High-power medical CT scanning device

Abstract

The invention discloses a high-power medical CT scanning device, and relates to the technical field of medical CT, the high-power medical CT scanning device comprises an X-ray generation system, a scanning system, a detector system, a data acquisition system and a computer system; the X-ray generation system comprises a high-power electron beam and an X-ray target assembly, the X-ray target assembly comprises a target belt, a main transmission roller and a driven roller, the target belt is connected to the main transmission roller in a transmission mode, and the main transmission roller drives the target belt to move; a cooling medium is introduced into the driven roller to carry out surface contact cooling on the target belt, the high-power electron beam bombards the target belt to generate X-rays, and the solid target belt is rapidly cooled through surface contact by introducing the cooling medium into the driven roller. According to the medical CT scanning device, electron beam deposition heat can be effectively dispersed, the power of the medical CT scanning device is greatly improved, and the problem that a traditional solid target belt is damaged due to insufficient heat dissipation is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical CT, and more specifically, to a high-power medical CT scanning device. Background Art

[0002] Medical CT scanner is an important medical imaging device and an important tool for disease diagnosis and treatment evaluation in the medical field. It uses X-rays to perform tomographic scans on the human body, receives X-ray signals passing through the human body through detectors, and reconstructs a three-dimensional image of the human body through computer processing. It can clearly display the structure of various parts of the human body and help doctors accurately diagnose diseases such as tumors and cardiovascular diseases.

[0003] At present, the technology of medical CT scanners is constantly developing, and high-power medical CT scanners with strong penetration ability, fast scanning speed and high imaging quality are becoming a development trend. High-power medical CT scanners often require high-power X-rays. High-power X-rays have the following advantages: strong penetration ability, can easily penetrate thicker or denser parts of the human body, obtain clear deep tissue images, help to find hidden lesions; can shorten the scanning time, reduce motion artifacts caused by patient body movement, etc., and improve image accuracy; can support more special imaging technologies, such as CT angiography, clearly present vascular morphology and lesions, which is conducive to improving diagnostic efficiency and accuracy, and can better meet clinical diagnosis and treatment needs. However, when high-power medical CT scanners use high-power X-rays, most of the electron beam power will eventually be deposited and accumulated in the target material in the form of heat. When the electron beam power is too large, the temperature of the target will rise sharply, and the thermal stress will increase accordingly, which will cause damage to the target layer. The damage to the target layer will directly affect the quality and stability of X-ray generation, making the uniformity of X-ray emission worse, causing artifacts in CT images, and thus reducing the diagnostic accuracy of medical CT scanners. At the same time, the increase in thermal stress may also cause the physical structure of the target material to deform, affecting its coordination with other components. In addition, when medical CT scanners use high-power X-rays and are in high-temperature operation for a long time and cannot be effectively and intelligently controlled to dissipate heat, it will affect the performance and life of the entire CT scanner, causing many inconveniences and potential risks to the hospital's diagnosis and treatment. Summary of the invention

[0004] The purpose of the present invention is to provide a high-power medical CT scanning device, which aims to partially or completely solve the problems raised in the above-mentioned background technology, and can disperse the heat deposited by the electron beam while using a high-power X-ray source, thereby reducing the damage caused by insufficient heat dissipation of the traditional solid target belt. In order to achieve the purpose of the present invention, the present invention provides the following technical solutions:

[0005] A high-power medical CT scanning device comprises: an X-ray generating system, a scanning system, a detector system, a data acquisition system and a computer system; the X-ray generating system is connected to the scanning system, the scanning system is connected to the detector system, the detector system is connected to the data acquisition system, and the data system is connected to the computer system;

[0006] The X-ray generating system includes a high-power electron beam and an X-ray target assembly. The X-ray target assembly includes a target belt, a main drive roller, and a driven roller. The target belt is drivably connected to the main drive roller, and the main drive roller drives the target belt to move. A cooling medium is passed into the driven roller to perform surface contact cooling on the target belt. The high-power electron beam bombards the target belt to generate X-rays.

[0007] The scanning system performs mobile scanning on the scanned object; the detector system is used to receive the X-rays after passing through the scanned object; the data acquisition system collects the information output by the detector system and outputs it to the detector system; the computer control system receives the information collected by the detector system and reconstructs the collected information into a tomographic image of the scanned object.

[0008] Optionally, the target tape is a solid target tape, and the solid target tape is made of tungsten.

[0009] Optionally, the driven roller is made of copper alloy.

[0010] Optionally, the medical CT scanning device further includes: a cooling control system, which adaptively adjusts the flow rate of the cooling medium according to the temperature of the solid target tape.

[0011] Optionally, the cooling control system includes: a target band temperature data acquisition module, which acquires the time queue of the collected target band real-time temperature values; a target band temperature local time series feature extraction module, which performs time series feature extraction based on the local time scale on the time queue of the target band real-time temperature values ​​to obtain a sequence of local time series implicitly associated feature vectors of the target band real-time temperature; a target band temperature time series feature full time domain aggregation module, which performs feature propagation aggregation based on time series gradient attenuation on the sequence of local time series implicitly associated feature vectors of the target band real-time temperature to obtain a target band real-time temperature principal component time series propagation reasoning vector; and a control instruction generation module, which generates a control instruction for the cooling medium flow rate based on the target band real-time temperature principal component time series propagation reasoning vector.

[0012] Optionally, the target band temperature local time series feature extraction module includes: a local sampling unit, which performs local sampling on the time queue of the target band real-time temperature value based on a predetermined sampling window to obtain a sequence of local sampling windows of the target band real-time temperature value; a temperature local time series feature extraction unit, which inputs the local sampling windows of each target band real-time temperature value in the sequence of local sampling windows of the target band real-time temperature value into a temperature local time series feature extractor based on a forward long short-term memory network model to obtain a sequence of target band real-time temperature local time series implicitly associated feature vectors.

[0013] Optionally, the target band temperature time series feature full time domain aggregation module includes:

[0014] The current target zone temperature characteristic principal component extraction unit extracts the current target zone real-time temperature local time series implicit correlation characteristic vector from the sequence of the target zone real-time temperature local time series implicit correlation characteristic vector, and performs principal component extraction based on the characteristic value to obtain the current target zone real-time temperature local time series characteristic principal component representation vector;

[0015] The historical target band temperature characteristic principal component extraction unit defines other target band real-time temperature local time series implicitly associated characteristic vectors in the sequence of target band real-time temperature local time series implicitly associated characteristic vectors as historical target band real-time temperature local time series implicitly associated characteristic vectors to obtain a sequence of historical target band real-time temperature local time series implicitly associated characteristic vectors, and performs principal component extraction based on eigenvalues ​​on each historical target band real-time temperature local time series implicitly associated characteristic vector in the sequence of historical target band real-time temperature local time series implicitly associated characteristic vectors to obtain a sequence of historical target band real-time temperature local time series characteristic principal component representation vectors;

[0016] A time series gradient attenuation weight calculation unit calculates the time series gradient attenuation weight of each historical target band real-time temperature local time series characteristic principal component representation vector in the sequence of historical target band real-time temperature local time series characteristic principal component representation vectors relative to the current target band real-time temperature local time series characteristic principal component representation vector to obtain a sequence of historical target band temperature local characteristic time series gradient attenuation weights;

[0017] The historical information dynamic propagation aggregation unit calculates the weighted sum of the sequence of principal component representation vectors of the local temporal characteristics of the historical target zone temperature based on the sequence of temporal gradient attenuation weights of the local temporal characteristics of the historical target zone temperature to obtain the principal component significant transfer aggregation representation vector of the local temporal characteristics of the historical target zone temperature;

[0018] The target zone temperature principal component feature full time domain aggregation unit calculates the positional sum of the historical target zone real-time temperature local feature principal component significant transfer aggregation representation vector and the current target zone real-time temperature local time series feature principal component representation vector to obtain the target zone real-time temperature principal component time series propagation inference vector.

[0019] Optionally, the control instruction generation module inputs the target belt real-time temperature principal component time series propagation reasoning vector into the decoder-based cooling medium flow dynamic control module to obtain a control instruction, and the control instruction includes: a recommended value of the cooling medium flow value at the next time point.

[0020] The beneficial effects of the application of the present invention are as follows:

[0021] (1) In the present invention, a high-power medical CT scanning device uses a high-power electron beam, which bombards a solid target tape to generate X-rays. The main drive roller drives the target tape to move so that the heat deposited by the electron beam does not continuously act on a point on the target disk. The driven roller quickly cools the target tape by surface contact by passing a cooling medium. In this way, the heat deposited by the electron beam can be effectively dispersed, the power of the X-ray source is greatly improved, and the problem of damage to the traditional solid target tape due to insufficient heat dissipation is solved;

[0022] (2) In the present invention, a high-power medical CT scanning device uses an intelligent cooling control system and adopts a deep learning-based data analysis technology to perform fine-grained time series analysis on the temperature data of the target belt, so as to capture the time series variation characteristics of the temperature data of the solid target belt in each local time domain, and then obtain the temperature variation information of the whole time domain by performing information propagation aggregation on the temperature characteristics of the target belt in each local time domain, and based on this, intelligent recommendation of the cooling medium flow rate is performed. In this way, it is possible to respond to the change of the temperature of the target belt in real time, dynamically and adaptively adjust the flow rate of the cooling medium, improve the accuracy of the cooling medium flow rate control, and ensure that the temperature of the target belt is always within a safe range under different working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings of the present invention. Obviously, the drawings described below only relate to some embodiments of the present invention, rather than limiting the present invention.

[0024] Figure 1 This is a schematic structural diagram of a high-power medical CT scanning device applied for by the present invention.

[0025] Figure 2 A schematic diagram of the structure of an X-ray target assembly for a high-power medical CT scanning device applied for by the present invention Figure 1 .

[0026] Figure 3 The second structural diagram of an X-ray target assembly of a high-power medical CT scanning device applied for by the present invention.

[0027] Figure 4A schematic diagram of the working principle of a cooling control system for a high-power medical CT scanning device applied for by the present invention.

[0028] Figure 5 This is a schematic diagram of the components of the cooling control system applied for in the present invention. DETAILED DESCRIPTION

[0029] The technical solutions in the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the present invention, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present invention.

[0030] In the existing field of medical CT technology, a medical CT scanner generally includes an X-ray generating system, a detector system, a data acquisition system, a scanning system and a computer system. The X-ray generating system generally includes a cathode and an anode, which are used for the electron beam to hit the anode target surface under the acceleration of a high-voltage electric field to generate X-rays. The cathode and the anode: The cathode mainly includes a filament. When current passes through the filament, the filament heats up to generate an electron cloud. The anode is the target of electron impact, and its target surface is usually made of a material with a high atomic number and a high melting point (such as tungsten or tungsten alloy); the detector system is used to receive the X-rays after passing through the scanned object; the data acquisition system: and the detector system, are used to collect the signals output by the detector system; the scanning system, which usually includes a scanning frame and a scanning bed, is used to support the X-ray generating system and the detector system, enabling them to perform precise rotational motion around the scanned object; the computer system is used to control the coordinated work of the X-ray generating system, the detector system, the data acquisition system, and the scanning system, and reconstruct the data collected by the detector into a tomographic image, which can help doctors observe the tomographic image from different angles and methods, and more accurately analyze the location, size, shape and relationship of the lesion with the surrounding tissue. These are common technical knowledge in the existing field of medical CT technology.

[0031] like Figures 1 to 5 As shown, the present invention proposes a high-power medical CT scanning device, including: an X-ray generating system 100, a scanning system 200, a detector system 300, a data acquisition system 400 and a computer system 500; the X-ray generating system 100 is connected to the scanning system 200, the scanning system 200 is connected to the detector system 300, the detector system 300 is connected to the data acquisition system 400, and the data acquisition system 400 is connected to the computer system 500;

[0032] The X-ray generating system includes a high-power electron beam 104 and an X-ray target assembly. The X-ray target assembly includes a target belt 105, a main drive roller, and a driven roller. The target belt 105 is drivably connected to the main drive roller, and the main drive roller drives the target belt to move. A cooling medium is passed through the driven roller to perform surface contact cooling on the target belt. The high-power electron beam 104 bombards the target belt 105 to generate X-rays.

[0033] The scanning system performs mobile scanning on the scanned object; the detector system is used to receive the X-rays after passing through the scanned object; the data acquisition system collects the information output by the detector system and outputs it to the detector system; the computer control system receives the information collected by the detector system and reconstructs the collected information into a tomographic image of the scanned object.

[0034] like Figure 2 , Figure 3 As shown, in the present application, the X-ray target assembly includes a main drive roller, a driven roller and a target belt, the main drive roller includes a first drive roller 101, a second drive roller 102 and a third drive roller 103, the driven roller includes a first driven roller 106 and a second driven roller 107, the target belt 105 is located between the first driven roller 106 and the second driven roller 107, the driven roller may include multiple groups, such as 2 groups, 4 groups, 6 groups, etc., the first driven roller 106 and the second driven roller 107 may be driven separately and move independently. Of course, in some embodiments, the first drive roller 101 may also drive the first driven roller 106 to rotate, and the rotation of the first driven roller 106 may also drive the second driven roller 107 to rotate. Similarly, the second drive roller 102 may also drive the first driven roller 106 to rotate, and the rotation of the first driven roller 106 may also drive the second driven roller 107 to rotate.

[0035] like Figure 2 , Figure 3 As shown in the present invention application, a high-power electron beam 104 can generate high-power X-rays by bombarding a target belt 105. The main drive roller drives the target belt to rotate so that the heat deposited by the electron beam will not continuously act on a point of the target disk. The driven roller passes a cooling medium, so that when the target belt contacts the cooling medium surface in the driven roller, the target belt can be quickly cooled. This target belt cooling and heat dissipation method can effectively disperse the heat deposited by the high-power electron beam, reduce the damage to the target belt caused by insufficient heat dissipation, and thus greatly improve the power of the X-rays, thereby obtaining high-power X-rays and high-power medical CT scanning devices.

[0036] In some embodiments, the target belt is a solid target belt, and the solid target belt and the driven roller can be made of materials with high thermal conductivity to ensure good heat exchange efficiency between the cooling medium and the target belt, and to ensure that the target belt can still maintain good thermal conductivity under high-power electron beam bombardment. For example, the solid target belt is made of tungsten, and the driven roller is made of copper alloy.

[0037] However, in the process of cooling the target belt by passing the cooling medium through the driven roller, the cooling effect of the target belt is closely related to the flow rate of the cooling medium. Specifically, a suitable flow rate can ensure a sufficient heat exchange rate and prevent the target belt from being too hot; while too large or too small a cooling medium flow rate may lead to energy waste or insufficient cooling. Traditional cooling medium flow control usually relies on fixed parameter settings and cannot dynamically adapt to the real-time temperature changes of the target belt, thereby affecting the cooling efficiency and the service life of the target belt. The applicant has discovered the above technical problems and provided a corresponding overall solution.

[0038] like Figure 4 and Figure 5 As shown, the medical CT scanning device also includes: a cooling control system 600, which includes: a target band temperature data acquisition module 601, which acquires a time queue of collected target band real-time temperature values; a target band temperature local time series feature extraction module 602, which performs time series feature extraction based on a local time scale on the time queue of target band real-time temperature values ​​to obtain a sequence of local time series implicitly associated feature vectors of target band real-time temperature; a target band temperature time series feature full time domain aggregation module 603, which performs feature propagation aggregation based on time series gradient attenuation on the sequence of local time series implicitly associated feature vectors of target band real-time temperature to obtain a target band real-time temperature principal component time series propagation reasoning vector; and a control instruction generation module 604, which generates a control instruction for the flow rate of cooling medium based on the target band real-time temperature principal component time series propagation reasoning vector.

[0039] In some embodiments, the cooling control system adaptively adjusts the flow rate of the cooling medium according to the temperature of the solid target tape.

[0040] In some embodiments, the target belt temperature data acquisition module 601 acquires a time queue of the target belt real-time temperature values ​​collected by the temperature sensor. It should be understood that in the application of a high-power X-ray source, a large amount of heat will be generated when a high-power electron beam bombards a solid target belt. If the heat cannot be removed in a timely and effective manner, the temperature of the target belt will rise, which may cause damage to the target belt. Therefore, real-time monitoring of the temperature change of the target belt and forming a time queue of temperature values ​​can provide a basis for subsequent temperature data analysis.

[0041] Exemplarily, the cooling control system 600 may include a temperature sensor, which may respectively monitor and acquire the real-time temperature value of the real-time monitoring target belt 105, and the real-time temperature value of the target belt may change with time; the target belt temperature data acquisition module 601 may acquire the real-time temperature value of the target belt acquired based on the temperature sensor, and as the target belt moves gradually over time during its movement, multiple real-time temperature values ​​of the target belt that are sequentially acquired in time may at least form a time queue of the real-time temperature value of the target belt, and multiple real-time temperature values ​​of the target belt that are sequentially acquired in time may at least form a time queue of the real-time temperature value of the target belt, thereby completing the time queue of the real-time temperature value of the target belt acquired by the temperature data acquisition module of the target belt 105.

[0042] In some embodiments, the target belt temperature local time series feature extraction module 602 extracts the time series feature based on the local time scale of the target belt real-time temperature value to obtain a sequence of implicitly associated feature vectors of the local time series of the target belt real-time temperature. It should be understood that the amount of temperature data collected after a long period of work may be very large. Directly processing the entire data set will lead to excessive consumption of computing resources and may affect the real-time performance of the cooling medium flow control. Through the extraction of the time series feature of the local time scale, the change trend of the target belt temperature can be better understood and predicted, which is very important for the subsequent formulation of cooling strategies.

[0043] Optionally, the target band temperature local time series feature extraction module 602 includes: a local sampling unit, which performs local sampling on the time queue of the target band real-time temperature value based on a predetermined sampling window to obtain a sequence of local sampling windows of the target band real-time temperature value; a temperature local time series feature extraction unit, which inputs the local sampling windows of each target band real-time temperature value in the sequence of local sampling windows of the target band real-time temperature value into a temperature local time series feature extractor based on a forward long short-term memory network model to obtain a sequence of target band real-time temperature local time series implicitly associated feature vectors.

[0044] In some embodiments, the local sampling unit is used to perform local sampling on the time queue of the target band real-time temperature value based on a predetermined sampling window to obtain a sequence of local sampling windows of the target band real-time temperature value. It should be understood that, considering that a large amount of temperature data may be included, directly processing such a large-scale data set will not only lead to a large consumption of computing resources and affect the real-time performance of the cooling medium flow control, but also lead to the loss of local detail change information in the time queue of the target band real-time temperature value. Therefore, in order to improve the computing efficiency and ensure that the subtle changes in the target band temperature are fully captured, the present invention application further performs local sampling on the time queue of the target band real-time temperature value based on a predetermined sampling window, so as to refine the granularity of the time series analysis and obtain a sequence of local sampling windows of the target band real-time temperature value, thereby facilitating the parallel processing of local time domain temperature data and learning a more detailed time series feature representation of the target band temperature.

[0045] In some embodiments, the temperature local time series feature extraction unit is used to input the local sampling windows of each target band real-time temperature value in the sequence of local sampling windows of the target band real-time temperature value into the temperature local time series feature extractor based on the forward long short-term memory network model to obtain a sequence of target band real-time temperature local time series implicit correlation feature vectors. It should be understood that in order to fully explore the target band temperature time series change pattern in each local time window, the present invention application adopts the temperature local time series feature extractor based on the forward long short-term memory network model to process the local sampling windows of each target band real-time temperature value in the sequence of local sampling windows of the target band real-time temperature value respectively, so as to utilize the long short-term memory characteristics of the forward long short-term memory network model, fully capture the time series correlation dependency of the temperature data in each local time window, learn the time series change trend and periodic law of the target band temperature, and thus obtain the sequence of target band real-time temperature local time series implicit correlation feature vectors.

[0046] In some embodiments, the target band temperature time series feature full time domain aggregation module 603 performs feature propagation aggregation based on time series gradient attenuation on the sequence of implicitly associated feature vectors of the local time series of the target band real-time temperature to obtain the target band real-time temperature principal component time series propagation reasoning vector. It should be understood that in order to deeply understand the global time series changes of the target band temperature, the present invention further performs feature propagation and aggregation on the time axis for the sequence of implicitly associated feature vectors of the local time series of the target band real-time temperature, so as to integrate the temperature features in each local time domain into a global feature representation, thereby providing more comprehensive data support for the adaptive control of the cooling medium flow rate.

[0047] It should be noted that, considering that in the process of time series analysis and reasoning of target temperature time series data, the influence of early temperature data features on the inference prediction results of temperature state and cooling medium flow at the current time point may gradually weaken as the time span increases. Therefore, in order to better simulate and reason about the full-time domain dynamic changes of target temperature, the present invention proposes a feature propagation aggregation method based on the time gradient attenuation mechanism, which dynamically adjusts the weight of feature propagation according to the time position information of the implicitly associated feature vector of the local time series of each target temperature, so as to ensure that when performing feature propagation aggregation, it can pay more attention to recent temperature changes while retaining global information, thereby improving the time sensitivity and accuracy of cooling medium flow control.

[0048] Optionally, the target band temperature time series feature full time domain aggregation module 603 includes:

[0049] The current target band temperature feature principal component extraction unit 6031 extracts the current target band real-time temperature local time series implicitly associated feature vector from the sequence of the target band real-time temperature local time series implicitly associated feature vector, and performs principal component extraction based on the eigenvalue to obtain the current target band real-time temperature local time series feature principal component representation vector;

[0050] The historical target band temperature feature principal component extraction unit 6032 defines other target band real-time temperature local time series implicitly associated feature vectors in the sequence of target band real-time temperature local time series implicitly associated feature vectors as historical target band real-time temperature local time series implicitly associated feature vectors to obtain a sequence of historical target band real-time temperature local time series implicitly associated feature vectors, and performs principal component extraction based on eigenvalues ​​on each historical target band real-time temperature local time series implicitly associated feature vector in the sequence of historical target band real-time temperature local time series implicitly associated feature vectors to obtain a sequence of historical target band real-time temperature local time series feature principal component representation vectors;

[0051] The time series gradient attenuation weight calculation unit 6033 calculates the time series gradient attenuation weight of each historical target band real-time temperature local time series characteristic principal component representation vector in the sequence of historical target band real-time temperature local time series characteristic principal component representation vectors relative to the current target band real-time temperature local time series characteristic principal component representation vector to obtain a sequence of historical target band temperature local characteristic time series gradient attenuation weights;

[0052] The historical information dynamic propagation aggregation unit 6034 calculates the weighted sum of the sequence of principal component representation vectors of the local temporal characteristics of the historical target belt temperature based on the sequence of temporal gradient attenuation weights of the local temporal characteristics of the historical target belt temperature to obtain the principal component significant transfer aggregation representation vector of the local temporal characteristics of the historical target belt temperature;

[0053] The target band temperature principal component feature full-time domain aggregation unit 6035 calculates the historical target band real-time temperature local feature principal component significant transfer aggregation representation vector and the current target band real-time temperature local time series feature principal component representation vector by position and adds them together to obtain the target band real-time temperature principal component time series propagation inference vector.

[0054] Optionally, the current target band temperature feature principal component extraction unit 6031 calculates the covariance matrix of the implicitly associated eigenvectors of the current target band real-time temperature local timing series to obtain the covariance matrix of the current target band real-time temperature local timing series features; performs eigenvalue decomposition on the current target band real-time temperature local timing series feature covariance matrix to obtain a set of eigenvalues ​​of the current target band real-time temperature local timing series features principal component and a corresponding set of eigenvectors of the current target band real-time temperature local timing series features principal component; selects the current target band real-time temperature local timing series features principal component eigenvector corresponding to the maximum value in the set of eigenvalues ​​of the current target band real-time temperature local timing series features principal component as the current target band real-time temperature local timing series features principal component representation vector.

[0055] Optionally, the time gradient attenuation weight calculation unit 6033 includes: a time propagation absolute attenuation factor calculation subunit, which calculates the time propagation absolute attenuation factor of each historical target band real-time temperature local time series characteristic principal component representation vector in the sequence of historical target band real-time temperature local time series characteristic principal component representation vector relative to the current target band real-time temperature local time series characteristic principal component representation vector, so as to obtain a sequence of time propagation absolute attenuation factors; a time span modulation subunit, which modulates each time propagation absolute attenuation factor in the sequence of time propagation absolute attenuation factors in the time dimension based on the time span between each historical target band real-time temperature local time series characteristic principal component representation vector in the sequence of historical target band real-time temperature local time series characteristic principal component representation vector and the current target band real-time temperature local time series characteristic principal component representation vector, so as to obtain a sequence of time span modulated propagation attenuation factors; a weight screening subunit, which inputs the sequence of time span modulated propagation attenuation factors into the information transfer screening module based on the gating function to obtain a sequence of historical target band temperature local characteristic time gradient attenuation weights.

[0056] In some embodiments, the dot division vector between the principal component representation vector of the local timing characteristics of the real-time temperature of the historical target band and the principal component representation vector of the local timing characteristics of the real-time temperature of the current target band is calculated, and the base-two logarithmic value of the absolute value of each eigenvalue in the dot division vector is calculated to obtain the absolute standard representation vector of the timing propagation attenuation; the dot product vector between the absolute standard representation vector of the timing propagation attenuation and the principal component representation vector of the local timing characteristics of the real-time temperature of the historical target band is calculated, and the exponential function value with base e and the sum of the eigenvalues ​​of the dot product vector as the exponent is calculated to obtain the absolute attenuation factor of the timing propagation.

[0057] In some embodiments, the timestamp difference between the principal component representation vector of the real-time local temperature characteristics of the historical target band and the principal component representation vector of the real-time local temperature characteristics of the current target band is calculated, and the timestamp difference is rounded down to obtain a time span factor; the timing propagation absolute attenuation factor of the principal component representation vector of the real-time local temperature characteristics of the historical target band is divided by the natural exponential function value of the time span factor to obtain a time span modulation propagation attenuation factor.

[0058] In some embodiments, the weight screening subunit, based on a preset threshold, masks the sequence of time span modulated propagation attenuation factors to obtain a sequence of masked time span modulated propagation attenuation factors; and inputs the sequence of masked time span modulated propagation attenuation factors into the Sigmoid function for normalization to obtain a sequence of historical target band temperature local characteristic temporal gradient attenuation weights.

[0059] In summary, in the present invention, the target zone temperature time series feature full time domain aggregation module can use the information time series dynamic propagation formula to process the sequence of implicit correlation feature vectors of the local time series of the target zone real-time temperature to obtain the target zone real-time temperature main component time series propagation reasoning vector, and the information time series dynamic propagation formula includes:

[0060] X={x1,x2,...,x i ,...,x p}

[0061] h i = PCA(x i )

[0062]

[0063] Where X represents the sequence of implicitly associated feature vectors of the local time series of the target zone real-time temperature, p is the number of vectors in the sequence of implicitly associated feature vectors of the local time series of the target zone real-time temperature, and x p represents the implicit correlation feature vector of the local time series of the current target zone real-time temperature, x1, x2 and x i denote the first, second and i-th historical target zone real-time temperature local time series implicit correlation feature vectors in the sequence of historical target zone real-time temperature local time series implicit correlation feature vectors, PCA(·) denotes the principal component analysis function, h i represents the principal component representation vector of the local time series characteristics of the real-time temperature of the i-th historical target zone, h p The principal component representation vector of the local time series characteristics of the current target zone real-time temperature, (·) T represents the transpose of the vector, n represents the length of the implicit correlation feature vector of the local time series of the real-time temperature of the historical target zone, C iThe sample covariance matrix of the implicit correlation feature vector of the local time series of the real-time temperature of the i-th historical target zone is the feature covariance matrix of the local time series of the real-time temperature of the historical target zone, U i represents the matrix composed of the set of principal component eigenvectors of the local time series characteristics of the real-time temperature of the historical target belt obtained by performing eigenvalue decomposition on the covariance matrix of the local time series characteristics of the real-time temperature of the i-th historical target belt, Λ i represents the diagonal matrix composed of the set of principal component eigenvalues ​​of the local time series characteristics of the real-time temperature of the historical target zone obtained by performing eigenvalue decomposition on the covariance matrix of the local time series characteristics of the real-time temperature of the i-th historical target zone, diag is the diag function used to construct the diagonal matrix, λ i1 , i2 , ik and λ im represents the first, second, kth and mth historical target zone real-time temperature local time series characteristic principal component eigenvalues ​​in the set of the i-th historical target zone real-time temperature local time series characteristic principal component eigenvalues, m is the number of eigenvalues ​​in the set of the historical target zone real-time temperature local time series characteristic principal component eigenvalues, u i1 、u i2 、u iz and u im denote the first, second, zth and mth principal component eigenvectors of the local time series characteristics of the real-time temperature of the i-th historical target zone, respectively. argmax(·) denotes the index corresponding to the maximum value. z denotes the index of the maximum value in the set of principal component eigenvalues ​​of the local time series characteristics of the real-time temperature of the historical target zone. represents the jth eigenvalue of the principal component representation vector of the local time series characteristics of the real-time temperature of the i-th historical target zone, represents the j-th eigenvalue of the principal component representation vector of the local time series feature of the current target zone real-time temperature, |·| represents the absolute value, log represents the logarithmic function with base 2, l represents the length of the principal component representation vector of the local time series feature of the current target zone real-time temperature, exp(·) represents the exponential function with base e, d i The absolute attenuation factor of the time series propagation of the principal component of the local time series characteristic of the real-time temperature of the i-th historical target zone, t p and t i They respectively represent the timestamps of the local temporal implicit correlation feature vector of the current target zone real-time temperature and the local temporal implicit correlation feature vector of the i-th historical target zone real-time temperature, Indicates rounding down, e i represents the modulation propagation attenuation factor of the i-th time span, mask(·) represents mask processing, sigmoid represents the Sigmoid function, θ represents the preset threshold, and w irepresents the attenuation weight of the temporal gradient of the local characteristic of the i-th historical target zone temperature, V represents the temporal propagation inference vector of the principal component of the real-time temperature of the target zone, n is a positive integer, which is set according to the actual situation, and n can be equal to or not equal to p.

[0064] In the present invention, firstly, the last time node in the sequence of the local temporal implicit correlation feature vector of the target zone real-time temperature is regarded as the current time node, and the other time nodes are regarded as historical nodes. The PCA analysis is performed on the local temporal implicit correlation feature vector of the target zone real-time temperature at each time node to remove the noise interference in the data, highlight the key information, and extract the most representative principal component; then, in order to capture the trend of the target zone temperature characteristics of each historical time node over time and the information loss, the characteristic principal component change rate of each historical time node relative to the current time node and its corresponding time series propagation absolute attenuation factor are further calculated to enhance the model. In order to improve the ability to understand information transmission over long time intervals, the time span factor is introduced to adjust the above-mentioned absolute attenuation factor of time series propagation to simulate the gradual forgetting process of data in time, so that the model can pay more attention to the recent temperature change trend and reduce excessive dependence on early data; then, based on the gating mechanism, the absolute attenuation factor of time series propagation after time span modulation is screened, the most influential propagation path is identified and the corresponding weight is assigned accordingly, so as to perform weighted aggregation of the principal component characteristics of each time node, thereby effectively integrating the long-term change trend and short-term fluctuation of the target belt temperature, and generating the target belt real-time temperature principal component time series propagation inference vector. In this way, an in-depth understanding of the global time domain dynamic changes of the target belt temperature can be effectively achieved, providing a strong information basis for the intelligent recommendation of the cooling medium flow rate.

[0065] Optionally, the control instruction generation module 604 inputs the target belt real-time temperature principal component time series propagation reasoning vector into the decoder-based cooling medium flow dynamic control module to obtain a control instruction, which includes: a recommended value of the cooling medium flow value at the next time point.

[0066] In some embodiments, the target belt real-time temperature principal component time series propagation inference vector is input into a decoder-based cooling medium flow dynamic control module to obtain a control instruction, including: using the decoder to decode and regress the target belt real-time temperature principal component time series propagation inference vector to obtain the control instruction, the formula is: Among them, V is the target belt real-time temperature principal component time series propagation inference vector, Y is the control instruction, W is the weight matrix of the decoder, Represents matrix multiplication.

[0067] In the present application, each target band real-time temperature local time series implicit correlation feature vector in the sequence of target band real-time temperature local time series implicit correlation feature vectors respectively represents the local time series implicit correlation feature of the target band real-time temperature value. When it is input into the feature time series dynamic propagation network based on the node feature principal component gradient attenuation, due to the sample repetition between each local sampling window, there will be a more significant feature redundancy expression between the sequence of target band real-time temperature local time series implicit correlation feature vectors, resulting in insufficient heterogeneity between the node feature principal components, and the feature manifold of the target band real-time temperature principal component time series propagation reasoning vector obtained after the dynamic propagation of the time series gradient attenuation has feature distribution imbalance in the high-dimensional feature space, thereby affecting the accuracy of the control instructions obtained by the cooling medium flow rate dynamic control module based on the decoder. Based on this, the target band real-time temperature principal component time series propagation reasoning vector can be optimized, and the optimized target band real-time temperature principal component time series propagation reasoning vector is formed including:

[0068] The target belt real-time temperature principal component time series propagation reasoning vector is randomly sampled L times to obtain L target belt temperature time series characteristic random perturbation eigenvalues; the L target belt temperature time series characteristic random perturbation eigenvalues ​​are ordered from large to small to obtain the target belt temperature time series characteristic random perturbation ordered vector; the target belt real-time temperature principal component time series propagation reasoning vector is calculated to obtain the target belt temperature time series characteristic energy modulation eigenvalue;

[0069] Based on the energy modulation eigenvalue of the target band temperature time series feature, the randomly perturbed ordered vector of the target band temperature time series feature is semantically scaled to obtain the randomly perturbed ordered vector of the semantically enhanced target band temperature time series feature; the product of the randomly perturbed ordered vector of the semantically enhanced target band temperature time series feature and the transposed vector of the target band real-time temperature principal component time series propagation reasoning vector is calculated to obtain the orthogonal modulation matrix of the best sample prompt of the target band temperature time series feature;

[0070] Calculate the matrix product between the best sample prompt orthogonal modulation matrix of the target band temperature time series characteristic and the target band real-time temperature principal component time series propagation reasoning vector to obtain the optimized target band real-time temperature principal component time series propagation reasoning vector;

[0071] The optimized target belt real-time temperature principal component time series propagation reasoning vector is input into the cooling medium flow rate dynamic control module based on the decoder to obtain the control instruction. The optimized target belt real-time temperature principal component time series propagation reasoning vector is formed to include an optimization formula, and the optimization formula includes:

[0072] S i =∈ i ,i=1,2,...,L

[0073]

[0074] u=average(V)

[0075]

[0076] V order =Order{S i}

[0077]

[0078] in, represents a normal distribution, represents the Gaussian density probability function, u and σ 2 They represent the mean and variance of the target zone real-time temperature principal component time series propagation inference vector, ∈ i represents a random variable that follows a normal distribution, L represents the number of random sampling, S i represents the random perturbation eigenvalue of the target belt temperature time series characteristic, V represents the target belt real-time temperature principal component time series propagation inference vector, v i represents the i-th eigenvalue of the target zone real-time temperature principal component time series propagation reasoning vector, average(·) represents the average value of the calculated vector, L1 represents the length of the target zone real-time temperature principal component time series propagation reasoning vector, Order{S i} indicates that the random perturbation characteristic values ​​of the target temperature time series characteristics are arranged in order from large to small, V order represents the ordered vector of random perturbation of the target temperature time series characteristic, e represents the energy modulation characteristic value of the target temperature time series characteristic, represents the square of the vector's two-norm, ⊙ represents the point multiplication by position, represents matrix multiplication, (·) T Represents the transpose of a vector, M t represents the orthogonal modulation matrix of the best example of the target band temperature timing characteristics, and V' represents the optimized target band real-time temperature principal component timing propagation inference vector.

[0079] In the present invention, firstly, a set of new feature value sets based on the original features but containing slight changes is obtained by performing L random sampling on the target band real-time temperature principal component time series propagation inference vector, so as to introduce randomness to explore different aspects of the feature space, and by simulating the noise or variation in the actual data, it is helpful for the model to better generalize to unseen data; then, these target band temperature time series feature random perturbation feature values ​​are arranged in order from large to small to form a target band temperature time series feature random perturbation ordered vector, so as to identify which perturbation directions may be the most important for the current problem, and such an ordering not only helps the algorithm focus on those feature differences that can best distinguish different categories, but also helps to reveal hidden data structures or patterns; Next, the inner product of the target band real-time temperature principal component time-series propagation inference vector is calculated to obtain the target band temperature time-series feature energy modulation eigenvalue. The target band temperature time-series feature energy modulation eigenvalue reflects the overall strength or "energy" of the original feature vector, and provides a benchmark for measuring the importance of the feature. Using this target band temperature time-series feature energy modulation eigenvalue as a reference point, the previously formed target band temperature time-series feature random perturbation ordered vector is semantically scaled to construct a semantically enhanced target band temperature time-series feature random perturbation ordered vector. In this way, those disturbances that appear to be more important in the original feature space are amplified, thereby enhancing the model's sensitivity to changes in key features, while also retaining relatively small but potentially useful disturbance information.

[0080] Then, the product of the randomly perturbed ordered vector of the semantically enhanced target temperature time series features and the transposed vector of the target real-time temperature principal component time series propagation inference vector is calculated to obtain the orthogonal modulation matrix of the best sample prompt of the target temperature time series features. The purpose is to find the way in which the features interact with each other under two different perspectives. The formed matrix can be regarded as a conversion tool for capturing and representing the complex linear relationship between features. Finally, the final feature structure modulation is completed by applying the above-constructed orthogonal modulation matrix of the best sample prompt of the target temperature time series features to the original target real-time temperature principal component time series propagation inference vector, so as to make the features more in line with the requirements of the objective function, reduce redundant information, and maximize the retention of useful information, so as to improve learning efficiency and prediction performance.

[0081] Therefore, the feature manifold optimization process of the target belt real-time temperature principal component time series propagation inference vector is explained, which extracts the best prompt sample based on the fine-grained target retrieval of random perturbations, and then guides the adversarial robustness learning based on the best prompt sample to complete the structure of the high-dimensional manifold of the feature vector in the feature space to improve the global balance of the feature expression. In this way, the accuracy of the control instructions obtained by the decoder-based cooling medium flow rate dynamic control module is improved.

[0082] In the high-power medical CT scanning device applied for by the present invention, firstly, the high-power medical CT scanning device uses a high-power electron beam, which bombards the solid target belt to generate X-rays, and the main drive roller drives the target belt to move, so that the heat deposited by the electron beam will not continuously act on a point of the target disk, and the driven roller quickly cools the target belt by passing the cooling medium through the surface contact. In this way, the heat deposited by the electron beam can be effectively dispersed, the power of the X-ray source is greatly improved, and the problem of damage to the traditional solid target belt due to insufficient heat dissipation is solved; in addition, the high-power medical CT scanning device uses an intelligent cooling control system, and uses a data analysis technology based on deep learning to perform fine-grained time series analysis on the temperature data of the target belt to capture the time series change characteristics of the temperature data of the solid target belt in each local time domain, and then through the information propagation aggregation of the target belt temperature characteristics in each local time domain, to obtain the temperature change information of the whole time domain, and based on this, the intelligent recommendation of the cooling medium flow rate is performed. In this way, it is possible to respond to changes in the target belt temperature in real time, dynamically and adaptively adjust the flow rate of the cooling medium, improve the accuracy of the cooling medium flow rate control, and ensure that the temperature of the target belt is always within a safe range under different working conditions.

[0083] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0084] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0085] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0086] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0087] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A high-power medical CT scanning device, characterized in that: include: X-ray generation system, scanning system, detector system, data acquisition system and computer system; The X-ray generating system is connected to the scanning system, the scanning system is connected to the detector system, the detector system is connected to the data acquisition system, and the data acquisition system is connected to the computer system; The X-ray generating system includes a high-power electron beam and an X-ray target assembly. The X-ray target assembly includes a target belt, a main drive roller, and a driven roller. The target belt is drivably connected to the main drive roller, and the main drive roller drives the target belt to move. A cooling medium is passed into the driven roller to perform surface contact cooling on the target belt. The high-power electron beam bombards the target belt to generate X-rays. The scanning system performs mobile scanning on the scanned object; the detector system is used to receive the X-rays after passing through the scanned object; the data acquisition system collects the information output by the detector system and outputs it to the detector system; the computer control system receives the information collected by the detector system and reconstructs the collected information into a tomographic image of the scanned object.

2. A high-power medical CT scanning device according to claim 1, characterized in that: The target belt is a solid target belt, and the solid target belt is made of tungsten.

3. A high-power medical CT scanning device according to claim 2, characterized in that: The driven roller is made of copper alloy.

4. A high-power medical CT scanning device according to claim 3, characterized in that: The medical CT scanning device also includes: a cooling control system, which adaptively adjusts the flow rate of the cooling medium according to the temperature of the solid target belt.

5. A high-power medical CT scanning device according to claim 4, characterized in that: The cooling control system includes: a target belt temperature data acquisition module, which acquires the time queue of the collected target belt real-time temperature values; a target belt temperature local time series feature extraction module, which performs time series feature extraction based on the local time scale on the time queue of the target belt real-time temperature values ​​to obtain a sequence of local time series implicitly associated feature vectors of the target belt real-time temperature; a target belt temperature time series feature full time domain aggregation module, which performs feature propagation aggregation based on time series gradient attenuation on the sequence of local time series implicitly associated feature vectors of the target belt real-time temperature to obtain a target belt real-time temperature principal component time series propagation reasoning vector; and a control instruction generation module, which generates a control instruction for the cooling medium flow rate based on the target belt real-time temperature principal component time series propagation reasoning vector.

6. A high-power medical CT scanning device according to claim 5, characterized in that: The target band temperature local time series feature extraction module comprises: a local sampling unit, which performs local sampling on the time queue of the target band real-time temperature value based on a predetermined sampling window to obtain a sequence of local sampling windows of the target band real-time temperature value; a temperature local time series feature extraction unit, which inputs the local sampling windows of each target band real-time temperature value in the sequence of local sampling windows of the target band real-time temperature value into a temperature local time series feature extractor based on a forward long short-term memory network model to obtain a sequence of implicitly associated feature vectors of the target band real-time temperature local time series.

7. A high-power medical CT scanning device according to claim 6, characterized in that: The full time domain aggregation module of target temperature time series characteristics includes: The current target zone temperature characteristic principal component extraction unit extracts the current target zone real-time temperature local time series implicit correlation characteristic vector from the sequence of the target zone real-time temperature local time series implicit correlation characteristic vector, and performs principal component extraction based on the characteristic value to obtain the current target zone real-time temperature local time series characteristic principal component representation vector; The historical target band temperature characteristic principal component extraction unit defines other target band real-time temperature local time series implicitly associated characteristic vectors in the sequence of target band real-time temperature local time series implicitly associated characteristic vectors as historical target band real-time temperature local time series implicitly associated characteristic vectors to obtain a sequence of historical target band real-time temperature local time series implicitly associated characteristic vectors, and performs principal component extraction based on eigenvalues ​​on each historical target band real-time temperature local time series implicitly associated characteristic vector in the sequence of historical target band real-time temperature local time series implicitly associated characteristic vectors to obtain a sequence of historical target band real-time temperature local time series characteristic principal component representation vectors; A time series gradient attenuation weight calculation unit calculates the time series gradient attenuation weight of each historical target band real-time temperature local time series characteristic principal component representation vector in the sequence of historical target band real-time temperature local time series characteristic principal component representation vectors relative to the current target band real-time temperature local time series characteristic principal component representation vector to obtain a sequence of historical target band temperature local characteristic time series gradient attenuation weights; The historical information dynamic propagation aggregation unit calculates the weighted sum of the sequence of principal component representation vectors of the local temporal characteristics of the historical target zone temperature based on the sequence of temporal gradient attenuation weights of the local temporal characteristics of the historical target zone temperature to obtain the principal component significant transfer aggregation representation vector of the local temporal characteristics of the historical target zone temperature; The target zone temperature principal component feature full time domain aggregation unit calculates the historical target zone real-time temperature local feature principal component significant transfer aggregation representation vector and the current target zone real-time temperature local time series feature principal component representation vector by position to obtain the target zone real-time temperature principal component time series propagation inference vector.

8. A high-power medical CT scanning device according to claim 7, characterized in that: The control instruction generation module inputs the target belt real-time temperature principal component time series propagation reasoning vector into the decoder-based cooling medium flow dynamic control module to obtain the control instruction, which includes: the recommended value of the cooling medium flow value at the next time point.

Images