Method for determining workflow phase of medical system in medical environment

Abstract

A method for determining a workflow stage of a medical system (201) in a medical environment, comprising: obtaining, by a processor, a signal (300, 400, 500) from at least one sensor (202, 203, 204, 205), where the at least one sensor (202, 203, 204, 205) is external to the medical system (201) and is configured to monitor the medical system (201) (S100); providing, by the processor, a workflow stage determination algorithm (401, 501) configured to determine workflow stage data of the medical system (201) (S200); the workflow stage data (S300) is determined by the processor by utilizing the workflow stage determination algorithm (401, 501) and the obtained signal (300, 400, 500).

Description

Technical Field

[0001] The present invention relates to a method for determining a workflow stage of a medical system in a medical environment, an apparatus for determining a workflow stage of a medical system in a medical environment, a system, and a computer program. Background Art

[0002] Medical processes of medical systems (e.g., diagnosis in an MRI system) are well known in the prior art. These medical processes typically include different workflow stages. It is crucial to know in what workflow stage such a medical process is.

[0003] It is now apparent that there is a need to provide a method for determining a workflow stage of a medical system. Summary of the Invention

[0004] In view of the above, an object of the present invention is to provide a method that allows for an improved determination of the workflow stage of a medical system in a medical environment.

[0005] The subject matter of the independent claims solves these and other objects, which become apparent after reading the following description. The present invention provides a method, an apparatus, a system, and a computer program. The dependent claims relate to preferred embodiments of the present invention.

[0006] In one aspect of the present disclosure, there is provided a method for determining a workflow stage of a medical system in a medical environment, including:

[0007] obtaining, by a processor, a signal from at least one sensor, wherein the at least one sensor is external to the medical system and is configured to monitor the medical system;

[0008] providing, by the processor, a workflow stage determination algorithm configured to determine workflow stage data of the medical system;

[0009] determining, by the processor, the workflow stage data by using the workflow stage determination algorithm and the obtained signal.

[0010] As used herein, the term "workflow stage" will be understood broadly and may refer to any step or state in a process performed by a medical system. The workflow stage may include a preparation stage, a startup stage, an operation stage, a waiting stage, a subsequent stage, a startup stage of a part of the medical system, a positioning stage of a part of the medical system or an object to be processed by the medical system. The workflow stage may be described by workflow stage data.

[0011] As used herein, the term "workflow stage data" is to be understood broadly and can relate to any data configured to describe a workflow stage. Workflow stage data can include the time duration of a workflow stage, the time of a workflow stage, the name of a workflow stage (e.g., the startup stage), the medical system status (e.g., the medical system is off, the medical system is paused). Workflow stage data can include the workflow stage.

[0012] As used herein, the term "medical system" is to be understood broadly and can relate to any system used in a medical environment. A medical system can include one or more medical units used in combination (e.g., an MRI system and a patient table). A medical system can include one or more subunits (e.g., the positioning unit of an MRI system, the coil of an MRI system). A medical system can include an MRI system, an X-ray system, and / or a CT system.

[0013] As used herein, the term "medical environment" is to be understood broadly and can relate to any environment in which a medical system is located. A medical environment can be a treatment room in a hospital or a medical practice.

[0014] As used herein, the term "signal" is to be understood broadly and can relate to any data provided by a sensor. A signal can include analog data or digital data. A signal can include raw data or at least partially processed data. A signal can be, for example, a noise signal, a temperature signal, a vibration signal, an acoustic signal, and / or an infrared signal. A signal can include different signals from one or more sensors. A signal can be obtained by a processor by means of a transmission module (e.g., a wireless data interface, a bus system between the processor and the sensor).

[0015] As used herein, the term "sensor" is to be understood broadly and can relate to any measuring device configured to measure a physical value. A sensor can be a vibration sensor, an acoustic emission sensor, a temperature sensor, and / or a noise sensor. A sensor can be a single entity or distributed over several entities. A sensor can be based on one or more physical measurement principles. A sensor can include one or more sensors (e.g., a sensor array, or a temperature sensor and a vibration sensor).

[0016] As used herein, the term "external" is to be understood broadly and means that the sensor is not part of the medical system. In other words, the sensor is a separate unit relative to the medical system. The sensor can be configured to operate independently of the medical system. The sensor (e.g., a vibration sensor) can be placed on the housing of the medical system to monitor the medical system.

[0017] As used herein, the term "workflow phase determination algorithm" will be understood broadly and can relate to any logic unit configured to determine the workflow phase of a medical system. The workflow phase determination algorithm can be an analytical algorithm and / or a heuristic algorithm. The workflow phase determination algorithm can include a regression model or an AI model (e.g., a machine learning model). The workflow phase determination algorithm can include a classification algorithm.

[0018] In other words, the basic idea of the present invention can include: analyzing signals from one or more sensors placed near the medical system in order to determine the workflow phase of the medical system. The method can be a computer-implemented method executed by a processor. This can be advantageous because it helps to determine at which stage the medical process performed by the medical system currently is. Information about the workflow phase can be used to analyze the efficiency of the medical system and / or the medical process. This can be beneficial for identifying bottlenecks and / or malfunctions. This can be beneficial for triggering accompanying processes or analyses. This can be advantageous in terms of quality control. This can be advantageous in terms of reliability because a method independent of the medical system is used (due to external sensors and / or an external processor).

[0019] Due to the aging population worldwide, the scarcity of clinical staff, and the pressure on healthcare budgets, there may be a strong need to improve the efficiency of clinical operations and enhance the daily experience of clinical staff. Generally, these needs can be expressed in terms of the so-called quadruple aim: improved patient experience, better health outcomes, improved staff experience, and lower cost of care.

[0020] The proposed invention can address the following problems: Many different modalities may have to be connected and equipped with additional sensors in order to obtain an overall overview of all relevant processes in a care environment (e.g., a catheter laboratory or a similar hospital environment). The design of sensors typically may involve expensive and lengthy trajectories (involving formal verification, validation, and regulatory release), and a large installation base needs to be established before the widespread benefits of such a system can be obtained. Ideally, a method independent of a specific vendor can be applied, which is potentially capable of sensing a wide variety of instruments used in clinical processes (including third-party devices).

[0021] Another problem overcome by the present invention may be that, without the present invention, all fault detection and prediction algorithms would need to run in parallel in an always-on mode. This may have some serious technical drawbacks: Edge devices with high computing capabilities may be required to support algorithm execution. This may consume high power and may result in expensive devices. The algorithms may continuously process sound / vibration / RF signals, even when it is irrelevant. This may lead to false positives or false warnings. If run for a longer period in the presence of unexpected or noisy inputs (e.g., background sounds or sounds from other components / parts of a medical system), some algorithms may drift or get stuck in local loops.

[0022] For this reason, it would be advantageous to intelligently select the correct algorithm at the right moment (i.e., during those time periods when an estimate can be optimally made (e.g., for gantry angle detection) or when a fault is likely to occur (e.g., an arc discharge can only occur when an X-ray is generated and the high-voltage system is active)).

[0023] Another additional benefit may be that the present invention allows triggering of a target anomaly detector that can selectively trigger discarding data or sending data to the cloud. Anomaly detection may be easier when the anomaly detection only needs to focus on specific workflow phases or subunits of a medical system where normal behavior is known.

[0024] To achieve the above objectives, it may be necessary to monitor workflow phases to populate dashboards and identify inefficiencies (slow / delayed processes, waiting times, repeated diagnostic processes, instrument downtime, etc.) and reduce the administrative burden on healthcare staff. The latter can be accomplished, for example, by collecting data from different modalities used in the clinical process and (e.g., using artificial intelligence algorithms) identifying key information and placing the key information on the workflow phase timeline. Conversely, key information can also be identified by tracking important phase transitions in the workflow.

[0025] Another reason for classifying the workflow into different phases may be that specific elements of the system are active in each phase and generate characteristic sounds, RF fields, and / or other measurable signals. Generally, faults or degradations of subunits of a medical system may only become detectable during these active phases. This would be advantageous for continuously performing different target monitoring and fault prediction algorithms (each target monitoring and fault prediction algorithm in the workflow phase that may be related to the faulty element).

[0026] This may mean that the ability to track (patient, staff, and instrument) workflows may be a key enabling factor for all requirements (dashboard population, management automation, and target prediction algorithms).

[0027] The present invention can propose to equip the staff and the environment with one or more sensors (e.g., wearing RF-ID badges, position sensors for locating mobile instruments).

[0028] The present invention can focus on the implementation of different monitoring and fault prediction algorithms, each algorithm being implemented at the right moment.

[0029] The present invention proposes a method that allows for patient / process workflow stage detection by processing signals from one or more sensors that can be sensitive in all directions (i.e., omni-sensitive). The sensors can be arranged next to or on the modalities used in the healthcare process. The results can be applied to dashboard filling, management automation, or triggering of target algorithms for predictive maintenance or fault monitoring.

[0030] The present invention can generally be used in a hospital environment for locations where multiple modalities and devices are used to diagnose or treat patients. The proposed system can include, for example, a computed tomography (CT) system, an electrically adjustable hospital bed, a C-arm X-ray system, a magnetic resonance (MR) system, an ultrasound system, a catheter system, a respiratory system (such as a ventilator). Some of these systems can have a fixed position in the room. Some of these systems can be flexibly moved around or even implemented on a mobile cart (depending on the process or process stage).

[0031] In such an environment, multiple sensors (e.g., multiple sensors of different types) can be positioned in place. The sensors can be arranged next to or on the instrument (e.g., to sense vibrations), or the sensors can be arranged at a certain distance from the instrument (e.g., to detect sounds or RF electromagnetic emissions from the instrument) or a combination thereof.

[0032] An interesting and important observation may be that a very important part of the signal content is outside the audible spectrum and can only be recorded using a broadband microphone with a high sampling frequency. Therefore, the sensors used in the proposed method can be broadband microphones with a high sampling frequency.

[0033] For example, in the field of CT systems, the following workflow stages can be determined: collimator adjustment, gantry rotation (acceleration, steady-state rotation, coasting), and X-ray generation.

[0034] The present invention can allow for uniquely identifying workflow-related operations, such as: movement of the patient table in different directions and angles, detection of the opening and closing of an air fan and / or adjustment at different speeds.

[0035] In the field of C-arm X-ray systems, the following workflow phases can be determined: the movement of the C-arm along the ceiling track, the movement of the C-arm around the patient in different orientations, and the movement of the cantilever monitor. This can be applied to both stationary C-arm systems and mobile C-arm systems.

[0036] In the field of CT systems, the following workflow phase can be determined: the movement of the CT system on a ground track in the case of a CT configuration on a track.

[0037] In the field of MR systems, the following workflow phases can be determined: the detection of the magnetic field (in particular the magnetic field modulation gradient) in the MR system and / or the detection of RF activation pulses.

[0038] In the field of DXR systems, the following workflow phase can be determined: the movement of the XRT and / or the detector along a ceiling suspension (e.g., a track system).

[0039] According to one embodiment, the method may include:

[0040] using, by the processor, the workflow phase data to select at least one evaluation algorithm corresponding to the workflow phases included in the workflow phase data for a medical system; and / or

[0041] providing, by the processor, the determined workflow phases for further processing, wherein the further processing includes displaying the workflow phase data and / or analyzing the workflow phase data.

[0042] The term "evaluation algorithm" as used herein will be understood broadly and may relate to any algorithm configured to analyze a workflow phase. An evaluation algorithm may relate to an algorithm configured to analyze a sub-unit of a medical system (e.g., a coil or a patient table of an MRI system). An evaluation algorithm may detect a fault in at least a part of a medical system (e.g., start-up noise of a coil of an MRI system). An evaluation algorithm may locate a fault in at least a part of a medical system (e.g., an uncommon vibration in a drive of a C-arm of a projection X-ray imaging system). The method may provide multiple evaluation algorithms.

[0043] The proposed method for determining the workflow phases of a medical system in a medical environment may include selecting one or more evaluation algorithms according to the determined workflow phases. For example, when the workflow phase is "determine start-up of MRI coils", the evaluation algorithm "quality control coil of MRI system" is selected and executed.

[0044] The proposed method for determining the workflow stage of a medical system in a medical environment may include providing a selection table including one or more evaluation algorithms. Based on the determined workflow stage, one or more evaluation algorithms may be selected from the selection table.

[0045] The evaluation algorithm may be executed on a separate entity (e.g., the cloud). In the case of only selecting the evaluation algorithm, the processor executing the above method may not necessarily need to be powerful enough to execute the evaluation algorithm. The processor may also only provide a trigger signal for another entity to execute the selected evaluation algorithm. This may be advantageous in terms of the resource efficiency of the processor.

[0046] This may be advantageous in terms of the quality control, reliability, efficiency, and / or usability of the medical system. This may be further advantageous because the evaluation algorithm is only executed when needed. This may save resources and may lead to better results (due to fewer noise signals from other workflow stages).

[0047] The display of the determined workflow stage data may include presenting the determined workflow stage on a screen, dashboard, tablet, and / or mobile phone. This may be advantageous in terms of user-friendliness, reduced complexity, and / or real-time analysis capabilities.

[0048] The analysis of the workflow stage data may involve a management analysis of the workflow stage (e.g., comparison of the determined workflow stage duration with a reference value). This may be advantageous in terms of improved efficiency. As used herein, the term "management analysis" will be understood broadly and may involve any process analysis of the workflow stage. The management analysis may include determining one or more key performance indicators, such as system availability and / or productivity.

[0049] Workflow stage determination can be used to establish a timeline and use it as the basis for a post-process report. Additionally, automated management (e.g., extracting key frames from a fluoroscopy stream) can benefit from the workflow stage data. The term management may include management tasks (e.g., extracting CT images for document processing).

[0050] According to one embodiment, the at least one evaluation algorithm may include at least one of the following: a fault detection algorithm for at least a part of the medical system, a position estimation algorithm for at least a part of the medical system, a fault localization algorithm for at least a part of the medical system, and a fault prediction algorithm for at least a part of the medical system.

[0051] As used herein, the term "fault detection algorithm" will be understood broadly and can relate to any algorithm configured to detect faults in a workflow phase and / or the corresponding part of a medical system (e.g., a coil of an MRI system) that is active during that workflow phase. The fault detection algorithm can include anomaly detection. This can be advantageous as it allows targeting of the monitoring subunit.

[0052] As used herein, the term "position estimation algorithm" will be understood broadly and can relate to any algorithm that allows determination of the position of a movable part of a medical system (e.g., the C-arm of a projection X-ray imaging system). The position estimation algorithm can include multi-agent models. For example, based on a noisy signal indicating the position of a part of the medical system, the position of that part of the medical system can be determined. This can be advantageous in terms of quality control as it reveals the current position of a part of the medical system. This information can be used for further analysis or can be used to readjust the position of the part of the medical system.

[0053] As used herein, the term "fault localization algorithm" will be understood broadly and can relate to any algorithm configured to identify faults in a part of a medical system. The fault can relate to at least a part of the medical system or even to a specific area of a part of the medical system. This can be advantageous in terms of quality control.

[0054] As used herein, the term "fault prediction algorithm" will be understood broadly and can relate to any algorithm configured to predict faults based on the acquired signals and / or additional information / data related to the medical system. The fault prediction algorithm can be based on anomaly analysis. This can be advantageous in terms of quality control.

[0055] In one embodiment, the method can include: executing the at least one selected evaluation algorithm and, in particular, presenting the results of the at least one selected evaluation algorithm executed on a user interface. The method can execute the selected evaluation algorithm on a processor. Alternatively, the method can execute the selected evaluation algorithm on a separate entity (e.g., on a workstation or in the cloud). This can be advantageous as it allows use of a processor with low performance requirements since the evaluation algorithm with high capability requirements is executed on a separate powerful entity.

[0056] The method can perform target monitoring and / or target fault prediction algorithms at relevant times (i.e., during relevant workflow stages of the process). For this purpose, the method can use a workflow stage determination algorithm to enable and / or trigger a set of evaluation algorithms focusing on specific faults of the system and / or specific components (e.g., an evaluation algorithm focusing on faults in the collimator, an evaluation algorithm for detecting and locating arc discharges, and / or an evaluation algorithm for predicting or locating other mechanical (wear-induced) faults).

[0057] The benefits of selecting and executing the correct evaluation algorithms can be that different evaluation algorithms are only active when needed. Thus, these evaluation algorithms do not consume unnecessary computing power or energy, or create unnecessary false alarms or warnings.

[0058] In one embodiment, the method can include: the processor analyzes the determined workflow stage data based on corresponding reference workflow stage data to obtain an analysis result, and the analysis result particularly includes the identification of an anomaly in the workflow stage data. The method can include: the processor transmits the obtained signal from the at least one sensor to an analysis platform via a data connection for further analysis based on the analysis result or to delete the obtained signal based on the analysis result, and the analysis platform is particularly a platform on a cloud system.

[0059] As used herein, the term "anomaly" will be broadly understood and can refer to any identified deviation of the determined workflow stage data from a reference value in the corresponding reference workflow stage data.

[0060] As used herein, the term "analysis platform" will be broadly understood and can refer to any separate entity related to the processor. The separate entity can be configured to execute one or more analysis algorithms. The analysis platform can be a cloud application or embedded in a workstation.

[0061] In other words, the method can analyze the obtained workflow stage data by comparing the obtained workflow stage data with the reference data of the corresponding determined workflow stage. In the case of detecting an anomaly in the workflow stage data, the raw data obtained from the sensor is sent to the analysis platform for further analysis, otherwise the raw data is deleted. This can be advantageous because unimportant data does not have to be stored or analyzed. This can advantageously save resources.

[0062] In one embodiment, the at least one selected algorithm may be executed on a cloud system, wherein the processor and the cloud system are connected by a data connection, and wherein, in particular, the processor is implemented on an edge device. The data connection may include a bus system or a wireless data interface. This may be advantageous in terms of resource efficiency because the processor for executing the method may be in a lower performance configuration compared to the cloud system. The processor may be part of an edge device (e.g., a tablet or a smartphone). The edge device may be a device at a medical system site.

[0063] In one embodiment, the method may include: performing signal processing on the acquired signal, wherein the signal processing includes signal conditioning, digitization, transformation, and in particular feature extraction. This may be advantageous in terms of data quality, reliability, and quality control.

[0064] Before determining the algorithm for the workflow phase, some signal processing is performed. The signal processing may include conditioning and / or digitization and / or feature extraction. Feature extraction may be accomplished, for example, by spectral analysis using wavelet transform or short-time Fourier transform (STFT). Feature extraction may include using one or more band-pass filters in combination with a power detector to extract at least part of the unique spectral fingerprint (i.e., features) of the medical system at a certain workflow phase.

[0065] In one embodiment, the workflow phase determination algorithm may be based on a machine learning algorithm that is trained to predict the workflow phase data, in particular the workflow phase included in the workflow phase data. The machine learning model may preferably include decision tree, naive Bayes classification, nearest neighbor, neural network, convolutional neural network, generative adversarial network, support vector machine, linear regression, logistic regression, random forest, and / or gradient boosting algorithm.

[0066] In one embodiment, the workflow phase determination algorithm may be based on a classification algorithm, in particular a classification algorithm configured to determine the workflow phase based on the acquired signal.

[0067] In one embodiment, the at least one sensor may include at least one of the following: an acoustic sensor, a vibration sensor, an electromagnetic sensor, and a thermal sensor. The at least one sensor may be placed near the medical system or on the surface of the medical system or at least part of the medical system for the intended use.

[0068] In one embodiment, the medical system may include one or more parts, and the medical system may include at least one of the following: a CT system, an X-ray system, an MR system, an ultrasound system, a catheter system, a respiratory system, and a patient table.

[0069] In one embodiment, the signal obtained can originate from a source configured to generate a synthetic signal, and the source can be located on the medical system. The synthetic signal can be, for example, a noise signal or a thermal signal. By comparing the signal obtained with a corresponding reference signal, the position of at least part of the medical system can be advantageously determined. For example, a sensor can thus capture a sound source originating from a rotating gantry.

[0070] Another aspect relates to a device for determining the workflow stage of a medical system in a medical environment, the device including modules for performing the steps of the above method. The modules can include a processor, a computing unit, a desktop PC, a workstation. The modules can be a single entity or distributed over several entities. The modules can include one or more interfaces for communicating with one or more sensors and / or other entities (such as, a cloud, an analysis platform) and / or a user.

[0071] Another aspect relates to a system, the system including: the above device, a medical system, at least one sensor, and optionally a user interface. The medical system can be an MRI system, a CT system, etc. The system can include a screen or a dashboard for presenting the determined workflow stage and / or analyzing the results of the one or more evaluation algorithms.

[0072] Another aspect relates to a computer program including instructions which, when executed by a computer, cause the computer to perform the above method. Another aspect relates to a computer-readable medium including instructions which, when executed by a computer, cause the computer to perform the above method.

[0073] A computer program may be stored on a computer unit, which may also be part of an embodiment. The computer unit may be configured to execute or cause the execution of the steps of the above-described method. Additionally, the computer unit may be configured to operate the components of the above-described device. The computing unit can be configured to automatically operate and / or execute the commands of a user. The computer program may be loaded into the working memory of a data processor. Thus, the data processor may be equipped to execute the method according to one of the foregoing embodiments. This exemplary embodiment of the invention covers both computer programs that use the invention from the start and computer programs that turn existing programs into programs that use the invention by means of an update. Additionally, the computer program may be capable of providing all the necessary steps to implement the process of an exemplary embodiment of the above-described method. According to a further exemplary embodiment of the invention, a computer-readable medium is provided, for example, a CD-ROM, a USB stick, etc., on which a computer program is stored, which computer program has been described in the previous sections. The computer program may be stored and / or distributed on a suitable medium (for example, an optical storage medium or a solid-state medium provided together with or as part of other hardware), but may also be distributed in other forms (for example, via the Internet or other wired or wireless telecommunication systems). However, the computer program may also be presented via a network such as the World Wide Web and be downloadable from such a network into the working memory of a data processor. According to a further exemplary embodiment of the invention, a medium for making a computer program available for download is provided, which computer program is arranged to execute the method according to one of the foregoing embodiments of the invention.

[0074] Note that the above embodiments may be combined with each other regardless of the aspects involved. Thus, the method may be combined with the structural features of devices and / or systems of other aspects, and similarly, devices and systems may be combined with the features of each other and may also be combined with the features described above with respect to the method.

[0075] These and other aspects of the invention will become apparent and be elucidated with reference to the embodiments described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Exemplary embodiments of the invention will be described with reference to the following drawings.

[0077] Figure 1 is a schematic diagram of a method for determining the workflow phases of a medical system in a medical environment;

[0078] Figure 2 schematically shows a device for determining the workflow phases of a medical system in a medical environment;

[0079] Figure 3Schematically shows a system for determining the workflow phases of a medical system in a medical environment;

[0080] Figure 4 Shows an exemplary signal analysis result;

[0081] Figure 5 Schematically shows an example for selecting an evaluation algorithm; and

[0082] Figure 6 Schematically shows an example for anomaly detection.

[0083] List of reference numerals:

[0084] S100 Obtain signal

[0085] S200 Provide workflow phase determination algorithm

[0086] S300 Determine workflow phase data

[0087] 100, 207 Devices

[0088] 200 System

[0089] 201 Medical system

[0090] 202, 203, 204, 205 Sensors

[0091] 206 Data connection

[0092] 300, 400, 500 Signals

[0093] 301 Spectrogram

[0094] 302 Start-up phase

[0095] 303 Accelerated rotation phase

[0096] 304 Steady rotation phase

[0097] 305 X-ray emission phase

[0098] 306 Arc phase

[0099] 307 Gantry glide phase

[0100] 401, 501 Workflow phase determination algorithms

[0101] 402 Selection algorithm

[0102] 403, 404, 405, 406, 503 Evaluation algorithms

[0103] 502 Cache unit

[0104] 504 Data connection

[0105] 505 Analysis platform

[0106] 506 Local side

[0107] 507 External side Detailed implementation manners

[0108] Figure 1 It is a schematic diagram of a method for determining a workflow stage of a medical system in a medical environment according to the present disclosure.

[0109] Step 100 includes obtaining, by a processor, signals from at least one sensor, where the at least one sensor is external to the medical system and is configured to monitor the medical system. In this example, the processor is part of an edge device (such as a tablet computer) used in the medical environment of the medical system. In this example, the medical system is a CT system. In this example, the at least one external sensor is an acoustic sensor arranged separately from the CT system. The acoustic sensor is configured to measure acoustic signals (i.e., sounds here). The acoustic sensor is connected to the edge device (by means of a wireless interface) to transmit signals from the acoustic sensor.

[0110] Step 200 includes providing, by the processor, a workflow stage determination algorithm (S200) configured to determine workflow stage data of the medical system.

[0111] In this example, the workflow stage determination algorithm is a classification algorithm based on an AI algorithm trained to determine the workflow stage of the medical system. The AI algorithm can be trained with existing data from the field or from a test setup. Alternatively, the workflow stage determination algorithm can be an analysis algorithm.

[0112] In this example, the obtained signals are additionally processed, for example. The signal processing of the obtained signals can include signal conditioning, digitization, transformation, and especially feature extraction. The optional signal processing is performed before step S300. In this example, the signal processing is performed by the processor. Feature extraction can be accomplished, for example, by spectral analysis using wavelet transform or short-time Fourier transform (STFT).

[0113] Step 300 includes determining, by the processor, workflow stage data by using the workflow stage determination algorithm and the obtained signals. In this example, the workflow stage data can include one of the following: duration, time, assignment to a workflow stage, workflow status. In this example, the workflow stage "the gantry of the CT system is moving" is determined.

[0114] Additionally, the method may include: using, by a processor, workflow stage data to select, for a medical system, at least one evaluation algorithm corresponding to a workflow stage included in the workflow stage data; and / or providing, by the processor, the determined workflow stage data for further processing, where the further processing includes displaying the workflow stage data and / or analyzing the workflow stage data.

[0115] In this example, the method selects a position estimation algorithm for the gantry of a CT system from a plurality of evaluation algorithms. The position estimation algorithm may use the acquired sensor signals to determine the position (i.e., gantry angle) of the gantry of the CT system.

[0116] Additionally, other evaluation algorithms may be selected, for example, a fault prediction algorithm for the collimator of a CT system, or a fault detection algorithm for mechanical wear.

[0117] The method may also execute the selected evaluation algorithm. In this example, such an execution operation may be performed in a cloud application. Thus, the method triggers such an execution operation in the cloud. The processor and the cloud system may be connected via a data connection. Such an execution operation may alternatively be performed by the processor.

[0118] The determined workflow stage data may be presented to a user on an edge device, a screen, and / or a dashboard. Results from the evaluation algorithm may also be presented to the user via the above-mentioned module and / or user interface (such as an edge device).

[0119] Additionally, the method may further include analyzing the determined workflow stage data based on corresponding reference workflow stage data to obtain an analysis result. The analysis result may particularly include an identification result of an anomaly in the workflow stage data. The processor may transmit the acquired signals from at least one sensor to an analysis platform (particularly a platform on a cloud system) via a data connection for further analysis based on the analysis result. Alternatively, the processor may delete the acquired signals based on the analysis result. In this example, no anomaly was detected in the workflow stage "the gantry of the CT system is moving", and thus the acquired signals have been deleted accordingly.

[0120] Figure 2 Device 100 for determining the workflow stage of a medical system in a medical environment is shown.

[0121] Device 100 includes modules for performing the steps of the above method. In this example, the module includes a processor of an edge device. In this example, the module includes an interface communicating with an acoustic sensor and at least one interface communicating with other entities (such as a cloud, an analysis platform) and a user.

[0122] Figure 3 System 200 is shown for determining the workflow phases of a medical system in a medical environment.

[0123] System 200 includes device 207, for example, the device described above in the context of device 100. The medical environment can be a hospital. System 200 includes medical system 201. In this example, medical system 201 is a CT system. System 200 includes four acoustic sensors 202 to 205 that are linked to the above-described device 207 by means of a wired connection 206. System 200 may also optionally include a user interface (not shown) and / or a screen or dashboard (not shown) for presenting the determined workflow phases and / or the results of one or more evaluation algorithms. Note that the number of acoustic sensors is merely exemplary, and more or fewer sensors may be provided.

[0124] Figure 4 An exemplary signal analysis result is shown.

[0125] Figure 4 The signal waveform 300 of the acoustic signal of the sensor in the time series and the corresponding spectrogram 301 are shown. The signal is from a sensor monitoring a CT system. In the spectrogram, different workflow phases of the CT system can be identified. Segment 302 of the spectrogram relates to the start phase including collimator adjustment. Segment 303 of the spectrogram relates to the accelerating rotation phase of the CT system. Segment 304 of the spectrogram relates to the stable rotation phase of the CT system. Segment 305 of the spectrogram relates to the X-ray emission phase of the spectrogram. Segment 306 of the spectrogram relates to the arc phase of the CT system. Segment 307 relates to the gantry coasting phase of the CT system.

[0126] Figure 5 An example for selecting an evaluation algorithm according to the present disclosure is schematically shown.

[0127] The method obtains a signal 400 from a sensor. A workflow stage determination algorithm 401 determines a workflow stage based on the obtained signal. The method then selects, by means of a selection algorithm 402, one or more evaluation algorithms 403, 404, and / or 405 based on the determined workflow stage. The evaluation algorithm 403 may include a fault detection / prediction algorithm for a collimator. The evaluation algorithm 404 may include a fault detection / prediction algorithm for arc discharge detection. The evaluation algorithm 405 may include a fault detection / prediction algorithm for mechanical wear. The selection algorithm 402 may trigger the execution of one or more evaluation algorithms. The selection algorithm 402 may also select a position estimation algorithm 406. The position estimation algorithm 406 may relate to a gantry estimation algorithm that provides an estimated gantry angle as a result. The position estimation algorithm 406 may receive the obtained sensor signal 400 as an input. The position estimation algorithm 406 may provide, for example, the estimated gantry angle to the evaluation algorithms 404 and 405 as an input. The evaluation algorithms 403 to 405 may provide one or more evaluation results for further processing or display to a user.

[0128] For example, when the position estimation algorithm is used to estimate the gantry angle, the position estimation algorithm may include preprocessing, sound source detection, and a controller loop. The preprocessing may include filtering the raw data obtained by the sensor to capture only the relevant frequencies of the sound source that the sensor is attempting to track. The sound source detection may include detecting when the source moves past a certain predetermined point on the gantry rotation path. This can be achieved by using amplitude modulation or frequency modulation. The amplitude modulation may relate to the aspect where the amplitude is highest when the source passes to the sensor (e.g., a microphone). The frequency modulation may relate to the aspect that due to the speed difference between the sensor and the rotating sound source, the frequency of the observed signal may be shifted due to the Doppler effect. A possible detection algorithm may utilize the frequency modulation to observe the Doppler effect and map it to the gantry orientation. The controller loop may relate to the aspect that the detection of the sound source may introduce an error in the orientation estimation. To eliminate this error, for example, an oscillator may be used in the control loop. A possible implementation of the controller loop may be a digital phase-locked loop (DPLL). The output of the controller loop will be an angle corresponding to the orientation of the gantry angle. When a separate system then detects that a fault has occurred, the gantry orientation angle can be read out. To provide user feedback, a small interface window (a physical interface on an edge device or a digital interface in the cloud) may show the gantry orientation angle that is continuously tracked during operation.

[0129] Figure 6 An example for anomaly detection according to the present disclosure is schematically shown.

[0130] The method includes obtaining signal 500 from a sensor. A workflow stage determination algorithm 501 determines a workflow stage based on the obtained signal. The raw data of signal 500 can be cached by a cache unit 502 (i.e., the local storage device of the processor). An evaluation algorithm 503 can determine an anomaly by comparing the obtained signal with a reference signal for the corresponding workflow stage, based on the determined workflow stage and the obtained signal.

[0131] The method can include transmitting the raw data of signal 500 via a data connection 504 to an analysis platform 505 (e.g., the cloud) by means of a processor (not shown) for further analysis, so as to perform a detailed analysis in case an anomaly is detected. The method can include deleting the raw data of signal 500 by means of the processor in case no anomaly is detected. Figure 6 The left side 506 in [description] represents the local side (i.e., the edge side), while the right side 507 represents the external side related to the cloud.

[0132] Workflow stage determination can enable more efficient anomaly detection. For each workflow stage in the workflow, the currently incoming sensor signal can be compared with the expected sensor signal (derived from measurements during normal operation on a healthy instrument). Anomaly detection can include identifying and / or detecting deviations from the normal signal (i.e., anomalies) and / or detecting slowly evolving trends. When the sensor signal starts to deviate from the normal signal, an evolving trend can be detected. When such a deviation or trend is detected, the method can decide to send a sample of the raw data of the obtained signal from the sensor to the cloud for further detailed analysis, while under normal circumstances, the raw data of the sensor can be deleted.

[0133] The method can allow very limited local processing. The actual detailed analysis can be completed in the cloud without always sending all the raw data obtained from the sensor to the cloud.

Claims

1. A method for determining a workflow stage of a medical system (201) in a medical environment, comprising: obtaining, by a processor, signals (300, 400, 500) from at least one sensor (202, 203, 204, 205), wherein the at least one sensor (202, 203, 204, 205) is external to the medical system (201) and is configured to monitor the medical system (201) (S100); providing, by the processor, a workflow stage determination algorithm (401, 501) configured to determine workflow stage data of the medical system (201) (S200); determining, by the processor, the workflow stage data by utilizing the workflow stage determination algorithm (401, 501) and the obtained signals (300, 400, 500) (S300).

2. The method according to claim 1, comprising: selecting, by the processor, at least one evaluation algorithm (403, 404, 405, 406, 503) corresponding to the workflow stage included in the workflow stage data for the medical system (201) by using the workflow stage data; and / or providing, by the processor, the determined workflow stage for further processing, wherein the further processing includes displaying the workflow stage and / or analyzing the workflow stage.

3. The method according to claim 1 or 2, wherein The at least one evaluation algorithm (403, 404, 405, 406, 503) includes at least one of the following: a fault detection algorithm for at least a part of the medical system (201), a position estimation algorithm for at least a part of the medical system (201), a fault location algorithm for at least a part of the medical system (201), a fault prediction algorithm for at least a part of the medical system (201).

4. The method according to any one of the preceding claims, comprising: executing at least one selected evaluation algorithm (403, 404, 405, 406, 503) and presenting, in particular on a user interface, the result of the at least one selected evaluation algorithm (403, 404, 405, 406, 503) that has been executed.

5. The method according to any one of the preceding claims, comprising: analyzing, by the processor, the determined workflow stage data based on corresponding reference workflow stage data to obtain an analysis result, the analysis result particularly including the identification of an anomaly in the workflow stage data, transmitting, by the processor, the obtained signals (300, 400, 500) from the at least one sensor (202, 203, 204, 205) to an analysis platform (505) via a data connection (504) for further analysis based on the analysis result or deleting the obtained signals (300, 400, 500) based on the analysis result, the analysis platform being particularly a platform on a cloud system.

6. The method according to any one of claims 2 to 5, wherein, The at least one selected algorithm is executed on a cloud system, wherein the processor and the cloud system are connected by a data connection (504), and wherein, in particular, the processor is implemented on an edge device.

7. The method according to any one of the preceding claims, further comprising: Signal processing is performed on the acquired signals (300, 400, 500), wherein the signal processing includes signal conditioning, digitization, transformation, and in particular feature extraction.

8. The method according to any one of the preceding claims, wherein, The workflow stage determination algorithms (401, 501) are based on machine learning algorithms, which are trained to predict the workflow stage data, in particular the workflow stage included in the workflow stage data.

9. The method according to any one of the preceding claims, wherein, The workflow stage determination algorithms (401, 501) are based on classification algorithms, in particular classification algorithms configured to determine the workflow stage based on the acquired signals (300, 400, 500).

10. The method according to any one of the preceding claims, wherein, The at least one sensor (202, 203, 204, 205) includes at least one of the following: an acoustic sensor, a vibration sensor, an electromagnetic sensor, and a thermal sensor.

11. The method according to any one of the preceding claims, wherein, The medical system (201) includes one or more parts, and wherein the medical system (201) includes at least one of the following: a CT system, an X-ray system, an MR system, an ultrasound system, a catheter system, a respiratory system, and a patient table.

12. The method according to any one of the preceding claims, wherein, The acquired signals are derived from a source configured to generate synthetic signals, and wherein the source is positioned on the medical system (201).

13. An apparatus for determining a workflow stage of a medical system (201) in a medical environment, the apparatus including modules configured to perform the steps of the method according to any one of claims 1 to 12.

14. A system, comprising: The apparatus according to claim 13, the medical system (201), the at least one sensor (202, 203, 204, 205), and an optional user interface.

15. A computer program including instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 12; and / or a computer-readable medium including instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 12.

Images