A method and system for intelligent monitoring of standard operation of a radiotherapy process

Abstract

The application provides a kind of radiological medical process normative operation intelligent monitoring method and system, wherein the method comprises: acquiring the video stream of radiological medical scene in real time by image acquisition device;Radiological medical scene is divided into detection equipment area, patient activity area and medical activity area;The reference image in the state of no one in each area is calculated with real-time image numerical difference, and the pixel value change curve representing motion state is generated;According to the fluctuation characteristics of pixel value change curve, dynamically judge whether patient enters or leaves patient activity area, doctor enters or leaves medical activity area, and patient starts or ends the examination behavior in detection equipment area;Based on the behavior judgment result, the frame image is obtained by adaptively adjusting the frame extraction interval of real-time image;According to frame image, it is judged whether there is illegal behavior in radiological medical process.The monitoring method provided by the application optimizes the allocation of computing power, ensures the monitoring accuracy while reducing the consumption of computing resources.

Description

Technical Field

[0001] This disclosure generally relates to the field of radiomedical technology, and specifically to an intelligent monitoring method and system for standardized operation of radiomedical procedures. Background Technology

[0002] During radiological examinations, doctors must ensure that patients and their caregivers receive the necessary protective measures to minimize radiation exposure. However, some patients still experience insufficient protection during examinations, necessitating supervision of doctors' procedural compliance. Due to the widespread distribution and unique nature of radiological departments, relying solely on manual inspections is insufficient to meet practical needs. With the rapid development of intelligent detection and information technology, image recognition-based monitoring of healthcare worker behavior has become an important research direction in this field.

[0003] Currently, most image recognition monitoring methods adopt frame-by-frame recognition for response identification, which is a huge test of the hardware's computing power. Furthermore, continuous calculation will waste the hardware's computing resources. Moreover, sending the video stream to a remote server for processing is not conducive to real-time monitoring. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the existing technology, it is desirable to provide an intelligent monitoring method and system for standardized operation of radiotherapy procedures to solve the above problems.

[0005] The first aspect of this application provides an intelligent monitoring method for standardized operation of radiological medical procedures, including: Real-time video streams of radiotherapy scenarios are acquired using image acquisition equipment; The radiotherapy scene is divided into a detection equipment area, a patient activity area, and a doctor activity area. Numerical difference calculations are performed on the baseline image and real-time image in each area under unmanned conditions to generate pixel value change curves representing motion state; Based on the fluctuation characteristics of the pixel value change curve, the system dynamically determines whether the patient enters or leaves the patient activity area, whether the doctor enters or leaves the doctor activity area, and whether the patient begins or ends the examination in the detection device area. Based on the behavior judgment result, the frame extraction interval of the real-time image is adaptively adjusted to obtain the frame image; The frame images are used to determine whether any violations occurred during the radiotherapy process.

[0006] According to the technical solution provided in the embodiments of this application, the step of calculating the numerical difference between the reference image and the real-time image in each area under unmanned conditions includes: Convert the RGB pixel values ​​of the real-time image into grayscale pixel values; The pixel values ​​of the real-time image are compared with the pixel values ​​of the corresponding reference image to generate grayscale change curves for the patient's activity area, the detection device area, and the doctor's activity area.

[0007] According to the technical solution provided in the embodiments of this application, the dynamic judgment includes: When the continuous fluctuation of the grayscale change curve of the patient's activity area exceeds the first preset fluctuation threshold, it is determined that there are people in the patient's activity area and that the people are in an active state. When the continuous fluctuation of the grayscale change curve of the doctor's activity area exceeds the first preset fluctuation threshold, it is determined that there are people in the doctor's activity area and that the people are in an active state. When the continuous fluctuation of the grayscale change curve of the detection equipment area exceeds the first preset fluctuation threshold, it is determined that there are people in the detection equipment area and that the people are in an active state. When the continuous fluctuation of the grayscale change curve of each area exceeds the second preset fluctuation threshold but is less than the first preset fluctuation threshold, it is determined that there are people and scene fluctuations in that area.

[0008] According to the technical solution provided in the embodiments of this application, the adaptive adjustment of the frame extraction interval of the real-time image includes: When it is determined that there are people in the patient's activity area and the doctor's activity area and the people are in an active state, and there are no people in the detection equipment area, the frame extraction interval is set to a first duration. When it is determined that there are no people in the patient's activity area and the doctor's activity area, but there are people in the detection equipment area and they are active, the frame extraction interval is set to the second duration. When it is determined that there are no people in the patient's activity area, there are people in the detection equipment area and there are scene fluctuations, or there are people in the doctor's activity area and the people are in an active state, the frame extraction interval is set to the third duration. When it is determined that there are no people in the patient's activity area, there are people in the detection equipment area and there are scene fluctuations, and there are no people in the doctor's activity area, the frame extraction interval is set to the fourth duration.

[0009] According to the technical solution provided in the embodiments of this application, the adaptive adjustment of the frame extraction interval of the real-time image includes: Based on the dynamic judgment results of each region, a corresponding identification label is generated for each region; the identification label includes: whether there is continuous fluctuation in the grayscale change curve of each region, and whether the continuous fluctuation satisfies that it is less than or equal to a first preset fluctuation threshold and greater than a second preset fluctuation threshold. A combination of identification tags is established based on the identification tags corresponding to each region. The frame extraction strategy mapping table is then queried based on the identification tag combination to determine the corresponding frame extraction interval. The frame extraction strategy mapping table includes the correspondence between different identification tag combinations and frame extraction intervals.

[0010] According to the technical solution provided in the embodiments of this application, the step of determining whether there is any violation during radiotherapy based on the frame image includes: The bounding box of the identified object is marked, and the identified object includes at least detection equipment, doctors, patients and radiation protection devices; Calculate the distance between the bounding boxes corresponding to each of the identified objects in the frame image; When the distance between the patient and the boundary frame corresponding to the radiation protection device is greater than a preset distance threshold, it is determined that there is a violation in the radiotherapy process.

[0011] According to the technical solution provided in the embodiments of this application, the method further includes: If, after the detection equipment is activated, a boundary frame is identified within the patient's or doctor's activity area, it is determined that a violation has occurred during the radiological treatment process.

[0012] According to the technical solution provided in the embodiments of this application, the start-up of the detection device includes: Identify the brightness of the indicator light that detects the device's activation in the real-time image; When the brightness of the indicator light is greater than or equal to a preset brightness threshold, the detection device is activated.

[0013] According to the technical solution provided in the embodiments of this application, the method further includes: Real-time monitoring of light intensity in radiotherapy settings and noise levels in real-time images acquired by the image acquisition device; The first preset fluctuation threshold is updated using linear interpolation based on the light intensity and the noise level. When the light intensity is determined to be lower than a preset lower limit or the noise level exceeds a preset upper limit, an auxiliary light source is activated and the frame skipping interval is reduced to a preset safe duration.

[0014] A second aspect of this application provides an intelligent monitoring system for standardized operations in radiotherapy procedures, used to execute the intelligent monitoring method for standardized operations in radiotherapy procedures as described above, the system comprising: An image acquisition module, configured to acquire video streams of a radiotherapy scene in real time via an image acquisition device; A region division module is configured to divide the radiotherapy scene into a detection equipment area, a patient activity area, and a doctor activity area. A curve generation module is configured to perform numerical difference calculations on the reference image and the real-time image in each area under unmanned conditions, and generate a pixel value change curve representing the motion state. The behavior judgment module is configured to dynamically judge the patient's entry or exit from the patient activity area, the doctor's entry or exit from the doctor activity area, and the patient's start or end of the examination in the detection device area based on the fluctuation characteristics of the pixel value change curve. A frame-sampling adjustment module is configured to adaptively adjust the frame-sampling interval of a real-time image based on the behavior judgment result to obtain a frame image. A behavior recognition module is configured to determine whether there are any violations during the radiotherapy process based on the frame image.

[0015] Compared with existing technologies, the beneficial effects of this application are as follows: By dynamically adjusting the video frame extraction interval, the detection frequency is intelligently adapted according to the behavioral characteristics of the patient, doctor, and equipment area. The frame extraction frequency is reduced in non-critical stages, and the detection density is increased when critical behaviors occur. This effectively balances the computing power allocation of the edge computing module, avoids the waste of computing power caused by high frame rate processing at all times, and significantly reduces hardware resource consumption. Through the localized processing architecture based on edge computing, combined with the numerical difference calculation and analysis of pixel value change curves, the data transmission latency is greatly reduced, ensuring timely response to personnel behavior. At the same time, through the first preset fluctuation threshold and continuous fluctuation judgment mechanism, noise interference is effectively suppressed, improving the accuracy and reliability of behavior state recognition. Attached Figure Description

[0016] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A flowchart illustrating the steps of the intelligent monitoring method for standardized operation of radiotherapy procedures provided in Embodiment 1 of this application; Figure 2 This is a schematic diagram of the monitoring process of the intelligent monitoring method in Embodiment 1 of this application; Figure 3 This is a grayscale variation curve of each region in Embodiment 1 of this application; Figure 4 This is a distribution map of identification tags in each region of Embodiment 2 of this application; Figure 5 This is a schematic diagram of the structure of the intelligent monitoring system for standardized operation of radiotherapy procedures provided in Embodiment 5 of this application.

[0017] The reference numerals are as follows: 10, Image Acquisition Module; 20, Region Division Module; 30, Curve Generation Module; 40, Behavior Judgment Module; 50, Frame Extraction and Adjustment Module; 60, Behavior Recognition Module. Detailed Implementation

[0018] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Example 1 Please refer to Figure 1 This embodiment provides an intelligent monitoring method for standardized operation of radiotherapy procedures, including: S100: Acquires video streams of radiotherapy scenarios in real time through image acquisition equipment.

[0021] Specifically, in step S100, the image acquisition device in this embodiment is a video acquisition device. The video acquisition device acquires images of the radiotherapy scene in real time. Optionally, the video acquisition device is a surveillance camera. The surveillance camera is installed above the front of the detection equipment to ensure a complete view of the entire department, minimizing blind spots and preventing undetectable areas. This ensures that the detection equipment, patient, doctor, and radiation protection devices are all captured completely. The detection equipment is used to examine the patient's body in a radiotherapy setting, and the radiation protection equipment includes protective clothing such as lead aprons.

[0022] S200: Based on the target detection algorithm, the radiotherapy scene is divided into the detection equipment area, the patient activity area, and the doctor activity area.

[0023] Specifically, in step S200, the radiotherapy scene is first divided into regions using a pre-trained dataset to facilitate separate processing of different regions and reduce the computational load of subsequent processing. In this embodiment, the region division is performed using a scene segmentation algorithm or manual division. The radiotherapy scene is ultimately divided into three parts: the detection equipment area, the patient activity area, and the doctor activity area. The detection equipment area is the area in the radiotherapy scene where the detection equipment is placed; the patient activity area is the area accessed from the door to the detection equipment area; and the doctor activity area is the area where doctors enter the isolation room from the detection equipment area for observation. Figure 2 As shown.

[0024] S300: Performs numerical difference calculations on the baseline image and real-time image in each area under unmanned conditions, and generates pixel value change curves representing the motion state.

[0025] Specifically, in step S300, the reference image corresponding to each divided region is first acquired. The reference image is the image of the region when no one is present. The reference image is pre-stored in a preset database and can be retrieved from the preset database when needed. Next, the real-time image corresponding to each region is obtained based on the acquired video stream. The reference image and the real-time image in the corresponding region are then compared and numerically differentiated. The pixel value changes in the three regions are extracted and the results are recorded. The three regions are calculated in parallel to generate a pixel value change curve for each region. It should be noted that the numerical differentiation calculation can use background subtraction, inter-frame subtraction, or other difference operations. It can also use calculation methods such as pixel value changes, variance, or energy in different regions. The specific method can be selected according to the actual situation. The ultimate goal is to calculate the pixel value difference between the reference image and the real-time image.

[0026] Furthermore, the numerical difference calculation for the baseline image and real-time image in each unmanned area includes: S301: Convert the RGB pixel values ​​of the real-time image into grayscale pixel values.

[0027] Specifically, in step S301, the real-time image typically acquired by the image acquisition device is a real-time RGB image. After obtaining the real-time RGB image, the RGB pixel values ​​in the real-time RGB image are converted into grayscale pixel values, i.e., the real-time RGB image is converted into a grayscale image. In this embodiment, converting the RGB image into a grayscale image reduces the computational load when performing subsequent numerical difference calculations between the real-time image and the reference image, thereby saving computational resources. It should be noted that the reference image is stored in grayscale image format to reduce the processing operations on the reference image during numerical difference calculations. In this embodiment, the numerical difference calculation uses a background difference method.

[0028] S302: Subtract the pixel values ​​of the real-time image from the pixel values ​​of the reference image in the corresponding area point by point to generate grayscale change curves of the patient's activity area, detection device area, and doctor's activity area.

[0029] Specifically, in step S302, after obtaining the reference image and real-time image corresponding to each region, the pixel values ​​in each reference image and real-time image are collected respectively; then, a region mask is generated for the divided region, and OpenCV is used to subtract the pixel values ​​from the image for only specific regions to extract grayscale changes. The three regions are traversed to generate the grayscale change curves of the patient's activity area, the detection device area, and the doctor's activity area.

[0030] Several regions are defined as follows: In the baseline image, the pixel value of the patient's active area is... The pixel value of the doctor's activity area The pixel value of the detection device area In real-time images, the pixel value of the patient's active area is... The pixel value of the doctor's activity area The pixel value of the detection device area .

[0031] The ordinate of the grayscale curve can be obtained by subtracting the pixel values ​​of the corresponding regions. The specific calculation method is as follows:

[0032]

[0033]

[0034] in, The vertical axis represents the grayscale change curve of the patient's active area. The vertical axis represents the grayscale variation curve of the detection equipment area. The vertical axis represents the grayscale change curve of the doctor's activity area, and the corresponding grayscale change curve of the patient's activity area is represented by... This indicates that the grayscale change curve of the detection equipment area is obtained through... This indicates that the grayscale change curve of the doctor's activity area is achieved through... express, , and The horizontal axis represents time, and the vertical axis represents the fluctuation value.

[0035] Specifically, in this embodiment, the pixel value change curve is presented in the form of a grayscale change curve. By performing subsequent processing based on the fluctuations in the grayscale change curve, the motion state of each region can be judged in a single channel, reducing the computational resources occupied during the judgment process. In other embodiments, the pixel value change curve can also be presented directly in the form of an RGB change curve. Presenting it in the form of an RGB change curve is more intuitive, but it requires judging the motion state of each region in three channels, which consumes more computational resources.

[0036] S400: Based on the fluctuation characteristics of the grayscale change curve, dynamically determine whether the patient enters or leaves the patient activity area, whether the doctor enters or leaves the doctor activity area, and whether the patient begins or ends the examination behavior in the detection equipment area.

[0037] Specifically, in step S400, after obtaining the grayscale change curve corresponding to each region, the fluctuation amount in each grayscale change curve is obtained one by one. Based on the fluctuation amount, it is determined whether there is activity in the corresponding region. The fluctuation in the patient's activity area may be caused by the patient entering or leaving the patient's activity area. The fluctuation in the doctor's activity area may be caused by the doctor entering or leaving the doctor's activity area. The fluctuation in the detection device area may be caused by the actions of the patient and the doctor during the detection preparation stage.

[0038] Furthermore, the dynamic determination in step S400 includes: A1: When the continuous fluctuation of the grayscale change curve of the patient's activity area exceeds the first preset fluctuation threshold, it is determined that there are people in the patient's activity area and that the people are in an active state.

[0039] Specifically, in A1, the grayscale of the real-time image of the patient's activity area is first compared with that of the baseline image. When the grayscale difference exceeds a preset grayscale threshold, it indicates the presence of a person within the patient's activity area. Then, the grayscale change curve of the patient's activity area is identified. When there are continuous fluctuations in the data, the continuous fluctuations corresponding to the patient's active area are considered. Data collection was performed; then the continuous fluctuations were recorded. Compared with a first preset fluctuation threshold, when continuous fluctuations occur... When the fluctuation is less than or equal to the first preset fluctuation threshold, it is determined that the people in the patient's activity area are in a relatively static state. At this time, the people in the patient's activity area are likely the patient, but could also be the patient's companion or other personnel; when the fluctuation is continuous... If the fluctuation exceeds the first preset threshold, it is determined that there are not only people in the patient's activity area, but they are also active. This means that the patient, their companion, or other people may be entering or leaving the patient's activity area.

[0040] A2: When the continuous fluctuation of the grayscale change curve of the doctor's activity area exceeds the first preset fluctuation threshold, it is determined that there are people in the doctor's activity area and the people are in an active state.

[0041] Specifically, in A2, the grayscale of the real-time image of the doctor's activity area is first compared with that of a reference image. When the grayscale difference exceeds a preset grayscale threshold, it indicates the presence of a person within the doctor's activity area. Then, the grayscale change curve of the doctor's activity area is identified. When there are continuous fluctuations in the continuous fluctuations, Data collection was performed; then the continuous fluctuations were recorded. Compared with a first preset fluctuation threshold, when continuous fluctuations occur... When the fluctuation is less than or equal to the first preset fluctuation threshold, it is determined that the person in the doctor's activity area is in a relatively static state, and at this time, the person in the doctor's activity area is likely a doctor; when the fluctuation is continuous... When the fluctuation exceeds the first preset threshold, it is determined that there are not only people in the doctor's activity area, but they are also active. This means that the doctor may be approaching the testing equipment to prepare for the patient's test, or the doctor may have left the area and entered the observation room after completing the preparation.

[0042] A3: When the continuous fluctuation of the grayscale change curve of the detection device area exceeds the first preset fluctuation threshold, it is determined that there are people in the detection device area and the people are in an active state.

[0043] Specifically, in A3, the grayscale of the real-time image of the detection device area is first compared with that of the reference image. When the grayscale difference exceeds a preset grayscale threshold, it indicates the presence of a person within the detection device area. Then, the grayscale change curve of the detection device area is identified. When there is continuous fluctuation, for continuous fluctuation Data collection was performed; then the continuous fluctuations were recorded. Compared with a first preset fluctuation threshold, when continuous fluctuations occur... When the fluctuation is less than or equal to the first preset fluctuation threshold, it is determined that the person in the detection equipment area is in a relatively static state, which may be because the patient is in the preparation stage or during the detection process; when the fluctuation is continuous... If the fluctuation exceeds the first preset threshold, it is determined that there are not only people in the detection area, but they are also active. This means that at this time, it may be that a doctor is preparing for the patient's test in the detection equipment area, or the patient is stepping onto the detection equipment, or the patient is leaving the detection equipment.

[0044] A4: When the continuous fluctuation of the grayscale change curve of the detection device area exceeds the second preset fluctuation threshold but is less than the first preset fluctuation threshold, it is determined that there are people and scene fluctuations in the detection device area.

[0045] Specifically, in A4, for continuous fluctuations After data collection, in addition to comparing it with the first preset fluctuation threshold, it is also compared with the second preset fluctuation threshold. When continuous fluctuations occur... If the fluctuation is less than or equal to the first preset fluctuation threshold and greater than the second preset fluctuation threshold, it is determined that there is continuous fluctuation within the detection equipment area, but the fluctuation is small. That is, there is scene fluctuation within the detection equipment area. The scene fluctuation may be caused by the indicator light used by the detection equipment to indicate that the detection is in progress, or it may be caused by the indicator light used by the detection equipment to indicate that the detection is in progress.

[0046] The presence or absence of patients and doctors in each area indicates the behavior judgment result. After determining whether the patient and doctor are active or inactive based on the judgment conditions A1-A4 above, and whether there are scene fluctuations within the detection device area, the behavior judgment result can be determined. This embodiment judges behavior based on grayscale value differences. In other embodiments, behavior can also be judged through visible light differences, infrared differences, etc., which will not be elaborated here.

[0047] S500: Based on the behavior judgment result, adaptively adjust the frame extraction interval of the real-time image to obtain a frame image.

[0048] Specifically, in step S500, the frame extraction interval for different regions is adaptively adjusted based on the judgment result obtained in step S400. Then, the real-time image within the region is extracted according to the corresponding frame extraction interval, so that the frame extraction frequency is changed only when specific behaviors change, thereby improving the automation level of the system.

[0049] Further, the adaptive adjustment of the frame skipping interval of the real-time image in step S500 includes: B1: When it is determined that there are people in the patient's activity area and the doctor's activity area and the people are in an active state, and there are no people in the detection equipment area, the frame extraction interval is set to the first duration; B2: When it is determined that there are no people in the patient's activity area and the doctor's activity area, but there are people in the detection equipment area and they are active, the frame extraction interval is set to the second duration. B3: When it is determined that there are no people in the patient's activity area, there are people in the detection equipment area and there are scene fluctuations, or there are people in the doctor's activity area and the people are in an active state, the frame extraction interval is set to the third duration. B4: When it is determined that there are no people in the patient's activity area, there are people in the detection equipment area and there are scene fluctuations, and there are no people in the doctor's activity area, the frame extraction interval is set to the fourth duration.

[0050] Please refer to Figure 3 , Figure 3The real-time images of each region are divided into seven time periods based on fluctuation characteristics. Specifically, when B1 is satisfied, such as... Figure 3 The second time period from the left is shown in the middle, that is, when and When there are fluctuations, it indicates that the patient and doctor are moving within the patient's activity area and the doctor's detection area, respectively. This suggests that the patient and doctor are moving towards the detection area in preparation for testing. In this case, it is unnecessary to extract too many frames; the frame extraction interval can be set to the first duration. First duration Longer, meaning fewer frames are acquired in the same amount of time; when B2 is satisfied, such as Figure 3 The third time period from the left is shown in the middle, which is only... When there are fluctuations, especially large ones, it can be determined that the patient is lying down onto the testing device, or the doctor is helping the patient lie down onto the testing device. At this point, it is already in the preparation stage before the formal test, and it is required to check whether the patient is wearing the device correctly. Therefore, the frame extraction time can be set as the second duration. Second duration Compared to the first duration Shorter, meaning more frames need to be acquired within the same time frame; while satisfying B3, such as Figure 3 The fourth time period from the left is shown in the middle. Basically no fluctuation and and There are fluctuations, and If the fluctuations are small, it can be determined that the patient is ready. The indicator light on the testing equipment flashes, signaling readiness for testing, and the doctor prepares to leave. At this point, it is necessary to re-monitor whether the patient is wearing the protective gear correctly. Therefore, the frame skipping interval can be set to the third duration. The third duration Compared to the second duration Shorter, meaning more frames need to be acquired in the same amount of time; while meeting B4, such as Figure 3 The fifth time period from the left in the middle is shown, that is... and There were no fluctuations. The presence of fluctuations, though small, indicates that the patient has officially begun the examination. The indicator light on the testing equipment flashes to show that the test is in progress. Based on the previous work, the frame skipping interval can be slightly increased at this point, setting it to the fourth duration. Fourth duration It can be longer than the third duration. .

[0051] The above four scenarios represent four stages that require increased attention, thus necessitating rigorous adjustments to the frame skipping interval. In addition to these four scenarios, adaptive adjustment of the real-time image frame skipping interval also includes the following two conditions: B5: When it is determined that there are people in the patient's activity area, the detection equipment area, and the doctor's activity area, the frame extraction interval is set to the fifth duration. B6: When it is determined that there are no people in the patient's activity area, the detection equipment area, and the doctor's activity area, the frame extraction interval is set to the sixth duration.

[0052] Please refer to Figure 3 Specifically, when B5 is met, such as Figure 3 The sixth time period from the left is shown in the middle, that is, when , and When all three areas show significant fluctuations, it indicates the presence of people in the patient's activity area, the testing equipment area, and the doctor's activity area. It can be concluded that the patient completed the examination in the testing equipment area, the doctor entered the doctor's activity area, and the patient's accompanying person entered the patient's activity area. At this point, the examination is complete, and the frame extraction duration can be set to the fifth duration. Fifth duration Greater than the first duration and the fourth duration When B6 is satisfied, such as Figure 3 The first and seventh time periods from the left are shown in the middle, that is, when , and If there are no fluctuations, the entire medical room can be considered to be in an idle state, and no standardized operation monitoring is required. The frame extraction time can be set to the sixth interval. The sixth duration Greater than the fifth duration .

[0053] Optional, first duration It lasts for 2 seconds, the second duration. One second, third duration It is 0.5 seconds, the fourth duration. It lasts 1.5 seconds, the fifth duration. It lasts for 5 seconds, the sixth duration. It lasts for 10 seconds.

[0054] S600: Determine whether there are any violations during the radiotherapy process based on the frame image.

[0055] Specifically, in step S600, after obtaining the frame image through frame extraction in step S500, the system identifies whether any violations exist based on the frame image. The specific method is as follows: S601: Mark the bounding box of the identified object, which includes at least detection equipment, doctors, patients and radiation protection devices.

[0056] Specifically, in step S601, the bounding boxes of the identified objects are first established with a preset point in the frame image as the origin to obtain the spatial coordinate positions of multiple identified objects, and the target bounding boxes are used to mark them and record the detection results; the image recognition algorithm is used to detect the boundary regions of the multiple identified objects.

[0057] S602: Calculate the distance between the bounding boxes corresponding to each of the identified objects in the frame image.

[0058] Specifically, in step S602, this embodiment provides two methods for calculating the distance between bounding boxes, namely the following two approaches: Method 1, calculate any , The distance between the bounding boxes of the two identified objects:

[0059]

[0060]

[0061]

[0062]

[0063] in, To identify objects The coordinates of the top left corner of the bounding box To identify objects The coordinates of the bottom right corner of the bounding box To identify objects The coordinates of the top left corner of the bounding box For is compared to object The coordinates of the bottom right corner of the bounding box Represents the object to be identified The area of ​​the bounding box, Represents the object to be identified The area of ​​the bounding box, Represents the object to be identified and identification objects The distance between bounding boxes.

[0064] Method 2, calculate any , The distance between the bounding boxes of the two identified objects:

[0065] in, Represents the object to be identified The center position of the bounding box Represents the object to be identified The center position of the bounding box.

[0066] S603: When the distance between the patient and the boundary frame corresponding to the radiation protection device is greater than a preset distance threshold, it is determined that there is a violation in the radiotherapy process.

[0067] Specifically, in step S603, after obtaining the distance between the patient and the radiation protection device, the distance between the patient's bounding box and the radiation protection device's bounding box is compared with a preset distance threshold. When the distance between the bounding boxes is less than or equal to the preset distance threshold, the patient is considered to be protected, and thus no violation is identified. When the distance between the bounding boxes is greater than the preset distance threshold, the patient is considered not to be protected, and thus a violation is identified.

[0068] In other embodiments, the patient's bounding box and the radiation protection device's bounding box can be calculated first. If the overlapping area of ​​the two bounding box areas is greater than a set area threshold, the patient is considered not to be protected, and thus a violation is identified.

[0069] In step S603, in addition to judging whether there is a violation based on whether the patient is protected, it is also possible to judge whether the doctor is protected. For example, if the distance between the doctor and the boundary frame corresponding to the radiation protection device is greater than a preset distance threshold, it is determined that there is a violation in the radiotherapy process.

[0070] Example 2 This embodiment provides another intelligent monitoring method for standardized operation of radiotherapy procedures. The same content as in Embodiment 1 will not be repeated here; the difference lies in: The adaptive adjustment of the frame skipping interval of the real-time image includes: Based on the dynamic judgment results of each region, a corresponding identification label is generated for each region; the identification label includes: whether there is continuous fluctuation in the grayscale change curve of each region, and whether the continuous fluctuation satisfies that it is less than or equal to a first preset fluctuation threshold and greater than a second preset fluctuation threshold. A combination of identification tags is established based on the identification tags corresponding to each region. The frame extraction strategy mapping table is then queried based on the identification tag combination to determine the corresponding frame extraction interval. The frame extraction strategy mapping table includes the correspondence between different identification tag combinations and frame extraction intervals.

[0071] Specifically, in this embodiment, based on whether continuous fluctuations exist within the region, and whether the continuous fluctuations meet the criteria of being less than or equal to a first preset fluctuation threshold and greater than a second preset fluctuation threshold, all identification tag types are pre-determined. The criterion for determining whether continuous fluctuations exist within the region is that the continuous fluctuations are greater than the second preset fluctuation threshold; continuous fluctuations meeting the criteria of being less than or equal to the first preset fluctuation threshold and greater than the second preset threshold are defined as the presence of scene fluctuations. It should be noted that continuous fluctuations caused by human activity are considered to exceed the first preset fluctuation threshold. Based on this, the identification tag types include: There are continuous fluctuations within the region and scene fluctuations. There are continuous fluctuations within the region, but no scene fluctuations. There are no continuous fluctuations or scene fluctuations within the region; There are no continuous fluctuations within the region, but there are scene fluctuations. The presence of scene fluctuations but no continuous fluctuations within the region is considered an abnormal situation, requiring a system check. For ease of recording, continuous fluctuations are recorded as "1", and the absence of continuous fluctuations as "0". A frame extraction strategy mapping table is constructed based on the combination of identification tag types between different regions. The frame extraction strategy mapping table is shown in Table 1 below: Table 1

[0072] Next, after obtaining the identification identifiers corresponding to each region, the frame extraction strategy mapping table is queried based on the combination of identification identifiers between different regions to obtain the corresponding frame extraction interval.

[0073] For ease of explanation, please refer to Figure 3 The situation shown is constructed as follows: Figure 4 The identification tag distribution diagram shown indicates that continuous fluctuations of "1" are represented by high-level tags, and continuous fluctuations of "0" are represented by low-level tags.

[0074] This embodiment abstracts the fluctuation state of each region into high / low level tags, and identifies the tags and the frame extraction interval by matching the preset frame extraction strategy mapping table. The core is to simplify the condition judgment into a mapping query of tag combinations, reducing the real-time calculation burden.

[0075] Example 3 This embodiment provides another intelligent monitoring method for standardized operation of radiotherapy procedures. The same content as in Embodiment 1 will not be repeated here; the difference lies in: The method further includes: If, after the detection equipment is activated, a boundary frame is identified within the patient's or doctor's activity area, it is determined that a violation has occurred during the radiological treatment process.

[0076] Specifically, in this embodiment, after the detection device starts detecting the patient, if a bounding box is detected in the patient's activity area or the doctor's activity area, it proves that there are other people besides the patient in the radiotherapy scene, and at this time it is determined that there is a violation in the radiotherapy process.

[0077] Furthermore, the activation of the judgment and detection device includes: Identify the brightness of the indicator light that detects the device's activation in the real-time image; When the brightness of the indicator light is greater than or equal to a preset brightness threshold, the detection device is activated.

[0078] Specifically, the indicator lights for the detection device in the real-time image are monitored in real time, and the brightness of the indicator lights is obtained. The brightness of the indicator lights is compared with a preset brightness threshold. When the brightness of the indicator lights is less than or equal to the preset brightness threshold, the detection device is considered not to be started; when the brightness of the indicator lights is greater than or equal to the preset brightness threshold, the detection device is considered to be started.

[0079] Example 4 This embodiment provides another intelligent monitoring method for standardized operation of radiotherapy procedures. The same content as in Embodiment 1 will not be repeated here; the difference lies in: The method further includes: Real-time monitoring of light intensity in radiotherapy settings and noise levels in real-time images acquired by the image acquisition device; The first preset fluctuation threshold is updated using linear interpolation based on the light intensity and the noise level. When the light intensity is determined to be lower than a preset lower limit or the noise level exceeds a preset upper limit, an auxiliary light source is activated and the frame skipping interval is reduced to a preset safe duration.

[0080] Specifically, in this embodiment, the light intensity value in the radiotherapy scene is acquired in real time by the photosensor built into the image acquisition device. Simultaneously, based on the grayscale distribution of the current frame image, the image signal-to-noise ratio or mean square error is calculated, and the noise level is quantized. Next, the light intensity value With the set standard light intensity range Compare and set noise levels Compared to standard noise range Compare; Effect of light intensity: When At that time, the threshold is updated by linear interpolation:

[0081] in, and These are the fluctuation thresholds corresponding to the lower and upper limits of the standard light intensity range, respectively; Noise level impact: When At that time, the threshold is updated by linear interpolation:

[0082] in, and These are the fluctuation thresholds corresponding to the lower and upper limits of the standard noise range, respectively. Overall threshold: The final fluctuation threshold is ,in and These are the weighting coefficients for illumination and noise, respectively, and are optional. And allows for adjustments based on the actual scenario. and For example, increasing the lighting weight in low-light scenes.

[0083] When in or In extreme scenarios, auxiliary light sources are directly activated, and the frame extraction interval is simultaneously reduced to the shortest safe duration to ensure effective recognition of grayscale change curves.

[0084] Linear interpolation is used to smoothly adjust the threshold for illumination and noise, avoiding misjudgments caused by sudden environmental changes; hardware compensation is performed by enabling auxiliary light sources and adjusting the frame extraction strategy to ensure system robustness.

[0085] Example 5 Please refer to Figure 5 This embodiment provides an intelligent monitoring system for standardized operation of radiotherapy procedures, used to execute the intelligent monitoring method for standardized operation of radiotherapy procedures as described in Embodiments 1-3. The system includes: Image acquisition module 10, configured to acquire video streams of radiotherapy scenes in real time through image acquisition devices; The area division module 20 is configured to divide the radiotherapy scene into a detection equipment area, a patient activity area, and a doctor activity area. The curve generation module 30 is configured to perform numerical difference calculations on the reference image and the real-time image in the unmanned state of each region to generate a pixel value change curve representing the motion state. The behavior judgment module 40 is configured to dynamically judge the patient's entry or exit from the patient activity area, the doctor's entry or exit from the doctor activity area, and the patient's start or end of the examination in the detection device area based on the fluctuation characteristics of the pixel value change curve. A frame extraction adjustment module 50 is configured to adaptively adjust the frame extraction interval of a real-time image based on the behavior judgment result to obtain a frame image. The behavior recognition module 60 is configured to determine whether there are any violations during the radiotherapy process based on the frame image.

[0086] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for intelligent monitoring of standardized operations in radiotherapy, characterized in that, include: Real-time video streams of radiotherapy scenarios are acquired using image acquisition equipment; The radiotherapy scene is divided into a detection equipment area, a patient activity area, and a doctor activity area. Numerical difference calculations are performed on the baseline image and real-time image in each area under unmanned conditions to generate pixel value change curves representing motion state; Based on the fluctuation characteristics of the pixel value change curve, the system dynamically determines whether the patient enters or leaves the patient activity area, whether the doctor enters or leaves the doctor activity area, and whether the patient begins or ends the examination in the detection device area. Based on the behavior judgment result, the frame extraction interval of the real-time image is adaptively adjusted to obtain the frame image; The frame images are used to determine whether any violations occurred during the radiotherapy process.

2. The intelligent monitoring method for standardized operation of radiotherapy procedures according to claim 1, characterized in that, The calculation of numerical differences between the baseline image and the real-time image in each unmanned area includes: Convert the RGB pixel values ​​of the real-time image into grayscale pixel values; The pixel values ​​of the real-time image are compared with the pixel values ​​of the corresponding reference image to generate grayscale change curves for the patient's activity area, the detection device area, and the doctor's activity area.

3. The intelligent monitoring method for standardized operation of radiotherapy procedures according to claim 2, characterized in that, The dynamic judgment includes: When the continuous fluctuation of the grayscale change curve of the patient's activity area exceeds the first preset fluctuation threshold, it is determined that there are people in the patient's activity area and that the people are in an active state. When the continuous fluctuation of the grayscale change curve of the doctor's activity area exceeds the first preset fluctuation threshold, it is determined that there are people in the doctor's activity area and that the people are in an active state. When the continuous fluctuation of the grayscale change curve of the detection equipment area exceeds the first preset fluctuation threshold, it is determined that there are people in the detection equipment area and that the people are in an active state. When the continuous fluctuation of the grayscale change curve of the detection device area exceeds the second preset fluctuation threshold but is less than the first preset fluctuation threshold, it is determined that there are people and scene fluctuations in the detection device area.

4. The intelligent monitoring method for standardized operation of radiotherapy procedures according to claim 3, characterized in that, The adaptive adjustment of the frame skipping interval of the real-time image includes: When it is determined that there are people in the patient's activity area and the doctor's activity area and the people are in an active state, and there are no people in the detection equipment area, the frame extraction interval is set to a first duration. When it is determined that there are no people in the patient's activity area and the doctor's activity area, but there are people in the detection equipment area and they are active, the frame extraction interval is set to the second duration. When it is determined that there are no people in the patient's activity area, there are people in the detection equipment area and there are scene fluctuations, or there are people in the doctor's activity area and the people are in an active state, the frame extraction interval is set to the third duration. When it is determined that there are no people in the patient's activity area, there are people in the detection equipment area and there are scene fluctuations, and there are no people in the doctor's activity area, the frame extraction interval is set to the fourth duration.

5. The intelligent monitoring method for standardized operation of radiotherapy procedures according to claim 3, characterized in that, The adaptive adjustment of the frame skipping interval of the real-time image includes: Based on the dynamic judgment results of each region, a corresponding identification label is generated for each region; the identification label includes: whether there is continuous fluctuation in the grayscale change curve of each region, and whether the continuous fluctuation satisfies that it is less than or equal to a first preset fluctuation threshold and greater than a second preset fluctuation threshold. A combination of identification tags is established based on the identification tags corresponding to each region. The frame extraction strategy mapping table is then queried based on the identification tag combination to determine the corresponding frame extraction interval. The frame extraction strategy mapping table includes the correspondence between different identification tag combinations and frame extraction intervals.

6. The intelligent monitoring method for standardized operation of radiotherapy procedures according to claim 4 or 5, characterized in that, The step of determining whether there are any violations during radiotherapy based on the frame image includes: The bounding box of the identified object is marked, and the identified object includes at least detection equipment, doctors, patients and radiation protection devices; Calculate the distance between the bounding boxes corresponding to each of the identified objects in the frame image; When the distance between the patient and the boundary frame corresponding to the radiation protection device is greater than a preset distance threshold, it is determined that there is a violation in the radiotherapy process.

7. The intelligent monitoring method for standardized operation of radiotherapy procedures according to claim 6, characterized in that, The method further includes: If, after the detection equipment is activated, a boundary frame is identified within the patient's or doctor's activity area, it is determined that a violation has occurred during the radiological treatment process.

8. The intelligent monitoring method for standardized operation of radiotherapy procedures according to claim 7, characterized in that, The activation of the detection device includes: Identify the brightness of the indicator light that detects the device's activation in the real-time image; When the brightness of the indicator light is greater than or equal to a preset brightness threshold, the detection device is activated.

9. The intelligent monitoring method for standardized operation of radiotherapy procedures according to claim 8, characterized in that, The method further includes: Real-time monitoring of light intensity in radiotherapy settings and noise levels in real-time images acquired by the image acquisition device; The first preset fluctuation threshold is updated using linear interpolation based on the light intensity and the noise level. When the light intensity is determined to be lower than a preset lower limit or the noise level exceeds a preset upper limit, an auxiliary light source is activated and the frame skipping interval is reduced to a preset safe duration.

10. An intelligent monitoring system for standardized operation of radiotherapy procedures, characterized in that, The system is used to perform the intelligent monitoring method for standardized operation of radiotherapy procedures as described in any one of claims 1-9, the system comprising: Image acquisition module (10), the image acquisition module (10) is configured to acquire video streams of radiotherapy scenes in real time through image acquisition devices; A region division module (20) is configured to divide the radiotherapy scene into a detection equipment area, a patient activity area, and a doctor activity area. Curve generation module (30) is configured to perform numerical difference calculation on the reference image and real-time image in the unmanned state of each region, and generate a pixel value change curve representing the motion state. The behavior judgment module (40) is configured to dynamically judge the patient's entry or exit from the patient activity area, the doctor's entry or exit from the doctor activity area, and the patient's start or end of the examination behavior in the detection device area based on the fluctuation characteristics of the pixel value change curve. A frame-skipping adjustment module (50) is configured to adaptively adjust the frame-skipping interval of a real-time image based on the behavior judgment result to obtain a frame image. A behavior recognition module (60) is configured to determine whether there are any violations during the radiotherapy process based on the frame image.

Images