artificial intelligence (AI) box, or other device. For example, the electronic device may be connected to a detection device, and may process the signal detected by the detection device in the target intestine into a corresponding intestinal image, and execute the intestinal imaging recognition method based on the computer program set therein.

[0121] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0122] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0123] S1, during the inspection of the target intestine, identifying the intestinal region where the detection device is located at a first moment;

[0124] S2, when it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region, identifying a landmark portion in the target intestinal region; the target intestinal region includes at least two different intestinal segments;

[0125] S3, determining the first intestinal segment where the detection device is located at the second moment according to the identified landmark position; the first intestinal segment is an intestinal segment in the target intestinal area, and the second moment is later than the first moment.

[0126] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0127] In addition, in combination with the intestinal imaging recognition method provided in the above embodiment, an endoscope system may also be provided in this embodiment to implement the method. The endoscope system includes a display, a light source device, an endoscope and the above electronic device.

[0128] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0129] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0130] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

[0131] The term "embodiment" in this application refers to a specific feature, structure or characteristic described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is clearly or implicitly understood by those of ordinary skill in the art that the embodiments described in this application can be combined with other embodiments without conflict.

[0132] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A method for intestinal imaging recognition, characterized in that: include: During the inspection of the target intestine, identifying the intestinal region where the detection device is located at a first moment; When it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region, identifying a landmark portion in the target intestinal region; the target intestinal region includes at least two different intestinal segments; The first intestinal segment where the detection device is located at the second moment is determined according to the identified landmark position; the first intestinal segment is an intestinal segment in the target intestinal region, and the second moment is later than the first moment.

2. The intestinal imaging recognition method according to claim 1, characterized in that: The method further comprises: According to the first intestinal segment where the detection device is located at the second moment, the result identified as the target intestinal area before the second moment is updated from the target intestinal area to the second intestinal segment in the target intestinal area; the marking site is located between the first intestinal segment and the second intestinal segment.

3. The intestinal imaging recognition method according to claim 1 or 2, characterized in that: The method further comprises: Identifying a low cleanliness area in the target intestine whose cleanliness is lower than a preset cleanliness threshold; Cleanliness statistics are performed on all the identified low-cleanliness areas to obtain statistical results of intestinal cleanliness.

4. The intestinal imaging recognition method according to claim 3, characterized in that: The cleanliness statistics of all the identified low cleanliness areas are performed to obtain the statistical results of intestinal cleanliness, including: Determine a target time period for identifying a target low cleanliness area; the target low cleanliness area is any of the low cleanliness areas; Identify whether there is a cleaning event for the target low-cleanliness area within a preset period after the target period; The target low cleanliness area where the cleaning event occurs is removed from the low cleanliness area, and cleanliness statistics are performed on the remaining low cleanliness areas to obtain statistical results of intestinal cleanliness.

5. The intestinal imaging recognition method according to claim 4, characterized in that: The identifying whether there is a cleaning event for the target low cleanliness area within a preset period after the target period includes: If water absorption and / or flushing behavior for the target low cleanliness area is identified within a preset period after the target period, it is determined that a cleaning event for the target low cleanliness area exists within the preset period after the target period.

6. The intestinal imaging recognition method according to claim 4, characterized in that: The method further comprises: For the target low-cleanliness area where the cleaning event occurs, the intestinal area is identified, and the identification result of the intestinal area at the corresponding time in the target time period is updated.

7. The intestinal imaging recognition method according to claim 3, characterized in that: The method further comprises: An abnormal intestinal region in the target intestine is identified.

8. The intestinal imaging recognition method according to claim 7, characterized in that: The method further comprises: The quality analysis result of the detection by the detection device is generated based on the identification result of each intestinal segment in the target intestine, the statistical result of the intestinal cleanliness and the identification result of the abnormal intestinal area.

9. The intestinal imaging recognition method according to any one of claims 1 or 2, characterized in that: The process of inspecting the target intestine belongs to the stage of withdrawing the scope of the detection device.

10. An intestinal imaging recognition device, characterized in that: include: A first identification module, a second identification module and a determination module; wherein: The first identification module is used to identify the intestinal region where the detection device is located at a first moment during the process of inspecting the target intestinal tract; The second identification module is used to identify the landmarks in the target intestinal region when it is determined that the intestinal region where the detection device is located at the first moment is the target intestinal region; the target intestinal region includes at least two different intestinal segments; The determination module is used to determine the first intestinal segment where the detection device is located at a second moment according to the identified landmark position; the first intestinal segment is an intestinal segment in the target intestinal area, and the second moment is later than the first moment.

11. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the intestinal imaging identification method according to any one of claims 1 to 9.

12. An endoscope system, comprising: A display, a light source device, an endoscope, and the electronic apparatus as claimed in claim 11.

Images