Method and system for monitoring events associated with an X-ray tube

Abstract

The present invention is entitled a method and system for monitoring events associated with an X-ray tube. The method of the present invention involves generating one or both of a failure prediction indication of an X-ray tube or an estimate of the remaining useful life of the X-ray tube. In one implementation, the complexity of a regression model is selected based on an operating point utilized by an imaging system of the X-ray tube, wherein the regression model estimates coefficients used by a static tube model to estimate the health (e.g., thickness) of an electron emitter of the X-ray tube, which in turn can be used to predict the remaining useful life of the electron emitter of the X-ray tube. In another implementation, replacement of an X-ray tube or a component of a filament drive circuit coupled to the X-ray tube can be detected.

Description

Background Art

[0001] The subject matter disclosed herein relates to X-ray tubes, and in particular to monitoring events associated with X-ray tubes.

[0002] Non-invasive imaging techniques allow images of internal structures or features of a subject (a patient, manufactured goods, luggage, packages, or passengers) to be obtained non-invasively. In particular, such non-invasive imaging techniques rely on various physical principles (such as differential transmission of X-rays through a target volume) to acquire data and construct an image or otherwise represent the internal features of the subject.

[0003] In such X-ray based non-invasive imaging contexts, an X-ray tube is typically used to generate X-rays that pass through a subject. Examples of imaging systems that employ X-ray tubes include, but are not limited to, the following systems: radiography, mammography, tomosynthesis, C-arm angiography, fluoroscopy, and computed tomography (CT) systems, among others. During an examination or imaging sequence, the X-rays emitted by the X-ray tube in such a system are generated in response to a control signal.

[0004] Typically, an X-ray tube includes a cathode and an anode. An emitter within the cathode may emit a stream of electrons in response to heat generated by an applied current and / or an electric field generated by an applied voltage. The anode may include a target that is impacted by the electron stream. Due to the impact of the electron beam, the target may generate X-ray radiation to be emitted toward the imaging volume.

[0005] In practice, such X-ray tubes have a finite useful life. However, even X-ray tubes of the same type and model may have different useful lives. However, it may be difficult to estimate or predict the useful life of an X-ray tube. As a result, an X-ray tube may be proactively replaced while its useful life remains or may be used until it fails, resulting in unplanned downtime that is inconvenient for patients and staff. Summary of the invention

[0006] Certain embodiments commensurate with the scope of the originally claimed subject matter are summarized below. These embodiments are not intended to limit the scope of the claimed subject matter, but rather, these embodiments are intended only to provide a brief overview of possible forms of the subject matter. In fact, the subject matter may include a variety of forms that may be similar to or different from the embodiments described below.

[0007] In one embodiment, a method for constructing an X-ray tube model for assessing the health of an X-ray tube is provided. The method includes obtaining training data points for the corresponding X-ray tube after the X-ray tube is installed in an imaging system. The method also includes determining a number of operating points used in the imaging system for the corresponding X-ray tube. The method further includes selecting a regression model from a plurality of regression models based on the number of operating points, wherein each regression model uses the training data points to derive corresponding values ​​of a plurality of coefficients.

[0008] In another embodiment, a method for assessing the health of an X-ray tube is provided. The method includes acquiring data points of a corresponding X-ray tube installed in an imaging system. The method also includes processing the data points using a trained X-ray tube model to calculate an estimated variable related to emitter resistance. The method further includes calculating a trend indicator based on the estimated variable related to emitter resistance. The method even further includes monitoring the trend indicator for sudden changes. The method further includes providing an indication when a sudden change occurs on a particular day.

[0009] In another embodiment, a method for monitoring replacement of an X-ray tube or a component of a filament drive circuit coupled to an X-ray tube is provided. The method includes acquiring data points of a corresponding X-ray tube installed in an imaging system. The method also includes processing the data points using a trained X-ray tube model to calculate an estimated variable related to emitter resistance. The method further includes calculating a trend indicator based on the estimated variable related to emitter resistance. The method even further includes monitoring the trend indicator for sudden changes in one direction. The method further includes marking the sudden change in the trend indicator as a replacement event for the X-ray tube or a component of a filament drive circuit coupled to the X-ray tube when the sudden change in the trend indicator is detected. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] These and other features, aspects, and advantages of the present subject matter will be better understood when the following detailed description is read with reference to the accompanying drawings, in which like characters refer to like parts throughout the several views, and in which:

[0011] Figure 1 is a schematic diagram of an embodiment of a computed tomography (CT) system configured to acquire CT images of a patient and process the images according to aspects of the present disclosure;

[0012] Figure 2 shows a block diagram of an X-ray tube according to aspects of the present disclosure;

[0013] Figure 3 An example of a static X-ray tube model describing electrical parameters and their interrelationships according to aspects of the present disclosure is shown;

[0014] Figure 4 shows an example graph plotting an indicator of X-ray tube life in accordance with aspects of the present disclosure;

[0015] Figure 5 shows an example graph plotting a combined indicator of X-ray tube life in accordance with aspects of the present disclosure;

[0016] Figure 6 shows a flow chart illustrating the creation and use of an X-ray tube model for assessing X-ray tube lifetime in accordance with aspects of the present disclosure;

[0017] Figure 7 shows example graphs plotting combined metrics of various X-ray tubes according to aspects of the present disclosure;

[0018] Figure 8 A drawing according to various aspects of the present disclosure is shown. Figure 7 Example curves of the linearized version of the combined indicator;

[0019] Fig. 9 A flow chart illustrating creation and use of an X-ray tube model for assessing X-ray tube life according to aspects of the present disclosure is shown, the model showing an example curve plotting an indicator of X-ray tube life;

[0020] Fig.10 A plot of mA at 5 ms / average mA within a time window showing unstable mA behavior is shown;

[0021] Fig.11 The present disclosure includes Fig.10 Trend indicator of erratic mA behavior shown in;

[0022] Fig.12 A flow chart illustrating a method for constructing an X-ray tube model for assessing the health of an X-ray tube according to aspects of the present disclosure is shown;

[0023] Fig.13 shows an example graph of a linearized combination index of an X-ray tube using a more complex linear regression model according to aspects of the present disclosure;

[0024] Fig.14 shows an example graph of a linearized combination index of an X-ray tube using a simpler linear regression model according to aspects of the present disclosure;

[0025] Fig.15 shows a flow chart illustrating a method for detecting a sudden change in a tube life curve according to aspects of the present disclosure;

[0026] Fig.16A flow chart illustrating a method for correcting sudden changes in a tube life curve and retraining a linear regression model according to aspects of the present disclosure is shown;

[0027] Fig.17 shows an example graph of a linearized combination index (without correction) for an X-ray tube in accordance with aspects of the present disclosure;

[0028] Fig.18 shows an example graph of a linearized combination index (corrected) for an X-ray tube according to aspects of the present disclosure;

[0029] Fig.19 shows an example graph of a linearized combined metric (without retraining) for an X-ray tube in accordance with aspects of the present disclosure;

[0030] Fig. 20 shows an example graph of a linearized combined metric (retrained) for an X-ray tube in accordance with aspects of the present disclosure; and

[0031] Fig.21 A flow chart illustrating a method for monitoring X-ray tube replacement according to aspects of the present disclosure is shown. DETAILED DESCRIPTION

[0032] One or more specific embodiments will be described below. In order to provide a concise description of these embodiments, all features of an actual implementation may not be described in the specification. It should be understood that in the development of any such actual implementation, as in any engineering or design project, many implementation-specific decisions must be made to achieve the developer's specific goals, such as complying with system-related and business-related constraints that may vary from implementation to implementation. In addition, it should be understood that such development efforts may be complex and time-consuming, but are still routine tasks for design, fabrication, and manufacturing for ordinary technicians who benefit from this disclosure.

[0033] When introducing elements of various embodiments of the present subject matter, the articles "a," "an," "the," and "said" are intended to indicate that there are one or more of the recited elements. The terms "comprising," "including," and "having" are intended to be inclusive and mean that there may be additional elements in addition to the listed elements. Furthermore, any numerical examples in the following discussion are intended to be non-limiting, and thus the appended numerical values, ranges, and percentages are within the scope of the disclosed embodiments.

[0034] Although the following discussion is generally provided in the context of medical imaging, it should be understood that the present technology is not limited to such medical contexts. Indeed, the examples and explanations are provided in such medical contexts only to facilitate explanation by providing examples of realistic specific implementations and applications. However, the methods of the present invention may also be used with X-ray tubes in other contexts, such as X-ray tubes used in non-destructive inspection of manufactured parts or goods (i.e., quality control or quality review applications) and / or non-intrusive inspection of packages, boxes, luggage, etc. (i.e., security or screening applications). In general, the methods of the present invention may be desirable in any imaging or screening context that utilizes an X-ray tube.

[0035] As described herein, an X-ray tube, such as an X-ray tube used in a computed tomography (CT) imaging system, may fail for a variety of reasons, including wear failure of a rotor (which rotates an anode that emits X-rays) and / or an electron emitter (e.g., a filament) (which emits electrons toward the anode). As used herein, the phrase "X-ray tube failure" may encompass situations where an emitter of an X-ray tube fails due to a loss of capacity of the X-ray tube, even if other emitters are present. Thus, herein, the phrase "X-ray tube failure" may be understood to encompass failure of an emitter in an X-ray tube, even if the X-ray tube retains some X-ray generating capability. Such failures may be inconvenient for an operator of the imaging system, particularly in the event of an unexpected failure of the X-ray tube while the scanner is in use. According to the methods described herein, an algorithm is disclosed that automatically predicts failure of an X-ray tube and notifies sufficiently in advance to allow replacement components to be ordered and / or maintenance requests to be planned, and allows patient examinations to be appropriately planned so that disruption to the examination is minimized. In an alternative aspect, an algorithm may additionally or alternatively be provided that provides an estimate of the remaining useful life of an X-ray tube based on degradation of an electron emitter (referred to herein as an "emitter" or "electron emitter").

[0036] With respect to emitter failure, as the emitter is used, the emitter evaporates and thereby reduces in thickness. Therefore, a lower emitter current can achieve the temperature required to drive a specified tube current at a specified voltage. By monitoring the evolution of the emitter current required for the target tube current at any tube voltage, an estimate of how much of its life cycle the emitter has completed can be obtained. With this in mind, the method of the present invention involves predicting emitter failure before the emitter fails and estimating the remaining life of the emitter at any stage within the life of the X-ray tube. The method of the present invention is based on modeling the temperature of the emitter, the evaporation rate of the emitter, and the current that needs to be transmitted through the emitter to obtain the desired tube current and tube voltage set points. In one aspect, the algorithm of the present invention can handle all combinations of X-ray tube current (mA) and X-ray tube voltage (kV) set points without the need to monitor multiple indicator curves for different current and voltage set points.

[0037] In one specific implementation, the predictive failure algorithm generates an alarm only when approaching the end of the life of the transmitter. Therefore, the operator is not actively aware of the health of the transmitter until the useful life of the transmitter is about to end. To provide additional information, in another embodiment, an indicator can be provided that continuously informs the operator about the status of the transmitter over time. In one such approach, the corresponding algorithm can calculate the remaining useful life of the transmitter at the corresponding time.

[0038] However, before discussing detailed aspects of the present method, an example of an imaging system to which this method may be employed is described in order to provide a useful background. With this in mind, Figure 1 An embodiment of an imaging system 10 that uses an X-ray tube to acquire and image data according to the methods described herein is shown. In the illustrated embodiment, the system 10 is a computed tomography (CT) system that is designed to acquire X-ray projection data at one or more energy spectra, reconstruct the projection data into volume reconstructions, and process the image data for display and analysis. The CT imaging system 10 includes one or more X-ray sources 12, such as one or more X-ray tubes.

[0039] In some implementations, the X-ray source 12 can be positioned proximate to a filter assembly or beam shaper 22, which can be used to direct the X-ray beam 20, define the shape and / or extent of a high intensity region of the X-ray beam 20, control or define the energy distribution of the X-ray beam 20, and / or otherwise limit X-ray exposure to those portions of the patient 24 that are not within the region of interest. In practice, the filter assembly or beam shaper 22 can be incorporated within the gantry, between the source 12 and the imaging volume within the patient 24.

[0040] An X-ray beam 20 enters an area in which a subject (e.g., a patient 24) or an object of interest (e.g., a manufactured part, luggage, a package, etc.) is located. The subject attenuates at least a portion of the X-rays 20, resulting in attenuated X-rays 26, which impinge on a detector array 28 formed of a plurality of detector elements (e.g., pixels) as discussed herein. Each detector element generates an electrical signal that represents the energy deposition of an incident X-ray photon at the detector element location. The electrical signals are acquired and processed to generate one or more projection data sets. In the depicted example, the detector 28 is coupled to a system controller 30, which commands the acquisition of a digital signal generated by the detector 28.

[0041] The system controller 30 commands the operation of the imaging system 10 to perform filtering, inspection and / or calibration schemes, and processes the acquired data. The system controller 30 provides power, focus position, control signals, etc. for the X-ray inspection sequence relative to the X-ray source 12. According to certain embodiments, the system controller 30 can control the operation of the filter assembly 22, the CT gantry (or other structural support to which the X-ray source 12 and the detector 28 are attached), and / or the translation and / or tilting of the patient support during the inspection process.

[0042] In addition, via the motor controller 36, the system controller 30 can control the operation of the linear positioning subsystem 32 and / or the rotation subsystem 34 for moving components of the imaging system 10 and / or the subject 24. The system controller 30 may include signal processing circuitry and associated memory circuitry. In such embodiments, the memory circuitry may store programs, routines, and / or coded algorithms that are executed by the system controller 30 to operate the imaging system 10 (including the x-ray source 12 and / or the filter assembly 22), process digital measurements acquired by the detector 28, and / or monitor and / or estimate the health or remaining life of the x-ray tube emitter according to the steps and processes discussed herein. In one embodiment, the system controller 30 may be implemented as all or part of a processor-based system.

[0043] The X-ray source 12 may be controlled by an X-ray controller 38 included within the system controller 30. The X-ray controller 38 may be configured to provide the X-ray source 12 with power, timing signals, and / or focal spot size and position.

[0044] The system controller 30 may include a data acquisition system (DAS) 40. The DAS 40 receives data collected by the readout electronics of the detector 28, such as a digital signal from the detector 28. The DAS 40 may then convert and / or process the data for subsequent processing by a processor-based system, such as a computer 42. In certain embodiments discussed herein, circuitry within the detector 28 may convert the analog signal of the photodetector to a digital signal prior to transmission to the data acquisition system 40. The computer 42 may include or be in communication with one or more non-transitory memory devices 46 capable of storing data processed by the computer 42, data to be processed by the computer 42, or instructions to be executed by the image processing circuitry 44 of the computer 42. For example, a processor of the computer 42 may execute one or more sets of instructions stored on a memory 46, which may be an example of a memory of the computer 42, a memory of a processor, firmware, or the like.

[0045] The computer 42 may also be adapted to control features enabled by the system controller 30 (i.e., scanning operations and data acquisition), such as in response to commands and scanning parameters provided by an operator via an operator workstation 48. The system 10 may also include a display 50 coupled to the operator workstation 48 that allows the operator to view relevant system data, imaging parameters, raw imaging data, reconstruction data, base material images, and / or alternative material decomposition images, etc. In addition, the system 10 may include a printer 52 coupled to the operator workstation 48 and configured to print any desired measurement results. The display 50 and printer 52 may also be connected to the computer 42 (e.g., directly or via the operator workstation 48). Figure 1 ). In addition, the operator workstation 48 may include or be coupled to a picture archiving and communication system (PACS) 54. The PACS 54 may be coupled to a remote system or client 56, a radiology department information system (RIS), a hospital information system (HIS), or an internal or external network to enable others at different locations to access the image data.

[0046] Considering the previous discussion of the overall imaging system 10, Figure 2 A block diagram of an X-ray source in the form of an X-ray tube 60 is shown, similar to the X-ray tubes used in computed tomography (CT) machines. As shown, the X-ray tube 60 may include a cathode 62 and an anode assembly 64 enclosed in a housing 66. The anode assembly 64 includes a rotor 68 (e.g., a rotating anode disk) that can rotate an anode 70. The anode 70 and the rotor 68 may rotate about a fixed axis 76.

[0047] During operation of the X-ray tube 60, the anode 70 is electrically connected to the electron emitter structure of the cathode 62 (see Figure 3, emitter 112) when impacted by an electron beam 80 emitted by the emitter 112. In some X-ray tubes, an electrostatic potential difference exceeding 20 kV is formed between the cathode assembly 86 connected to the voltage source 82 and the anode 70. Therefore, the cathode assembly 86 accelerated toward the anode 70 can emit electrons, thereby generating X-rays 20 in response to the incident electron stream.

[0048] The X-ray tube 60 is supported by an anode assembly 64 and a cathode assembly 86, wherein a housing 66 defines a relatively low pressure region (e.g., a hermetically sealed housing). For example, the housing 66 may include glass, ceramic, stainless steel, or other suitable materials. The anode 70 may be made of any metal or composite, such as tungsten, molybdenum, copper, or any material that contributes to Bremsstrahlung radiation (i.e., decelerating radiation) when bombarded by electrons. The space between the cathode assembly 86 and the anode 70 may be evacuated to minimize electron collisions with other atoms and to enhance high voltage stability. Such evacuation may advantageously allow the magnetic flux to interact with the electron beam 80 (e.g., steering, focusing).

[0049] As mentioned above about Figure 1 As described, the X-ray controller 38 of the system controller 30 provides or controls electrical signals to the X-ray tube 60 (including to the cathode assembly 86), including power, timing signals, and / or focal spot size and position to the X-ray tube 60 during operation. Thus, the X-ray controller 38 has available data regarding electrical parameters used to operate the X-ray tube 60. Such electrical parameters may be retained over time in an operations log or file that records the parameters and electrical conditions of some or all exposure events associated with the emission of electrons by the emitter of the cathode assembly 86. An example of such a log may be a log of exposure information for the system acquired in real time and stored in a log file (or otherwise available) for subsequent retrieval and analysis.

[0050] In the example shown, the X-ray controller 38 (or the overall system controller 30, of which the X-ray controller is one aspect) may include a communication component 90, a processor 92, a memory 94, a storage device 96, an input / output (I / O) port 98, a display 100, and the like. Alternatively, some or all of these components may be shared with or may be equivalent components of the computer 42 of the imaging system 10 that additionally addresses the processing of acquired data and interacts with an operator and other connected devices. The communication component 90 may be a wireless or wired communication link that may facilitate communication with various other processors, etc. The processor 92 may be any type of computer processor or microprocessor capable of executing computer executable code (e.g., computer executable instructions). The memory 94 and the storage device 96 may be any suitable article of manufacture that may be used as a medium for storing processor executable code, data, and the like. These articles of manufacture may represent computer-readable media (e.g., any suitable form of memory or storage device) that may store processor executable code used by the processor 92 to execute the presently disclosed technology.

[0051] The memory 94 or storage device 96 may be used to store data downloaded via the communication component 90. The memory 94 and storage device 96 may be used to store data received via the I / O port 98, data analyzed by the processor 92, etc. The memory 94 and storage device 96 may be used to store data providing details regarding operating parameters of the X-ray tube 60, wherein if the data received via the I / O port 98 and / or the data analyzed by the processor 92 does not meet (e.g., exceeds) the operating parameters, the control system 38 may respond by adjusting the operation of the X-ray tube 60 or a system to which the X-ray tube 60 is associated.

[0052] The memory 94 or storage device 96 may also be used to store application programs 102, firmware, software, etc. The application programs 102, when executed by the processor 92, may also enable the X-ray controller 38 or the system controller 30 to assess the health of the electron emitters of the X-ray tube 60 as described herein. The application programs 102, when executed by the processor 92, may also enable the X-ray controller 38 or the system controller 30 to provide alarms or indications as described herein, such as regarding degradation of the emitters within the cathode assembly 86. It should be noted that the alarms or indications may be provided in any suitable manner (e.g., visual alarms and / or audible alarms) to alert an operator or maintenance personnel of an impending emitter failure.

[0053] With the foregoing in mind and to facilitate subsequent discussion, a number of variables and terms are provided herein related to the operation of the X-ray tube 60. Unless otherwise noted, the X-ray tube electrical parameters or variables noted below may be obtained or retrieved from a log or data storage area where exposure data or parameters (e.g., electronic X-ray tube operating parameters, scan and scanner details, etc.) are acquired in real time and stored for subsequent retrieval and analysis separate from the scan or inspection procedure. Specifically, as used herein, variables associated with preprocessing as described herein include: code (i.e., an inspection number, from which personnel can infer whether an inspection is a routine inspection or a maintenance inspection), time (i.e., the time when the generator log data message is sent), exposure number (Exp_Number) (i.e., the exposure index of the inspection), focus selection (Select_Focus) (i.e., the focus size setting value (e.g., 1 for a small focus and 2 for a large focus)), number of excitations per exposure (Num_of_Spits_per_Exp), inspection prescription (Rx_Option) (i.e., the type of CT inspection), average current and / or voltage during exposure, current and / or voltage at the beginning of exposure or near the beginning of exposure (e.g., at about 5ms), transmitter current and / or voltage during the warm-up phase, transmitter current and / or voltage at the end of exposure, current setting value (Select_mA), scanning mode (e.g., single or multiple exposures, scanning sequence, etc.), etc.

[0054] Taking into account the above parameters and variables, pre-processing of the X-ray tube electrical data as used herein may include limiting the data used to assess emitter lifetime to data points where the tube current is greater than a specified minimum value such as 0, the tube voltage is within an acceptable range, the CT exam (Rx_Option) and / or scan mode is of an appropriate type (e.g., the emitter drive current remains constant for an appropriate length of time, such as between 0.2 seconds and 5 seconds), the number of excitations per exposure is 0, the exam corresponds to the model employed, and the actual X-ray tube current is close to the indicated X-ray tube current, such as within 1%. Data points that do not meet these conditions may be discarded from the calculation of the indicator (as described above) as part of the data pre-processing, thereby reducing the risk of false positive indications due to corrupted or non-representative measurements. However, it may be noted that in certain implementations, some or all of this data may be retained in the total exposure count used to determine the slope or other derived parameters, as will be discussed in more detail below.

[0055] Pre-processed (i.e., cleaned) X-ray tube electrical data points may be provided as input to train or use an algorithm for predicting failure or estimating the remaining useful life of the X-ray tube emitter 112 as described herein. In one implementation, the algorithm of the present invention is based on the static tube model 110 ( Figure 3 ), the model is used to calculate a fault indicator or a remaining life estimate based on the residual value between the measured emitter (e.g., filament) drive current and the estimated emitter drive current. Specifically, given Figure 3 The static tube model shown in Figure 2 shows the X-ray tube current I 管 It can be characterized as the emitter (e.g., filament) drive current I 灯丝 and the X-ray tube voltage V 管 A function of . An example of the derivation and parameterization of such a model can be found in U.S. Patent No. 7,023,960, entitled “Method of adjusting the emission rate of radiation from a source of radiation”, issued to General Electric Company on April 4, 2006, which is incorporated herein in its entirety for all purposes. Although examples are generally described below in which an estimated emitter drive current is calculated based on an X-ray tube current, it is understood that it is also contemplated that the estimated X-ray tube current may be calculated based on the emitter drive current instead, and the estimated X-ray tube current as described below may be used in the same manner as the estimated emitter drive current. In addition, as described below, the extraction voltage applied to the emitter may also be utilized in the same manner as the estimated emitter drive current. In fact, the estimated value may be any variable related to the emitter resistance. For simplicity only, a specific implementation of the emitter drive current is discussed in detail below, but those skilled in the art will readily understand the applicability of this discussion to corresponding uses based on the estimated X-ray tube current, the extraction voltage applied to the emitter, or any variable related to the emitter resistance.

[0056] Various aspects of the model, such as the model coefficients, may be estimated by fitting the model with training data, which is typically collected from the beginning of the life of an X-ray tube. The predicted emitter (e.g., filament) drive current may then be calculated from the inverse static tube model. Typically, the X-ray tube current at a fixed X-ray tube voltage is a function of the emitter temperature. As the emitter becomes thinner due to use (i.e., evaporates), the resistance of the emitter increases, and thus the emitter drive current required to reach the target temperature becomes lower. In the case of a circular cross-section emitter (e.g., a filament), this approximately proportional relationship between the emitter current, temperature T, and filament diameter d may be characterized as:

[0057] (1)I灯丝 2 ∝T 3 d 3

[0058] The scaling factor depends on the emitter emissivity, which itself depends on the cathode geometry. In addition, the initial emitter thickness (e.g., filament diameter) may also fluctuate.

[0059] Taking this into account, if we assume constant temperature, then the quantity I 灯丝 2 / 3 must be proportional to the diameter or other measure of thickness. A first order approximation is the assumption that for fixed X-ray tube current and X-ray tube voltage settings, the temperature remains constant over the life of the emitter. However, in practice, it has been observed that this is not the case, as the temperature of the emitter does change as the thickness (e.g., diameter) of the emitter decreases, so I 灯丝 2 / 3 The proportionality relationship with the thickness of the emitter is not accurate. Note that hot spots have been observed to form on the emitters, which leads to a non-uniform temperature distribution on the emitter, which may be a possible explanation for this. However, in the context of certain specific implementations of the method of the present invention, it can be assumed that the general proportionality relationship holds with the same proportionality constant throughout the lifetime of the emitter. In order to account for the variation, the proportionality constant is not assumed to remain constant across the emitter. Instead, for each emitter, the proportionality constant is learned from the initial part of the lifetime of the emitter, and is assumed to remain constant throughout the lifetime of the emitter.

[0060] In order to normalize the variation of the emitter current at different kV-mA settings, some embodiments of the method of the present invention normalize to the current observed at the beginning of the life of the X-ray tube at each kV-mA setting. Using this normalization, the quantity I 灯丝 2 / 3 The fractional decrease in can be used as an indication to quantify the thinning of the emitter over time. In this way, the lifetime of the corresponding emitter at any lifetime t can be expressed as:

[0061] (2)

[0062] where the expression on the left hand side is evaluated at any fixed kV-mA setting. Based on this relationship, a fractional decrease in the emitter current to the power of 2 / 3 is equal to a fractional decrease in the diameter (or more generally thickness) of the emitter. This quantitative assessment of the decrease in emitter thickness can be used as an indicator of wear on the corresponding X-ray tube. In one specific implementation, this indicator can be expressed as I rel (t), where the superscript indicates that this metric captures the relative change in filament current. This can be defined as:

[0063] (3)

[0064] It may be noted that, in practice, the quantity I 灯丝 (0) is not measured directly, but rather estimated using an emitter model evaluated at the current kV-mA setting. As described herein, since the emitter model is learned using beginning of life data from an X-ray tube emitter, the estimate does correspond to the emitter current that will result from a new filament at the operating kV-mA setting.

[0065] With the foregoing in mind, in practice, the long-term feedback control system implemented in the X-ray generator automatically adjusts the emitter drive current from one exposure to the next so that the X-ray tube current remains close to the desired set point. Therefore, the measured emitter drive current will be lower than the current predicted using the regression coefficients learned in the initial stages of the emitter. Assuming that the constant tube current and tube voltage set points correspond to constant temperature, it can be assumed that I 灯丝 2 / 3 The relative decrease in thickness should be proportional to the relative decrease in thickness. Therefore, if the index I of Formula 3 is plotted against a measure of the total usage of the X-ray tube over the life of the X-ray tube (e.g., the cumulative number of exposures that have been applied using the X-ray tube), rel (t), a downward trending curve should be observed. As the transmitter approaches failure, the decrease in transmitter current accelerates. Therefore, significant changes in this indicator can be observed in terms of both magnitude and slope. An example of this indicator is shown in Figure 4 In this graph, the fraction of lifetime (x-axis) at any point is measured as the ratio of the total number of exposures that have occurred on the transmitter to the total number of exposures at which failure occurred at the transmitter.

[0066] like Figure 4 As seen in I rel The downward trend in the value of (t) accelerates towards the end of the life of the transmitter. Therefore, the slope of the indicator provides useful information that can be used to predict transmitter failure. According to the method of the present invention, a combined indicator can be used that combines the indicator value with the slope to produce a single indicator:

[0067] (4)I 组合 (t) = I rel (t)+50000|I rel (t)|S rel (t)

[0068] Where S rel (t) is I rel (t) Slope or rate of change relative to the exposure count. relThe scaling factor |I (t)| is added to the slope term to emphasize that the slope plays a more significant role as the indicator approaches the end of the life of the emitter, at which point the factor |I rel (t)| will be larger. In addition, in this example, a factor of 50,000 is adopted based on trial and error to ensure that the final combined indicator fully utilizes the slope information without inducing unnecessary fluctuations. Figure 5 The evolution of this indicator as a function of the lifetime fraction is shown.

[0069] Compare Figure 4 and Figure 5 ,It can be observed that the combined index decreases at a faster rate near the end of the life of the transmitter and reaches a smaller value at the end ,therefore, incorporating the slope information into the index helps to improve the detectability of ,imminent failures.

[0070] An example of a process and use is provided with respect to the generation and use of an emitter fault detection algorithm as used herein. The example is provided as a two-step process, wherein the first step involves modeling the relationship between the emitter drive current and the X-ray tube current and voltage, and the second step involves applying the constructed model to monitor incoming data and set alarms for potential faults.

[0071] Regarding the first step, an inverse static X-ray tube model is constructed (i 灯丝 =f(v,i)). The available data is divided into two subsets, one for small emitters and one for large emitters. The following steps are applied separately to each emitter size subset (i.e., separately to large emitter data and small emitter data) because the two emitters can be used independently and therefore they can age at different rates.

[0072] In this example and given the constraints above, training data points are then collected for linear regression model fitting. To ensure that the training set has enough data points, one approach is to use a certain threshold time window of data (e.g., the next 1 day, 7 days, or 31 days of data) as the training set after tube installation. If the number of training points is below a threshold (e.g., 1,000, 2,000, etc.), one more day of data is added to the training set until the number of training points is above the threshold. For a new X-ray tube, one week of data may be sufficient to build a model.

[0073] The mA-kV setpoints used in the training set are counted. In one embodiment, during the monitoring phase described below, the algorithm will not consider a new point for further analysis if its mA-kV setpoint does not appear in the training set.

[0074] Calculate the regression model i=f(v,i 灯丝) and use it to estimate the coefficients. Using these coefficients, an inverse static tube model is constructed. As described below, this model can be used to estimate the emitter drive current by using the estimated coefficients.

[0075] Go to Figure 6 , this first step is illustrated relative to the upper branch of the flowchart shown. As shown in this example, X-ray tube generator data points are collected over time 150. In this model training example, some threshold number of points that meet one or more pre-processing criteria (such as those discussed above) are collected for training. In the data points shown, pre-processing or "cleaning" occurs at step 152, the output of which is generator data points that meet the established pre-processing criteria and are suitable for model building (step 154). As described above, while data points can be discarded for the purposes of metric calculation, they can still be retained as part of the exposure counts in the regression used to determine the slope of the curve. Thus, as Figure 6 As shown, model building 154 is performed to generate a model 156 of the corresponding X-ray tube 60. The model 156 can then be used to monitor the corresponding X-ray tube 60 for impending emitter failure or remaining useful life estimation, as described with respect to the second step and as shown in FIG. Figure 6 is shown in the lower branch of the flowchart.

[0076] Regarding the second step, additional generator data points 150 are collected over time as the X-ray tube 60 is used. An estimated emitter drive current is calculated based on the new data points 150 and a model 156 (e.g., an inverse static X-ray tube model). In this example, the i derived from the inverse static tube model is 灯丝 is the emitter drive current.

[0077] Trend Indicator I rel (t) can be calculated as:

[0078] (5) Among them I 灯丝 (t) represents the filament current measured in the corresponding exposure indexed by t.

[0079] In one embodiment, outliers in the measurement results are identified and removed. For example, the measurement result I rel (t) can be used to calculate the slope that meets one or more criteria, such as but not limited to: rel (t) <Mean(W)+1.0×Std(W),I rel (t)>Mean(W)-1.0×Std(W), I rel (t) <Mean(W)+2.0×Std(W),I rel(t)>Mean(W)-2.0×Std(W), (mA_at_5ms(t)) / (Average_mA(t))<1.1, (mA_at_5ms(t)) / (Average_mA(t))>0.9, (mA_at_e xposure_end) / (Average_mA(t))>0.95, (mA_at_exposure_end) / (Average_mA(t))<1.05, (Average_mA(t)) / (Commanded mA)>0.95, (Average_mA(t)) / (CommandedmA)<1.05, (mA_at_exposure_end) / (Commanded mA)>0.98, (mA_at_exposure_end) / (Commanded mA)<1.02, etc., where W is the number of points containing the past 1000 points. rel (t-1000),…,I rel The purpose of some of these standards may be to reduce the impact of mA instabilities identified as abnormal behavior of mA_at_5ms / Average_mA.

[0080] For missing data, the time gap between two exposures can be checked to ensure it is above a threshold (e.g., 7 days, 31 days, etc.) If a gap is detected, the moving window is reset at the first valid data point after the time gap.

[0081] In one specific implementation, the effective observations in the moving window are used to calculate the slope S of the indicator. rel (t). By executing I rel The slope is calculated by linear regression of (t) on t, which is the raw exposure count before outliers are removed. In other words, t measures the count of the current measurement in the raw data, not the count in the data remaining after outliers are removed. In one embodiment, data points generated during maintenance checks are also included in the raw exposure counts used for the slope calculation. The slope S based on the indicator rel (t), calculate the combined index I according to formula 4 组合 (t).

[0082] Back to Figure 6, illustrating this second step with respect to the lower branch of the flowchart shown. As shown in this example, X-ray tube generator data points 150 are collected over time and cleaned (i.e., pre-processed) as described above. The data points are then used together with the model 156 to calculate the trend indicator and the combined indicator (step 160), as described above. If the alarm triggering criteria are met (decision box 162), an alarm is generated. If not, data collection and processing continues over time.

[0083] Although the combined index of Equation 4 is a good index for predicting transmitter failure, the value of the combined index is essentially a fractional change in transmitter lifetime, which has no easy to understand suitable interpretation. To address this problem, a linearized version of the index that directly measures the transmitter fractional remaining lifetime can be used.

[0084] For example and go to Figure 7 , shows a combined metric for various emitters. As shown, each of the emitters represented tends to follow the same evolution as the fractional lifetime of the emitter changes. Based on this observation, a model can be fitted to predict the fractional remaining lifetime from the metric value. This can take the form of a linearized version of the metric. In one such example, the linearized version of the metric is obtained by transforming the relative metric. The transformation can be given by:

[0085] (6)I 变换 (I) = max{1-a(1-e bI ),0}

[0086] Where a and b are parameters learned during the training phase. Figure 8 As shown in , the transformation converts the index value into a score between 0 and 1. The result of the transformation is as follows Figure 8 In practice, the linear transformation can be applied directly to the combined metric or to a smoothed version of the metric, depending on whether the goal is failure prediction or remaining useful life estimation, both of which are discussed in more detail below.

[0087] Regarding the estimation of the remaining useful life of the X-ray tube, I in Equation 4 组合 The combined metric of (t) may fluctuate significantly due to inaccuracies in the model, the presence of unfamiliar usage patterns, and outliers in the data. Therefore, a remaining useful life estimate based directly on this metric will be correspondingly noisy and subject to fluctuations. These problems can be addressed by smoothing the metric. Therefore, for remaining useful life estimation, the transformation of Equation 6 can be applied to I given in Equation 4 组合 In other words, the model of the form of Formula 6 can be used to compare the combined indicator (e.g. Figure 7 The curves associated with those shown in are modeled.

[0088] In this approach, the algorithm used to learn the model is applied to the combined metric of Equation 4 rather than the relative metric of Equation 3. Another difference is the calculation of the fractional remaining lifetime used in the training phase. Instead, since the usage rate remains constant over the entire life of the emitter at almost all observed sites, the fractional number of exposures remaining can be used as the fractional remaining lifetime. Therefore, the fractional cumulative wear is roughly equal to the fractional number of exposures. In view of this observation, there is little or no gain in choosing a computationally complex quantity (such as cumulative wear) rather than simply using exposure counts. Another variation in the implementation may employ a weighted fit (in which a larger weight is assigned to the error at the end of the emitter life) while performing the fit. This ensures that the fit is accurate near the end of the filament life, which is an interesting scenario for failure prediction.

[0089] With the foregoing in mind, in one example of a specific implementation, steps such as the following steps may be performed to learn a model that may be used to estimate the remaining useful life of an X-ray tube.

[0090] With respect to the learning model, for each failed transmitter, the combined trend index of Formula 4 is calculated using the available data. When the combined trend index is calculated in this way, a median filter is applied over an exposure window (e.g., over 300 exposures) to generate a smoothed index. The smoothed index at any time is the median of the index measured during the past window of valid exposures (e.g., 300 times), which has passed the outlier removal test and the missing data test described herein. In this example, S represents the smoothed index obtained by calculating the median of the past exposure window (e.g., 300 exposures).

[0091] For each data point, the fractional remaining life, denoted as R, can be calculated as follows: (7)

[0093]

[0094] The fractional remaining lifetime after any exposure can be denoted as R.

[0095] The weighted nonlinear regression of R on S is performed using the functional form of Formula 6 to obtain a nonlinear fit of the following form:

[0096] (8)R≈max{1-a(1-e bS ),0}

[0097] And the fitted values ​​of parameters a and b are obtained. The weight assigned to the error at each data point is (1-R) 2The ratio is proportional to emphasize the accuracy of the fit at data points close to the end of the filament life, where the value of R is close to 0 or equivalently. In one embodiment, only data points after the first 5000 exposures are used when performing the regression.

[0098] Once the model is learned, the values ​​of the parameters a and b are available. One aspect of this approach is that the model can be continuously updated as data from more failed filaments become available. Processed data from these filaments can also be added to the training set to improve the robustness and accuracy of the parameter estimates.

[0099] Once learning is complete, the model can be used to predict the remaining useful life of other transmitters. In one example, the steps involved in predicting the remaining useful life of a transmitter may include the following steps. For the current transmitter of interest, the available data for the corresponding transmitter is used to calculate a combined trend indicator according to Formula 4. In some exposure windows (e.g., 300 exposures), the normalized trend indicator so calculated is smoothed, such as using a median filter. Therefore, the smoothed indicator at any time is the median of the indicators measured during the past exposure windows. The smoothed indicator obtained by calculating the median of the past exposure windows can be denoted as S.

[0100] For each measurement of the smoothed indicator, the time difference between the earliest sample and the last sample is calculated. This time difference can be used to calculate the median. In some specific implementations, a threshold can be applied so that a time difference exceeding the threshold (e.g., 25 calendar days) is considered too large and the median is not calculated in this window due to the lack of current or "fresh" data.

[0101] The calculated value of S and the relationship shown in Equation 1 are used with the learned values ​​of parameters a and b to calculate an estimate of the remaining fractional life, R, as follows:

[0102] (9)R 估计 =max{1-a(1-e bS ),0}

[0103] In one implementation, the estimate of R may be reported differently depending on whether the model is being trained or has fully completed training. For example, in such an implementation, a value of "1" (i.e., full life) may be reported when the model is being trained, while a calculated value is returned instead when the model has completed training.

[0104] As can be appreciated with respect to the preceding steps, the output of the above algorithm is a fraction of the remaining lifetime of the emitter. In performing such an estimate, it is implicitly assumed that the usage rate of the emitter will remain substantially fixed or constant. The output of the above algorithm can be transformed into a remaining lifetime expressed in absolute calendar time, and with additional information, such as the current lifetime of the filament expressed in calendar time.

[0105] Although the smoothed combined index S described above can be used as a remaining life estimate, it may not be suitable for predicting impending failures because smoothing the index tends to average out sudden transitions in the index. In practice, sudden drops in the combined index value are common at the end of the life of a transmitter, which is one of the plausible explanations for incorporating a slope into the combined index. Therefore, for failure prediction, it may be more useful to rely on a processed version of the original combined index of Equation 4 rather than the smoothed index.

[0106] With this in mind, the following is an overview of a prediction algorithm for predicting emitter failures. The prediction algorithm described includes two aspects: modeling the relationship between the emitter drive current and the X-ray tube current and voltage, and applying the constructed model to monitor incoming data and set alarms for potential failures. Note that in one specific implementation, the data used by the prediction algorithm has undergone the pre-processing steps described above.

[0107] Regarding the steps to calculate the trend indicator I of formula 5 rel (t) and the combined index I of formula 4 组合 (t). The unsmoothed combined indicator is transformed according to the following formula to obtain the unsmoothed remaining fractional useful life estimate:

[0108] (10)

[0109] It may be noted that in this example, the transformation is performed on an unsmoothed combined indicator (rather than a smoothed indicator used for the above-described remaining useful life estimation) so that abrupt changes can have the effects described above. Alternatively, rather than applying no smoothing at all, less aggressive smoothing may be applied (e.g., using a smaller number of exposures for the window used for the median calculation).

[0110] The remaining days of the service life of the corresponding transmitter can be estimated as:

[0111] (11) Where N(t) is the total life of the filament obtained by calculating the difference between the current date stamp and the date the X-ray tube was installed.

[0112] For notification of an impending or upcoming transmitter failure, one or more alarm criteria may be established. For example, in one embodiment, an alarm or notification may be generated when the following conditions are met:

[0113] (12)max{D(t),D(t-1),D(t-2),D(t-3),D(t-4)} <Thr

[0114] Where Thr is a configurable threshold that represents the lead time before a transmitter failure needs to be notified. In this example implementation, the trigger criteria requires 5 consecutive points that meet the threshold, which helps reduce the impact of outliers.

[0115] It may be noted that in an implementation that performs both remaining useful life estimation and failure prediction, the above sequence of steps may be modified to facilitate the execution of both algorithms. For example, in the above implementation of remaining useful life estimation, smoothing of the combined index is described as being performed prior to applying the nonlinear transformation of Equation 6.

[0116] However, the smoothing operation is performed using a median filter, so if the smoothing is instead performed after the nonlinear transformation, the final estimate will not be significantly different. Therefore, in an alternative implementation, a nonlinear transformation can be performed first on the unsmoothed combined indicator, so that the result of the nonlinear transformation operation can be used for both the remaining useful life estimation and the alarm generation (i.e., failure prediction). Using this sequence of operations will no longer require repeated nonlinear transformations on the smoothed signal when calculating the remaining useful life estimate.

[0117] Fig. 9 An illustration of a suggested flow for performing all steps required for model training, remaining useful life estimation, and alarm generation is provided. In this example, data characterizing electrical operating parameters of each exposure event is acquired, such as at step 200 of reading log data from a log file. At step 202, the acquired log data may be cleaned as described herein to remove data points that are deemed unrepresentative or inappropriate for modeling or evaluation purposes using the model.

[0118] In the example shown, a determination is made as to whether the X-ray tube 60 in question is a new X-ray tube (decision block 206). As noted herein, this may be based on whether data for the tube exists for some threshold number of days (e.g., seven days) or based on other criteria, such as whether a new tube indicator or marker exists in a corresponding database, whether a model associated with the X-ray tube exists, etc. If it is determined that a new X-ray tube 60 exists, then the training mode for the illustrated tube is set to on (step 208). If the X-ray tube 60 is not new, then a trained model exists, and the new data point will be processed with the existing model for remaining life estimation and / or transmitter failure prediction.

[0119] In the illustrated implementation, once a new tube detection is determined, the acquired data points may be processed to remove data points associated with unreliable exposure events (step 212), unstable mA, etc. Based on whether the model training mode is set to on or off (decision box 216), one of two paths may be followed. With the model training mode set to on, exposure parameters based on the data points are stored or memorized (step 220) until a threshold amount of data is acquired (decision box 222). Once the threshold amount of data is acquired, the training mode is set to off (step 226), and the model is calculated as described herein (step 228).

[0120] Returning to step 216, when the model training mode is set to off, the acquired data points are processed to remove outliers, such as based on an absolute or statistical threshold (step 240). For each exposure event, an emitter (e.g., filament) metric is calculated (step 242). The metric can then be used in one or both of calculating the remaining lifetime percentage of the emitter (step 250) or calculating the remaining days of the emitter (step 252). In the case of calculating the remaining lifetime percentage of the emitter as described above, a smoothing step 254 can be performed before performing the calculation. In the case of calculating the remaining days of the emitter (step 252), the calculated days can be used to calculate (step 260) whether one or more trigger conditions are met to generate a notification or alarm of an impending emitter failure within a configured time period.

[0121] In another aspect, it will be appreciated that in addition to the remaining lifetime and / or transmitter failure prediction information that can be derived according to the method of the present invention, other information can also be derived. Fig.10 and Fig.11 , it can be seen that information about unstable mA conditions can also be ascertained. For example, see Fig.10 , the ratio between the 5ms mA and the average mA of the X-ray tube in question becomes very unstable at later time periods. Correspondingly, go to Fig.11 , it can be observed that once an unstable mA condition is present, the curve of the trend indicator is no longer smooth. Therefore, in addition to the other conditions described herein, the trend indicator method of the present invention can also be used to identify other X-ray tube problems, such as unstable mA. Unstable mA can be caused by problems unrelated to the emitter itself, such as fluctuating pressure inside the X-ray tube, which in turn can be caused by outgassing or vacuum leaks.

[0122] In some embodiments, the training dataset may not have sufficient learning points. Insufficient learning points may lead to incorrect estimation of model parameters, which may result in false alarms for predicting X-ray tube filament failures. In such cases, it may be necessary to reduce or simplify the complexity of the model (e.g., a regression model) (e.g., by reducing the number of coefficients to be estimated or by using a model with a different form (e.g., different coefficients)). Since the complexity of the model depends on the availability of different kV-mA settings in the training data, the following strategies can be utilized to count the number of operating points (e.g., kV-mA settings). These strategies include: given any kV, if there is only one available mA, the corresponding kV-mA setting will not be used for model construction and metric calculation; given any kV, for two uncertain mAs (mA1 < mA2), if the conditions (mA2 - mA1) ≥ 20 and (mA2 - mA1) / mA1 ≥ 0.2 are not met, the two mAs are grouped as one mA; count the number of unique tube voltage settings; and count the number of unique tube current settings within each tube voltage setting.

[0123] Based on the available kV-mA settings in the training data, the regression model to be utilized can be selected. A larger number of available kV-mA settings leads to the selection of a more complex regression model with a larger number of coefficients to be estimated, while a smaller number of available kV-mA settings leads to the selection of a less complex regression model with fewer coefficients to be estimated. In some embodiments, regression models in different forms (e.g., with different coefficients) can be utilized. In addition to the number of available kV-mA settings, other criteria can be utilized to select the regression model. For example, an expansion criterion (SC) may have to be satisfied. For any kV and its corresponding mA, the given mA expansion (S) is defined as S = log(max(mAs)) - log(min(mAs)). To satisfy SC, the following criteria should be met. Assuming S max is the maximum mA expansion for all available kVs, there should be at least 2 kVs (if available ≥ 3 kVs) or 1 kV (if available 2 kVs), where their mA expansion is greater than 0.5x S max ; and Smax ≥ 1.

[0124] For example, assume that the training data indicates three unique tube voltage settings, namely 80 kV, 100 kV, and 120 kV, and for each tube voltage setting, there are three tube current settings (80 kV-20 mA, 50 mA, 255 mA; 100 kV-80 mA, 200 mA, 220 mA; 120 kV-120 mA, 180 mA, 200 mA). After grouping, the tube current settings at 200 mA and 220 mA are combined for the tube voltage setting at 100 kV, and the tube current settings at 180 mA and 200 mA are also combined for the tube voltage setting at 120 kV. The tube voltage setting at 80 kV still has three tube current settings. The mA spreads of the tube voltage settings at 80 kV, 100 kV, and 120 kV are 1.106, 0.398, and 0.176, respectively. Therefore, the first term of SC is not satisfied and a regression model with a reduced number of coefficients will be selected.

[0125] The foregoing examples assume that only a few extended kV settings are used (e.g., 80 kV, 100 kV, 120 kV, 140 kV). There may also be systems that use a continuous kV setting, or systems that use kV settings with small increments that do not inherently satisfy SC (e.g., 80 kV, 80.1 kV, 80.2 kV, ...). In this case, the SC method described for mA settings can also be applied to kV settings. Alternatively, the kV settings can be classified to inherently satisfy the extended criteria, such as by treating all kV settings close to a "nominal" kV setting as being at that nominal kV setting. For example, all kV settings for which 75 kV ≤ x < 85 kV are treated as 80 kV when determining the regression model to use.

[0126] In summary, Fig.12 , a method 300 for constructing an X-ray tube model for assessing the health of an X-ray tube is shown in FIG. The method 300 may be performed by a processor-based system as described above. The method 300 includes acquiring training data points for a corresponding X-ray tube after the X-ray tube is installed in an imaging system (block 302). The method 300 also includes determining a number of operating points (e.g., kV-mA settings) used in the imaging system for the corresponding X-ray tube (block 304). The method 300 further includes selecting a regression model from a plurality of regression models based on the number of operating points (block 306), wherein each regression model uses the training data points to derive respective values ​​of a plurality of coefficients. The method 300 even further includes constructing an X-ray tube model using a plurality of coefficients estimated by the selected regression model (block 308). Utilizing a less complex regression model (when fewer learning points are available) or a model in a different form (e.g., with different coefficients) may reduce or eliminate the triggering of false alarms in predicting X-ray tube filament failures.

[0127] Fig.13 and Fig.14 Example curves showing the linearized combined index for the filament current of an X-ray tube using a more complex regression model and a simpler regression model are provided, respectively. Lines 301 and 305 represent the linearized combined index and the alarm index, respectively. Fig.13 As shown, the linearization combination index 301 suddenly jumps from close to 0 to a very large value (e.g., 10e10 to 10e13), and then returns to a normal value (about 1.0), which will trigger the false alarm 305. Fig.14 As shown, employing a simpler regression model results in no false alarms being triggered and the linearized combined indicator 301 exhibits the expected decrease over time.

[0128] In some cases, replacement of certain parts or other maintenance actions (e.g., filament recalibration, heater plate replacement, etc.) may cause a sudden change (i.e., a jump or decrease of a certain and significant percentage within a short period of time (e.g., one or more days)) in the calculated index (e.g., trend index), which may lead to an erroneous estimate of tube life. In other cases, in some tubes, the combined index initially has an upward trend that may cause the combined index to be greater than one for a considerable period of time, which may lead to an inaccurate life estimate of the tube life. The following techniques monitor for these situations (e.g., sudden changes in the tube life curve) and correct the combined index.

[0129] Fig.15 A method 310 for detecting sudden changes in a tube life curve is shown. The method 310 can be performed by a processor-based system as described above. The method 310 includes monitoring for sudden changes in a trend indicator (box 312). Certain rules (e.g., utilized by the processor-based system discussed above) can be implemented to detect sudden changes in a tube life curve (e.g., a trend indicator). One such rule finds out whether, on a given day, the mean of the trend indicator of Formula 5 shows a significant change compared to previous days. The trend indicator is utilized (rather than a linearized combination indicator) to simplify calculations and reduce calculation time. The method 310 includes determining whether a certain number of exposures on a given day t is above a predetermined threshold or value (box 314). The rules for change detection are as follows: On a given day t, if a given number of exposures (e.g., 10 times or another predetermined number) is above a threshold, then the mean of the trend indicator of Formula 5 is calculated (Block 316), otherwise, take no action (ie, continue monitoring for sudden changes (Block 312)); calculate The minimum difference from the average trend indicator of the previous 7 days (Block 318); Determine Dif min Is it greater than a threshold (block 320), where if Dif min>μ, where μ is a threshold at a predetermined value (e.g., 0.006, which represents about 6% to 7% tube life), then the flag for day t is set to flag t =1 (box 322); and determine whether the number of days in the set window meets a certain value (box 324). For example, for a continuous 7-day window (or other window with a different number of days), if 4 of them (or another number of days) have a flag set to 1, then trigger an alarm (e.g., provide notification of an alarm) (box 326) to indicate that a change was detected. Note that the day when the change occurs will be the day with the first flag set to 1 in the 7-day window, not the day when the alarm is triggered. If the number of days does not meet the value, continue monitoring for sudden changes (box 312).

[0130] Once a change is detected, a correction needs to be made to the filament current (Filament_Current_in_Preheat) to move the trend indicator curve back to the correct position. Fig.16 A flow chart illustrating a method 328 for correcting a sudden change in a tube life curve and retraining a linear regression model is shown. The method 328 may be performed by a processor-based system as described above. For example, assuming a sudden jump or drop is detected on day t, the objective function of the correction procedure is to minimize the difference between the mean of the trend indicator before and after the change (block 330),

[0131] (13)

[0132] By using Formula 5, the mean of the trend indicator on day t+1 can be described as,

[0133] (14)

[0134] where α is the correction factor applied to the filament current after the change, and N t+1 is the number of exposures on day t+1.

[0135] By setting Formula 13 to 0, the following calculation is performed,

[0136] (15)

[0137] This leads to,

[0138] (16)

[0139] Therefore, the correction factor α may be calculated (block 332) as follows,

[0140] (17)

[0141] When calculating the correction factor, the method 328 includes applying the correction factor to the filament current to move the trend indicator to the correct position (block 334 ).

[0142] Fig.17 and Fig.18 Example graphs showing the linearized combination index of the filament current and the corresponding slope over time for an X-ray tube without and with correction are provided, respectively. Lines 327, 329, and 331 represent the linearized combination index, the slope, and the alarm index, respectively. Fig.17 As shown, the linearization combination index 303 suddenly jumps from about 1 to about 1.5 in the early stage. Fig.18 As shown, the sudden jump is detected and corrected by utilizing the above-mentioned techniques.

[0143] To address the issue that the tube life may have exceeded 100% of the remaining life, a continuous learning mechanism can be applied to dynamically update the trained linear regression model that describes the relationship between the tube voltage, tube current, and filament current. Similar to the change detection rules described above, whether retraining is needed is determined based on the calculated trend indicator. From Equation 5, it can be seen that the ideal situation is that the trend indicator will start from 0 and gradually decrease as the filament becomes thinner. However, in some cases, the value of the trend indicator can be much greater than 0. Therefore, the trained linear model no longer captures the evaporation thinning of the filament, and the model needs to be recalibrated.

[0144] return Fig.16 To retrain the model, we keep the moving window W with a size set to 1,000. Method 328 includes calculating the trend indicator in the moving window (box 336). Method 328 further includes determining whether the median value of the trend indicator is above a predetermined threshold (box 338). If the median value of the trend indicator in the moving window is above the threshold (e.g., set to 0.001 or another value), the linear regression model will be retrained by using the data points from the moving window (box 340), where outliers are removed based on the following rules. Given the trend indicator I in the moving window rel (t), then the outliers are identified as (mA_at_5ms(t)) / (Average_mA(t))>1.2, or (mA_at_5ms(t)) / (Average_mA(t))<0.8, or I rel (t)≥Mean(W)+3.0x Std(W), or I rel (t)≤Mean(W)-3.0x Std(W). If the median value of the trend indicator in the moving window is not above the threshold, then the method 328 includes determining not to further train the linear regression model (block 342).

[0145] It should be noted that the decision about retraining occurs after the change detection step. In other words, if a sudden change is detected, the filament current will be corrected first, and then the trend indicator will be calculated (which is used to determine whether model retraining is needed).

[0146] Fig.19 and Fig. 20 Example curves showing the linearized combination index of the filament current and the corresponding slope over time of the X-ray tube without and with retraining are provided, respectively. Lines 337 and 339 represent the linearized combination index and the slope, respectively. Fig.19 As shown, at the beginning of the tube life, the linearized combined index 337 jumps from about 1 to about 1.8 in the early stage. Fig. 20 As shown, after applying retraining, the linearized combination indicator 337 reaches a peak of about 1.0 in the early stage instead of 1.8.

[0147] The X-ray tube is the most frequently replaced part in an imaging system. When the initial X-ray tube is worn out, it is desirable to replace the approved / authorized X-ray tube (i.e., from the manufacturer / supplier) with another approved / authorized X-ray tube. However, due to third-party availability, the initial X-ray tube can be replaced with a non-original X-ray tube (i.e., not approved or authorized by the manufacturer / supplier). The non-original X-ray tube can be easily configured to be the original (i.e., approved / authorized X-ray tube) without notifying the manufacturer / supplier of the initial X-ray tube and the system. Although the actual X-ray tube is a third-party replacement, this action results in all doses and other critical parameters related to exposure being recorded / assigned to the initial approved / authorized X-ray tube. The manufacturer / supplier of the initial X-ray tube has an interest in knowing the status of any X-ray tube change actions and keeping a log of authorized X-ray tube replacements. These actions can be reviewed periodically based on authorized maintenance requests. Any differences can help the manufacturer / supplier of the original X-ray tubes identify systems operating on third-party X-ray tubes and take appropriate actions to ensure those third-party tubes are not configured as the original manufacturer / supplier's X-ray tubes.

[0148] In certain embodiments, the above-described techniques for predicting X-ray tube failure can be used to monitor filament current values ​​to automatically detect X-ray tube replacements. When a new X-ray tube is installed in the field as a replacement part or installed into a system for the first time, its power unit (i.e., generator) performs a filament calibration using the filament current. This value obtained during calibration is used to track the filament aging process. As described above, as the filament ages, the value of the current required to heat a specified filament to a specified temperature decreases because the filament itself decreases in size due to evaporation.

[0149] The trending algorithm described above enables tracking of filament current values ​​(which are typically decreasing trends over time). It should be noted that the trend may increase over time, depending on the variable being tracked related to the emitter resistance. In certain embodiments, the tracking of these filament current values ​​is performed discretely and operates as a core process that ensures complete invisibility and confidentiality. As described below, when an existing X-ray tube is exhausted and replaced with a new X-ray tube (or a component of the filament drive circuit that is connected to the X-ray tube (e.g., located outside the X-ray tube)), the trending algorithm (described above) and historical records will detect an anomaly (i.e., replacement of the X-ray tube or filament drive circuit component). A (suspected) X-ray tube change event (or filament drive circuit component change event) may trigger an alarm displayed on the system. For systems monitored by the manufacturer / supplier of the system, a (suspected) X-ray tube change event may send an alarm to the manufacturer / supplier. The manufacturer / supplier can verify the event by viewing the system identification number and the relevant maintenance request associated with the system (e.g., to see if an X-ray tube replacement is scheduled). In certain embodiments, for non-connected systems (i.e., not connected to the manufacturer / supplier via the Internet), detection of a (suspected) X-ray tube event may result in a notification (e.g., a pop-up alert) being displayed or provided to the customer to notify the customer to contact the manufacturer / supplier of the X-ray tube or the component of the filament drive circuit. If the X-ray tube replacement or filament drive circuit component replacement is performed by the original manufacturer / supplier of the initial X-ray tube or component (via a field engineer approved by the manufacturer / supplier), the notification or pop-up alert may be turned off (e.g., using the manufacturer / supplier's proprietary replacement X-ray tube). The pop-up alert may appear repeatedly until the manufacturer / supplier is contacted to investigate the matter. After a site visit by a field engineer from the manufacturer / supplier, the necessary actions may be taken (i.e., determining whether the initial X-ray tube was replaced with a third-party X-ray tube and turning off the alarm).

[0150] Fig.18A flow chart illustrating a method 344 for monitoring replacement of an X-ray tube or replacement of a filament drive circuit coupled to an X-ray tube (e.g., located external to the X-ray tube) is shown. The method 344 may be performed by a processor-based system as described above. The method 344 includes obtaining a baseline value of an emitter drive current of an X-ray tube (e.g., an initially approved or authorized X-ray tube) during manufacturing (box 346). The value may be stored in a memory and accessed by the processor-based system. The baseline value is obtained during a staged progression of the system in manufacturing, wherein the system records the filament aging current during the staged progression of manufacturing. Once the system is installed at a customer site and commissioned, the method 344 includes monitoring a variable related to emitter resistance (e.g., emitter drive current (e.g., filament aging current)) (box 348). Monitoring of a variable related to emitter resistance (e.g., emitter drive current) occurs as described above, wherein the data is stored and trended. Over time, the emitter drive current is expected to have a downward trend. As described above, another variable related to emitter resistance may have an upward trend.

[0151] The method 344 includes determining whether an abrupt change in one direction (e.g., an upward trend (e.g., a significant upward trend)) is observed in the emitter drive current data or other variables related to the emitter resistance (box 354). In some embodiments, the abrupt change (e.g., an upward trend) must reach a certain threshold (e.g., 20% or another value depending on the X-ray tube filament design). If no abrupt change (e.g., an upward trend (e.g., a significant upward trend)) is observed, the method 344 includes continuing to monitor the emitter drive current or other variables related to the emitter resistance (box 348). If an abrupt change is observed, the method 344 includes determining whether the date of the abrupt change (e.g., the upward trend) is consistent with a known change in the filament drive circuit components or the X-ray tube (box 355). If the abrupt change is consistent with a known change, the method 344 includes performing a filament calibration (box 350). The system records the calibration value with or without knowing whether the X-ray tube or component has been replaced. After the filament calibration, the method 344 further includes continuing to monitor the emitter drive current or other variables related to the emitter resistance (box 348). If the sudden change is not consistent with a known change, the method 344 includes marking a (possible) X-ray tube replacement event or a filament drive circuit component replacement event (box 356). In some embodiments (e.g., when the system is connected to the manufacturer / supplier via the Internet), if a sudden change is observed, the method 344 includes providing an alert to the supplier / manufacturer of the original X-ray tube or component (box 358). In some embodiments (e.g., when the system is not in communication with the manufacturer / supplier), if a sudden change is observed, the method 344 includes providing or causing a notification (e.g., a pop-up alert) to the customer of the possible event (box 360). The next action is based on determining whether the X-ray tube or component is from the original manufacturer / supplier (box 362). If the customer has not contacted the initial manufacturer / supplier, the notification or pop-up alert may continue. If the customer does contact the initial manufacturer / supplier, a field engineer may visit the site to study the X-ray tube. If the X-ray tube or component is a tube or component of the manufacturer / supplier, the notification or pop-up alert may be stopped or turned off via a received tube certificate or input from a maintenance menu (box 364). If the X-ray tube or component is obtained from a third party, the system will flag the dose log and record obtained from the third party X-ray tube or component (box 366). In some embodiments, confirmation from the customer that the original X-ray tube or component is replaced with a third party tube may eliminate the need for a field engineer visit, and the system will flag the dose log and record obtained from the third party X-ray tube or component (box 366).

[0152] Detecting X-ray tube replacement events enables the initial supplier / manufacturer to track customers who have switched to third-party X-ray tubes and focus on those customers for the advantages of X-ray tubes authorized or approved by the supplier / manufacturer. Increasing the number of authorized X-ray tubes may result in lower production costs due to increased volume. In addition, existing contracts between initial suppliers / manufacturers can be renegotiated with customers who switch to third-party X-ray tubes. Tube quality affects the functionality of components connected to the tube, and therefore, the warranty of generators, power modules, and high-voltage cables may need to be renegotiated. In addition, even if the extended maintenance contract covers the site, the X-ray tube rights have been exhausted by using a third-party X-ray tube, and early failures of other high-voltage components may also be at least partially charged. Similarly, even if the extended maintenance contract covers the site, the X-ray tube rights have been exhausted by using a third-party X-ray tube, solving image quality problems and dose records may also be charged. Even further, due to the presence of lower quality X-ray tubes, maintenance costs associated with X-ray tubes may also be transferred to customers. In addition, suppliers / manufacturers can track the quality of third-party X-ray tubes.

[0153] Technical effects of the subject matter disclosed herein include generating one or both of a failure prediction indication of an X-ray tube or an estimate of the remaining useful life of the X-ray tube. In one implementation, a trained static tube model is used to estimate the health (e.g., thickness) of an electron emitter of an X-ray tube, which in turn can be used to predict failure of the tube and / or estimate the remaining useful life of the tube. In another implementation, replacement of the X-ray tube can be detected.

[0154] This written description uses examples to disclose the subject matter, including the best mode, and also to enable those skilled in the art to practice the disclosed subject matter, including making and using any device or system and performing any included method. The patentable scope of the disclosed subject matter is defined by the claims, and may include other examples that occur to those skilled in the art. Such other examples are intended to fall within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they include equivalent structural elements with minor differences from the literal language of the claims.

Claims

1. A method for constructing an X-ray tube model for assessing the health of an X-ray tube, the method comprising: acquiring training data points of the corresponding X-ray tube after the X-ray tube is installed in the imaging system; determining a number of operating points for use in the imaging system for the respective X-ray tube; as well as selecting a regression model from a plurality of regression models based on the number of operating points, wherein each regression model uses the training data points to derive corresponding values ​​for a plurality of coefficients, The operating points include mA-kV set points that characterize each corresponding training data point. 2 . The method of claim 1 , wherein each of the plurality of regression models comprises a different number of coefficients for the plurality of coefficients, or a different form of the regression model. 3 . The method of claim 1 , comprising constructing the X-ray tube model using the plurality of coefficients estimated by the selected regression model.

4. The method of claim 3, wherein the X-ray tube model returns an estimate of a variable related to emitter resistance in response to an input data point.

5. The method of claim 3, wherein the X-ray tube model comprises a model relating X-ray tube voltage and electron emitter current with X-ray tube current.

6. A method for assessing the health of an X-ray tube, the method comprising: acquiring data points of a corresponding X-ray tube installed in an imaging system; processing the data points using a trained X-ray tube model constructed by the method of any one of claims 1 to 5 to calculate an estimated variable related to emitter resistance; calculating a trend indicator based on the estimated variable related to the transmitter resistance; Monitor said trend indicators for sudden changes; as well as An indication is provided when the sudden change occurs on a particular day.

7. The method of claim 6, wherein monitoring the trend indicator for sudden changes comprises determining whether a certain number of exposures is above a first predetermined threshold for the particular day.

8. The method according to claim 7 comprises, when the certain number of exposures is higher than the first predetermined threshold, calculating the mean of the trend indicator of the specific day and calculating the minimum value of the difference between the mean and the average trend indicator of the previous multiple days.

9. The method of claim 8, comprising determining whether the minimum value of the difference values ​​is greater than a second predetermined threshold.

Images