Medical linear accelerator anomaly detection method, device, equipment and medium

Abstract

The invention discloses a medical linear accelerator anomaly detection method, device and equipment and a medium, and the method comprises the steps: obtaining to-be-detected data collected by at least two sensors of a medical linear accelerator; converting the to-be-detected data acquired by the at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents a pixel value corresponding to an acquisition value of one sensor at a time point; and inputting the two-dimensional image data converted from the to-be-detected data into a trained anomaly detection model, and outputting an anomaly detection result. According to the invention, the technical problems of relatively single consideration factor and relatively high probability of false alarm and missing alarm of the existing anomaly detection system of the medical linear accelerator with a single sensor can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of abnormal identification of medical devices, and particularly to a method, device, equipment and medium for abnormal detection of a medical linear accelerator. Background Art

[0002] A medical linear accelerator is a highly precise medical device widely used in cancer radiotherapy. It accelerates electron beams to near the speed of light and makes these electron beams strike a metal target, thereby generating high-energy X-rays or electron beams. Through these rays, the linear accelerator can accurately irradiate the tumor area to destroy tumor cells and achieve the purpose of treating cancer. The medical linear accelerator is used to accurately irradiate tumors, and the effect and safety of radiotherapy depend on the accuracy and dose of the radiation beam. If the device fails or operates deviates, it may lead to inaccurate irradiation, resulting in damage to healthy tissues or failure to effectively irradiate the tumor, affecting the treatment effect. Therefore, the abnormal detection of a medical linear accelerator needs to be able to detect any abnormality in the system in time, avoid errors during the treatment process, and ensure the treatment safety of patients to the greatest extent.

[0003] A medical linear accelerator is a high-tech device involving high-energy rays, precise mechanical movements and complex electronic control systems. A failure in any part of the device (such as the cooling system, power supply system, position sensor, etc.) may cause the device to shut down and even cause permanent damage to the device. The abnormal detection system can detect potential failures in time and trigger an alarm by monitoring the device status in real time, thereby avoiding major failures, reducing maintenance costs and device downtime.

[0004] Radiotherapy is often a long-term and continuous process, and patients need to receive treatment multiple times. The stability of the device is crucial for the effectiveness of treatment. Through automated abnormal detection, the system can check any deviation or problem in the device operation at any time, ensure that each treatment is carried out according to the established standards and procedures, and avoid affecting the treatment effect of patients due to any instability of the device.

[0005] The abnormal detection system can not only detect faults, but also provide timely maintenance suggestions or warnings to help technicians intervene before the faults become serious. This proactive management method makes device maintenance more efficient, avoids emergency handling when the device fails, and thus improves the working efficiency of the device. Although operators are usually strictly trained, human factors may still lead to operation errors or judgment mistakes. Automated abnormal detection can reduce these risks by continuously monitoring the device status and issuing warnings when the system deviates, helping operators adjust in time to ensure that the device works as expected.

[0006] The anomaly detection of medical linear accelerators is not only about monitoring the equipment, but also a crucial link in ensuring patient safety, improving treatment quality, and ensuring the long-term stable operation of the equipment. Through real-time monitoring, precise feedback, and warning systems, anomaly detection can effectively prevent failures, ensure treatment accuracy, improve equipment reliability and treatment efficiency, which is essential for enhancing the quality and effectiveness of the entire radiotherapy process.

[0007] Currently, the anomaly detection of medical linear accelerators mainly relies on threshold detection of individual sensors. For example, temperature sensors are installed at different positions of the accelerator, and the upper and lower limits of the temperature sensors are controlled. Once the temperature exceeds the preset upper and lower limits, the linear accelerator will be alarmed to avoid equipment damage caused by overheating. This anomaly detection method considers relatively few factors, and the probability of false alarms and missed alarms is relatively high. Summary of the Invention

[0008] The present invention provides a method, device, equipment, and medium for anomaly detection of medical linear accelerators, which are used to solve the technical problems existing in the anomaly detection system of medical linear accelerators with individual sensors, such as relatively few factors considered and relatively high probabilities of false alarms and missed alarms.

[0009] According to one aspect of the present invention, there is provided a method for anomaly detection of medical linear accelerators, including: Obtaining measurement data collected by at least two sensors of a medical linear accelerator; Converting the measurement data collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the collected value of one sensor at a time point; Inputting the two-dimensional image data obtained by converting the measurement data into a trained anomaly detection model to output an anomaly detection result.

[0010] Optionally, after obtaining the measurement data collected by at least two sensors of the medical linear accelerator, it further includes: Preprocessing the measurement data; The preprocessing at least includes aligning the measurement data collected by at least two sensors in chronological order; if the data cannot be aligned, then performing interpolation processing on the measurement data collected by at least one sensor, and then aligning the measurement data collected by at least two sensors in chronological order.

[0011] Optionally, the converting the measurement data collected by at least two sensors into two-dimensional image data includes: Normalize the data to be measured collected by at least two sensors at each moment to obtain the gray values of the data to be measured at each moment, so that the data to be measured of each sensor is converted into a series of gray value points in chronological order; Stitch the data to be measured of at least two sensors after normalization processing based on a preset order to obtain the two-dimensional image data; each row in the two-dimensional image data represents the gray value corresponding to the data to be measured of a sensor at different moments, and each column represents the gray value corresponding to the data to be measured of different sensors at a moment.

[0012] Optionally, it further includes: Obtain the historical data collected by at least two sensors of multiple patients through a medical linear accelerator; Convert the historical data collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the acquisition value of a sensor at a time point; Label the two-dimensional image data converted from the historical data, and the type of label is the abnormal type of the medical linear accelerator; Use the two-dimensional image data corresponding to one patient as a sample to construct a data set; Train the constructed anomaly detection model based on the data set to obtain the trained anomaly detection model.

[0013] Optionally, after the step of obtaining the historical data collected by at least two sensors of multiple patients through a medical linear accelerator, it further includes: Preprocess the historical data; The preprocessing at least includes aligning the historical data collected by at least two sensors in chronological order; if the data cannot be aligned, perform interpolation processing on the historical data collected by at least one sensor, and then align the historical data collected by at least two sensors in chronological order.

[0014] Optionally, the step of converting the historical data collected by at least two sensors into two-dimensional image data includes: Normalize the historical data collected by at least two sensors at each moment to obtain the gray values of the historical data at each moment, so that the historical data of each sensor is converted into a series of gray value points in chronological order; Stitch the historical data of at least two sensors after normalization processing based on a preset order to obtain the two-dimensional image data; each row in the two-dimensional image data represents the gray value corresponding to the historical data of a sensor at different moments, and each column represents the gray value corresponding to the historical data of different sensors at a moment.

[0015] Optionally, the anomaly detection model is a deep learning model for processing images; The anomaly detection model includes a pre-trained ResNet model, a fully connected layer, and a classifier; The pre-trained ResNet model is used to extract features from the two-dimensional image data; The extracted features are input into the fully connected layer, and the fully connected layer outputs a feature map to the classifier for classification, and the classification result is output.

[0016] According to another aspect of the present invention, there is provided a device for detecting anomalies in a medical linear accelerator, including: An acquisition unit for acquiring measurement data collected by at least two sensors of a medical linear accelerator; A conversion unit for converting the measurement data collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the acquisition value of one sensor at a certain time point; An anomaly detection unit for inputting the two-dimensional image data obtained by converting the measurement data into a trained anomaly detection model and outputting an anomaly detection result.

[0017] According to another aspect of the present invention, there is provided an electronic device, the electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the anomaly detection of the medical linear accelerator according to any embodiment of the present invention.

[0018] According to another aspect of the present invention, there is provided a computer-readable storage medium, the computer-readable storage medium stores computer instructions, and the computer instructions are used to implement the anomaly detection of the medical linear accelerator according to any embodiment of the present invention when executed by a processor.

[0019] The technical solution of the embodiment of the present invention captures the associated sensor data across dimensions by acquiring the data collected by at least two sensors and converting the data of at least two sensors into two-dimensional image data, and enables the anomaly detection model to better capture the anomaly features in the multi-dimensional data, thereby improving the anomaly detection accuracy and reducing the false alarm rate and the missed alarm rate.

[0020] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0022] Figure 1 is a flowchart of a method for detecting anomalies in a medical linear accelerator according to Embodiment 1 of the present invention; Figure 2 is a flowchart of a method for detecting anomalies in a medical linear accelerator according to Embodiment 2 of the present invention; Figure 3 is a schematic diagram of converting the gray value sequence corresponding to the data collected by the sensor into two-dimensional image data in an embodiment of the present invention; Figure 4 is a schematic diagram of the architecture of an anomaly detection model in an embodiment of the present invention; Figure 5 is a structural diagram of a device for detecting anomalies in a medical linear accelerator according to Embodiment 3 of the present invention; Figure 6 is a schematic diagram of the structure of an electronic device for implementing the method for detecting anomalies in a medical linear accelerator according to Embodiment 4 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] In order to enable those skilled in the art to better understand the solutions of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "first", "second", etc. in the description, claims and above-mentioned drawings of the present invention are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0025] The present invention aims to solve the technical problem that the abnormal detection system of a medical linear accelerator with a single sensor has relatively single considerations and a relatively high probability of false alarms and missed alarms.

[0026] The present invention provides an abnormal detection system for a medical linear accelerator that simultaneously considers multi-modal data of multiple sensors. Specifically, the multi-dimensional time-series data of multiple sensors of the medical linear accelerator is converted into two-dimensional image data, and through transfer learning, a mature pre-trained image deep learning model is applied to the real-time sensor abnormal detection of the medical linear accelerator to achieve simultaneous abnormal detection of multiple sensors of the medical linear accelerator.

[0027] Embodiment 1 Figure 1 The figure is a flowchart of a method for abnormal detection of a medical linear accelerator provided by Embodiment 1 of the present invention. As Figure 1 shown, the method includes: S101. Obtain the data to be measured collected by at least two sensors of the medical linear accelerator.

[0028] The sensors on the medical linear accelerator may include a dose monitoring sensor, a position sensor, a temperature sensor, a pressure sensor, a radiation detector, a gas sensor, etc. The data to be measured may include the dose of the ray detected by the dose monitoring sensor, the position of the ray irradiation measured by the position sensor, the position movement of components such as the treatment couch and collimator of the monitoring accelerator, the temperature of the key components of the accelerator and the environmental temperature of the treatment room monitored by the temperature sensor, the vacuum degree of the vacuum system and the pressure distribution of the treatment couch monitored by the pressure sensor, the intensity and energy of the ray monitored by the radiation detector, and the concentration of harmful gases or special gases in the treatment room monitored by the gas sensor.

[0029] Since the anomalies in a medical linear accelerator may be caused by multiple factors jointly, in this embodiment, the measured data collected by at least two sensors in the medical linear accelerator can be acquired and used as the input for the subsequent anomaly detection model, so as to correlate sensor data across dimensions, enabling the anomaly detection model to better capture the anomaly features in multi-dimensional data.

[0030] S102. Convert the measured data collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the acquisition value of one sensor at a certain time point.

[0031] In this embodiment, the measured data collected by at least two sensors can be stitched together. That is, the measured data collected by one sensor can be used as a row or a column of data in the two-dimensional image data. Then, after stitching together the measured data collected by at least two sensors, a two-dimensional data matrix can be obtained, and further the two-dimensional data matrix can be converted into two-dimensional image data.

[0032] It should be noted that the measured data collected by one sensor may include multiple data points arranged in chronological order. After stitching together the measured data collected by at least two sensors, each time point corresponds to the data collected by different sensors. When the two-dimensional data matrix is converted into two-dimensional image data, the acquisition value of one sensor at a certain time point corresponds to a pixel value in the two-dimensional image data.

[0033] S103. Input the two-dimensional image data obtained by converting the measured data into the trained anomaly detection model, and output the anomaly detection result.

[0034] It should be noted that the anomaly detection model can be a deep learning model for processing image data. This deep learning model can use the two-dimensional image data obtained by converting the measured data collected by at least two sensors in the medical linear accelerator as training samples, and train this deep learning model in advance, enabling this deep learning model to learn the relationship between the combination of data in different dimensions and different types of anomalies. Furthermore, after training, the anomaly detection model can identify whether a certain type of anomaly occurs in the medical linear accelerator according to the input two-dimensional image data.

[0035] In this embodiment, the two-dimensional image data obtained by converting the measured data collected in real time by at least two sensors in the medical linear accelerator can be input into the trained anomaly detection model, and thus the anomaly result of the current medical linear accelerator can be identified.

[0036] The technical solution of the embodiment of the present invention obtains data collected by at least two sensors, and converts the data of at least two sensors into two-dimensional image data, so as to cross-dimensionally associate the sensor data, and enable the anomaly detection model to better capture the anomaly features in the multi-dimensional data, thereby improving the anomaly detection accuracy and reducing the false alarm rate and missed alarm rate.

[0037] Embodiment 2 Figure 2 It is a flowchart of a method for detecting anomalies in a medical linear accelerator provided by Embodiment 2 of the present invention. As Figure 2 shown, the method includes: S201. Obtain historical data collected by at least two sensors of a medical linear accelerator for multiple patients.

[0038] In this embodiment, historical data collected by multiple sensors in a medical linear accelerator during radiotherapy for different patients can be obtained, and the historical data corresponding to each patient is a data sample. Among them, historical data represents data that has been collected.

[0039] It should be noted that in this embodiment, the historical data may include the dose of the ray detected by the dose monitoring sensor, the position of the ray irradiation measured by the position sensor, the position movement of components such as the accelerator treatment couch and collimator monitored, the temperature of key components of the accelerator monitored by the temperature sensor and the ambient temperature of the treatment room, the degree of vacuum of the vacuum system and the pressure distribution of the treatment couch monitored by the pressure sensor, the intensity and energy of the ray monitored by the radiation detector, and the concentration of harmful gases or special gases in the treatment room monitored by the gas sensor. Each item of data in the historical data corresponds to a timestamp, that is, the historical data is time-domain data.

[0040] In one embodiment, it further includes: Preprocess the historical data; The preprocessing at least includes aligning the historical data collected by at least two sensors in chronological order; if the data cannot be aligned, the historical data collected by at least one sensor is interpolated and then the historical data collected by at least two sensors is aligned in chronological order.

[0041] It should be noted that the historical data includes multiple items of data collected by multiple sensors, and the duration of each item of data in the historical data of each data sample is the same. In this embodiment, the items of data in each data sample can be aligned in chronological order according to the time sequence. For an item of data that cannot be aligned, for example, if there is a data missing at a certain time point in an item of data, interpolation processing can be used to fill in the data at that time point.

[0042] S202. Convert the historical data collected by at least two sensors into two-dimensional image data. Each point in the two-dimensional image data represents the pixel value corresponding to the collected value of one sensor at a certain time point.

[0043] In this embodiment, the historical data collected by multiple sensors can be spliced. That is, the data collected by one sensor can be used as a row or a column of data in the two-dimensional image data. After splicing the data collected by multiple sensors, a two-dimensional data matrix can be obtained, and then the two-dimensional data matrix can be further converted into two-dimensional image data.

[0044] It should be noted that the data collected by one sensor can include multiple data points arranged in chronological order. After splicing the data collected by multiple sensors, each time point corresponds to the data collected by different sensors. When the two-dimensional data matrix is converted into two-dimensional image data, the collected value of one sensor at a certain time point corresponds to a pixel value in the two-dimensional image data.

[0045] In one embodiment, it specifically includes: Perform normalization processing on the historical data at each moment collected by at least two sensors to obtain the gray value of the historical data at each moment, so that the historical data of each sensor is converted into a series of gray value points arranged in chronological order. Splice the historical data of at least two sensors after normalization processing based on a preset order to obtain the two-dimensional image data. Each row in the two-dimensional image data represents the gray value corresponding to the historical data of one sensor at different times, and each column represents the gray value corresponding to the historical data of different sensors at a certain time.

[0046] In this embodiment, the historical data can be normalized. One method of normalization is to obtain the maximum value S nmax and the minimum value S nmin of the nth sensor of the historical data. Then the proportion p of the value S ni of the nth sensor at the i-th moment in the historical interval is:

[0047] Then use linear scaling to scale it to the corresponding gray value of the picture, and the gray value range is [0, 255]. Then the gray value S ni corresponding to the value S ni of the nth sensor at the i-th moment is expressed as:

[0048] Based on the above formula, the data at each moment collected by each sensor can be converted into gray values.

[0049] The pixel values obtained by converting the data collected by each sensor are spliced in sequence from top to bottom according to the preset sensor sorting order, and two-dimensional image data can be obtained. The two-dimensional image data can be specifically as Figure 3 shown Figure 3 The curve graph on the left in Figure 3 shows the data collected by one sensor. In the graph, the horizontal axis is the time axis and the vertical axis is the magnitude of the gray value; Figure 3 The middle bar gray graph in

[0050] is a sequence of gray values arranged in time order after the data collected by one sensor is converted into gray values;

[0051] The graph on the right in

[0052] is the two-dimensional image data. Each row in the two-dimensional image data represents a sequence of gray values obtained after the data collected by one sensor is converted. Each row in the two-dimensional image data has the same timestamp, and different rows represent sequences of gray values obtained after the data collected by different sensors are converted.

[0053] S203. Label the two-dimensional image data converted from the historical data. The type of label is the type of abnormality of the medical linear accelerator.

[0054] The abnormal types of the medical linear accelerator corresponding to different two-dimensional image data can be labeled. For example, the two-dimensional image data corresponding to the data collected when the medical linear accelerator is operating normally can be labeled with the type "normal"; the two-dimensional image data corresponding to the data collected when the medical linear accelerator is operating abnormally can be labeled with the type "abnormal"; of course, the specific types of abnormal types can also be labeled.

[0055] In one embodiment, the anomaly detection model is a deep learning model for processing images; The anomaly detection model includes a pre-trained ResNet model, a fully connected layer, and a classifier; The pre-trained ResNet model is used to extract features from the two-dimensional image data; The extracted features are input into the fully connected layer, and the fully connected layer outputs a feature map to the classifier for classification, and the classification result is output.

[0056] In this embodiment, the anomaly detection model can adopt a deep learning model for processing images. Specifically, transfer learning can be used, and a pre-trained model is fine-tuned. The pre-trained model adopts a feature extraction module of ResNet50 to extract features of two-dimensional image data. After the feature extraction module, a fully connected layer and a classifier are sequentially connected. After the features of the two-dimensional image data are input into the fully connected layer, the input two-dimensional image data is classified through the softmax layer. The specific structure of the anomaly detection model is as Figure 4 shown.

[0057] In one embodiment, cross-entropy loss can be used as the loss function; a relatively small learning rate is selected to avoid large-scale modification of the pre-trained weights; the batch size can be set to 32; the number of training epochs can be set to 10, which can be adjusted specifically according to the size of the dataset and the training situation. In addition, during the training process, the performance of the test set needs to be monitored, and early stopping is used to avoid overfitting. According to the test results, hyperparameters such as the learning rate and batch size are adjusted and re-trained. Finally, the trained model is frozen and saved.

[0058] S206. Obtain the data to be measured collected by at least two sensors of the medical linear accelerator.

[0059] In one embodiment, it further includes: preprocessing the data to be measured; The preprocessing at least includes aligning the data to be measured collected by at least two sensors in chronological order; if the data cannot be aligned, the data to be measured collected by at least one sensor is interpolated, and then the data to be measured collected by at least two sensors is aligned in chronological order.

[0060] It should be noted that the data to be measured includes multiple items of data collected by multiple sensors, and the duration of each item of data in each data sample is the same. In this embodiment, the data of each item in each data sample can be aligned in chronological order. For one item of data that cannot be aligned, for example, if there is a missing data point in one item of data, interpolation processing can be used to fill in the data at that time point.

[0061] S207. Convert the data to be measured collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the acquisition value of a sensor at a certain time point.

[0062] In one embodiment, it specifically includes: Normalize the measured data at each moment collected by at least two sensors to obtain the gray value of the measured data at each moment, so that the measured data of each sensor is converted into a series of gray value points in chronological order; Splice the measured data of at least two sensors after normalization based on a preset order to obtain the two-dimensional image data; each row in the two-dimensional image data represents the gray value corresponding to the measured data of a sensor at different moments, and each column represents the gray value corresponding to the measured data of different sensors at one moment.

[0063] In this embodiment, the method of normalizing the measured data can be the same as that of normalizing the historical data, and the method of splicing the measured data of multiple sensors after normalization based on a preset order to obtain two-dimensional image data is also the same.

[0064] S208: Input the two-dimensional image data after conversion of the measured data into the trained anomaly detection model, and output the anomaly detection result.

[0065] In this embodiment, the trained anomaly detection model can be deployed on the server. After the medical device runs once, the data collected by the sensor can be directly transmitted to the model server in real time, so that the model server can perform inference on the real-time sensor data and identify the anomaly category of the linear accelerator. If it is classified into the normal category, there is no subsequent processing; if it is classified into the abnormal category, an alarm is given and the abnormal category is displayed.

[0066] The present invention comprehensively considers all sensor data of the medical linear accelerator, can capture cross-dimensional associated anomalies, identify complex anomaly patterns, improve the accuracy of anomaly detection, and reduce the false alarm rate and missed alarm rate; moreover, the present invention can fuse multi-dimensional time-series data by splicing them into images, and then can be processed using an image deep learning model without training multiple models.

[0067] Embodiment III Figure 5 It is a schematic structural diagram of an anomaly detection device for a medical linear accelerator provided in Embodiment III of the present invention. As Figure 5 shown, the device includes: An acquisition unit 501, configured to acquire measured data collected by at least two sensors of a medical linear accelerator; A conversion unit 502, configured to convert the measured data collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the acquisition value of a sensor at a time point; Anomaly detection unit 503 is configured to input the two-dimensional image data obtained by converting the data to be measured into a trained anomaly detection model, and output an anomaly detection result.

[0068] The medical linear accelerator anomaly detection device provided by the embodiments of the present invention can implement the medical linear accelerator anomaly detection device provided by any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the implementation method.

[0069] Embodiment 4 Figure 6 FIG. shows a schematic structural diagram of an electronic device 10 that can be used to implement the embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as, for example, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, for example, a personal digital processor, a cellular phone, a smart phone, a wearable device (such as a helmet, glasses, a watch, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0070] As Figure 6 shown, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. The memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0071] A plurality of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0072] The processor 11 may be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a method for detecting anomalies in a medical linear accelerator.

[0073] In some embodiments, a method for detecting anomalies in a medical linear accelerator may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for detecting anomalies in a medical linear accelerator described above may be executed. Alternatively, in other embodiments, the processor 11 may be configured to execute a method for detecting anomalies in a medical linear accelerator by any other suitable means (e.g., by means of firmware).

[0074] The various embodiments of the systems and techniques described above in this document may be implemented in digital electronic circuitry, integrated circuit systems, field-programmable gate arrays (FPGA), application-specific integrated circuits (ASIC), application-specific standard products (ASSP), systems-on-a-chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs that may be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0075] The computer program for implementing the method of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowchart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0076] In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0077] To provide for interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide for interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, speech input, or tactile input).

[0078] The systems and techniques described herein can be implemented in a computing system that includes a back-end component (e.g., as a data server), or a computing system that includes a middleware component (e.g., an application server), or a computing system that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0079] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The client-server relationship is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0080] It should be understood that various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved, and no limitation is imposed herein.

[0081] The above specific embodiments do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for detecting anomalies in a medical linear accelerator, characterized in that, Including: Obtaining measurement data collected by at least two sensors of a medical linear accelerator; Converting the measurement data collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the collected value of one sensor at a certain time point; Inputting the two-dimensional image data after conversion of the measurement data into a trained anomaly detection model to output an anomaly detection result.

2. The abnormal detection method of the medical linear accelerator according to claim 1, wherein After obtaining the measurement data collected by at least two sensors of the medical linear accelerator, it further includes: Preprocessing the measurement data; The preprocessing at least includes data alignment of the measurement data collected by at least two sensors in chronological order; if the data cannot be aligned, interpolation processing is performed on the measurement data collected by at least one sensor, and then the measurement data collected by at least two sensors are aligned in chronological order.

3. The abnormal detection method of the medical linear accelerator according to claim 1, wherein, The converting the measurement data collected by at least two sensors into two-dimensional image data includes: Performing normalization processing on the measurement data collected by at least two sensors at each moment to obtain the gray value of the measurement data at each moment, so that the measurement data of each sensor are all converted into a series of gray value points in chronological order; Stitching the measurement data of at least two sensors after normalization processing based on a preset order to obtain the two-dimensional image data; each row in the two-dimensional image data represents the gray value corresponding to the measurement data of one sensor at different times, and each column represents the gray value corresponding to the measurement data of different sensors at a certain time.

4. The abnormal detection method of the medical linear accelerator according to claim 1, characterized in that It further includes: Obtaining historical data collected by at least two sensors of multiple patients passing through a medical linear accelerator; Converting the historical data collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the collected value of one sensor at a certain time point; Labeling the two-dimensional image data converted from the historical data, and the type of labeling is the anomaly type of the medical linear accelerator; Taking the two-dimensional image data corresponding to one patient as a sample to construct a data set; Training the constructed anomaly detection model based on the data set to obtain the trained anomaly detection model.

5. The abnormal detection method of the medical linear accelerator according to claim 4, wherein, After obtaining the historical data collected by at least two sensors of multiple patients passing through a medical linear accelerator, it further includes: Preprocessing the historical data; The preprocessing at least includes data alignment of the historical data collected by at least two sensors in chronological order; if the data cannot be aligned, interpolation processing is performed on the historical data collected by at least one sensor, and then the historical data collected by at least two sensors are aligned in chronological order.

6. The abnormal detection method of the medical linear accelerator according to claim 4, wherein, The converting the historical data collected by at least two sensors into two-dimensional image data includes: Performing normalization processing on the historical data collected by at least two sensors at each moment to obtain the gray value of the historical data at each moment, so that the historical data of each sensor are all converted into a series of gray value points in chronological order; Stitch the historical data of at least two sensors after normalization based on a preset order to obtain the two-dimensional image data; each row in the two-dimensional image data represents the gray values corresponding to the historical data of a sensor at different times, and each column represents the gray values corresponding to the historical data of different sensors at a certain time.

7. The abnormal detection method of the medical linear accelerator according to claim 4, wherein The anomaly detection model is a deep learning model for processing images; The anomaly detection model includes a pre-trained ResNet model, a fully connected layer, and a classifier; The pre-trained ResNet model is used to extract features from the two-dimensional image data; The extracted features are input into the fully connected layer, and the fully connected layer outputs a feature map to the classifier for classification, and outputs a classification result.

8. An abnormal detection device for a medical linear accelerator, characterized in that, It includes: An acquisition unit for acquiring measurement data collected by at least two sensors of a medical linear accelerator; A conversion unit for converting the measurement data collected by at least two sensors into two-dimensional image data; each point in the two-dimensional image data represents the pixel value corresponding to the acquisition value of a sensor at a certain time point; An anomaly detection unit for inputting the two-dimensional image data after converting the measurement data into a trained anomaly detection model and outputting an anomaly detection result.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the medical linear accelerator anomaly detection method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a processor to implement the medical linear accelerator anomaly detection method according to any one of claims 1-7 when executed.

Images