Monitoring the treatment of an object

Abstract

Device for monitoring the treatment of an object (4) comprising the device (1) - an optical unit (2) designed to direct ionizing radiation (3) onto the object (4); - a filter element (5) which can be arranged in a beam path of the ionizing radiation (3) in order to attenuate a portion of the ionizing radiation (3); - an imaging unit (6) configured to detect components of the ionizing radiation (3) passing through the object (4) in order to produce an image (8) of part of the object (4); - an eye-tracking system (26) designed to detect the eye movements of an observer (25); and a control unit (10) designed to control the position of the filter element (5) depending on the eye movement during a first operating mode; characterized by the fact that the control unit (10) is set up to - to identify a predefined eye movement sequence depending on the eye movement; - to switch from the first operating mode to a second operating mode once the eye-tracking sequence has been identified; and - during the second operating mode, to control the position of the filter element (5) depending on an image analysis of the image (8).

Description

[0001] The present invention relates to a device for monitoring the treatment of an object, comprising an optical unit configured to direct ionizing radiation onto the object, a filter element that can be arranged in a beam path of the ionizing radiation and configured to attenuate a portion of the ionizing radiation, an imaging unit configured to detect portions of the ionizing radiation passing through the object in order to produce an image of a portion of the object, a gaze detection system configured to detect the gaze movement of an observer, and a control unit configured to control the position of the filter element as a function of the gaze movement during a first operating mode.The invention further relates to a training device for a model for image analysis during the monitoring of the treatment of an object, a method for training a model for image analysis during the monitoring of the treatment of an object, as well as a method for monitoring the treatment of an object, a computer program and a computer-readable storage medium.

[0002] When treating an object with ionizing radiation, for example, during X-ray-guided catheterization in a therapeutic procedure on a patient or a non-medical procedure on another object, it can be advantageous to minimize the dose of ionizing radiation to which an observer, such as a physician, and / or the object is exposed. This can be achieved by inserting a filter element into the beam path of the ionizing radiation. This filter element defines a priority area within which the ionizing radiation is not attenuated to obtain the best possible image results, while outside this priority area, the ionizing radiation is attenuated.Since the position of a region of interest (ROI), which should be imaged with the highest possible image quality, can change during treatment, the position of the filter element can be controlled accordingly.

[0003] If the position of the filter element is controlled by an eye-tracking system based on the observer's gaze movement, the system may lose track of the user's or observer's current viewing position, meaning it cannot detect the gaze movement, or cannot detect it correctly, or for other reasons it may be impossible to position the filter element correctly. In such a case, the filter element can, for example, be completely removed from the optical path to ensure that the region of interest (ROI) continues to be imaged with good image quality.

[0004] A disadvantage is that this eliminates the filter element's primary function, namely dose reduction for the observer and / or object. Another disadvantage is that removing and reinserting the filter element into the beam path can lead to disturbing artifacts in the image.

[0005] Document US 5,278,887 A describes a device and method for reducing the X-ray dose during a fluorescence-assisted procedure. This involves the use of a filter element that allows X-rays to pass through undamped, enabling imaging of an area of ​​interest with high X-ray intensity and correspondingly high image quality.

[0006] Document US 8,754,388 B2 describes a radiation control system designed to minimize radiation exposure to a patient or system user. This system employs an attention monitoring system, which can be configured, for example, as a brain sensor or an eye sensor. An area of ​​interest is automatically determined based on this attention monitoring system.

[0007] Document DE 10 2016 219 708 A1 discloses a filter device for an X-ray machine, comprising a filter aperture element designed to be positioned in the X-ray machine in a beam path between a radiation source and a detector element. The filter aperture element has at least one hole. A holding device secures the filter aperture element to the X-ray machine, allowing the filter aperture element to be moved by the holding device in a vertical direction perpendicular to a principal plane of extension of the filter aperture element. This increases flexibility in selecting the size of a region of interest (ROI) of an X-ray image without altering any relevant image content.

[0008] Document DE 10 2015 221 638 A1 discloses a method for adjusting at least one radiation parameter of a radiation source in an X-ray device, wherein the radiation source is controlled according to the radiation parameter to provide X-rays. The radiation parameter is predetermined, as long as the filter element is not moved, depending on image data acquired by the radiation detector in a fluoroscopy area. At least one intermediate image is acquired, at least partially, within a movement interval during which the filter element is moved. After this intermediate image is acquired, the radiation parameter is determined independently of the image data of the intermediate image, or only image data of the intermediate image, acquired in an overlap area of ​​the radiation detector that lies within the fluoroscopy area for the entire acquisition interval, are considered when determining the radiation parameter.

[0009] Against this background, it is an object of the present invention to provide an improved concept for monitoring the treatment of an object using ionizing radiation, wherein the dose of ionizing radiation for the object and / or an observer can be reduced.

[0010] According to the invention, this problem is solved by a device for monitoring the treatment of an object, a training device for a model for image analysis during the monitoring of the treatment of an object, a method for training a model for image analysis during the monitoring of the treatment of an object, a method for monitoring the treatment of an object, a computer program, and a computer-readable storage medium according to the independent claims. Advantageous embodiments and further embodiments are the subject of the dependent claims.

[0011] The improved concept is based on the idea of ​​controlling the position of a priority area during an operating mode depending on the observer's eye movement, and, if a defined eye movement sequence of the observer is detected, controlling the position of the priority area depending on an image analysis or leaving it unchanged.

[0012] According to a first independent aspect of the improved concept, a device for monitoring the treatment of an object is described. The device comprises an optical unit configured to direct ionizing radiation, for example, X-rays, onto the object. The device also comprises a filter element, an imaging unit, a gaze detection system, and a control unit. The filter element, which is arranged in or on the optical unit, can be placed in or positioned within the beam path of the ionizing radiation to attenuate a portion of the ionizing radiation, particularly an intensity of that portion, in particular to define a priority area. The imaging unit is configured to detect portions of the ionizing radiation passing through the object in order to produce an image of a portion or area of ​​the object.The eye-tracking system is designed to detect the eye movement of an observer, in particular an observer of the image or a representation of the image, for example on an image display device.

[0013] The control unit, which is coupled to the eye-tracking system, for example, to obtain eye-tracking data generated by the eye-tracking system based on the detection of eye movements, is configured to control the position of the filter element during a first operating mode, particularly of the device and / or the control unit, depending on the eye movement, particularly on the eye-tracking data, specifically to control the position of the priority area. The control unit is configured to identify a predefined eye-tracking sequence, particularly of the observer, depending on the eye movement and to switch from the first operating mode to a second operating mode, particularly of the device or the control unit, when the eye-tracking sequence has been identified.The control unit is designed to control the position of the filter element during the second operating mode depending on an image analysis of the image, in particular a result of the image analysis, or to leave the position of the filter element unchanged during the second operating mode.

[0014] The object being treated might involve, for example, the use of a tool. In the case of medical treatment of a patient, the tool could be a catheter. The part of the object whose image is generated by the imaging unit includes, for example, the tool or a specific area of ​​the tool, such as the catheter tip. Controlling the position of the filter element during the first or second operating mode can, for example, be used to track the tool or the specific area of ​​the tool during the treatment, allowing the observer to obtain an accurate picture of the tool's position, particularly within the object, such as inside the patient's body.

[0015] The optical unit can, for example, receive ionizing radiation from a radiation source, which is optionally included in the device.

[0016] The filter element attenuates a portion of the ionizing radiation when positioned in the beam path, in order to minimize the dose of ionizing radiation to the observer and / or the object. Another portion of the ionizing radiation is not attenuated by the filter element, or only to a much lesser extent. This remaining portion of the radiation can therefore reach the object essentially unattenuated, allowing the image of that part of the object to be generated with the highest possible image quality, particularly with minimal noise. The priority region is specifically that area of ​​the object or on the object that receives the remaining portion of the ionizing radiation essentially unattenuated.

[0017] The tool, the distinguished area of ​​the tool, or an area of ​​the object within which the tool or the distinguished area of ​​the tool is located corresponds, for example, to a region of interest (ROI), since the distinguished area of ​​the tool, such as the catheter tip, is the part that the observer must keep in view as precisely as possible in order to be able to treat the object using the tool, in particular to be able to guide the tool as precisely as possible.Since the ROI can move during processing with the tool, for example due to a corresponding movement of the tool and / or the object, the priority area is controlled, for example, in such a way that it corresponds as closely as possible to the ROI or lies within the ROI, or the ROI lies within the priority area, so that the ROI can always be mapped with the highest possible quality.

[0018] The control unit is accordingly coupled to the eye-tracking system and, for example, to the imaging unit, and in particular to the filter element and / or the optical unit. In particular, the device, for example the optical unit, can include a drive system for moving the filter element in order to move the position of the filter element depending on control signals generated by the control unit for controlling the filter element.

[0019] The observer can be, in particular, a person who performs the treatment of the object using the tool, for example, a doctor. Specifically, the observer can be a user of the device. However, the observer can also be another person who observes the treatment of the object and does not necessarily perform it themselves. For example, the observer could be medical or technical personnel.

[0020] The eye-tracking system can, for example, include an eye tracker or be designed as an eye tracker itself. The eye tracker can be attached to or integrated into glasses or other devices worn by the observer, or it can be stationary within the observer's environment. The eye tracker can, for example, use active optical sensors and / or cameras to generate eye-tracking data based on the observer's eye movements.

[0021] The image includes, in particular, image data generated by the imaging unit based on the proportions of ionizing radiation passing through the object, in order to visually represent the object and / or the tool to the observer. The image data can be displayed as a visual image or representation on the image display unit. Furthermore, the image data can be analyzed using image analysis, for example, to identify the presumed region of interest (ROI) and thus control the filter element accordingly.

[0022] The control unit is specifically designed to repeatedly capture eye movements, compare the repeatedly captured eye movements with the predefined eye movement sequence, and identify the predefined eye movement sequence based on the result of the comparison.

[0023] The identification of the predefined eye movement sequence can therefore be understood, for example, as meaning that the control unit determines that the repeatedly recorded eye movement corresponds to the predefined eye movement sequence.

[0024] Identifying the predefined eye movement sequence based on the observer's eye movement can, for example, involve continuously or repeatedly capturing or monitoring the eye movement and comparing it with the predefined eye movement sequence, i.e., at least one predefined reference eye movement or a predefined sequence of predefined reference eye movements, which may be stored on a storage medium of the device. For this purpose, the control unit uses, for example, the eye-tracking data from the eye-tracking system.

[0025] If the control unit determines, by means of comparison, that a sequence of eye movements of the observer, in particular up to a specified tolerance, corresponds to the predefined eye movement sequence, it has accordingly identified the eye movement sequence and can switch itself or the device from the first to the second operating mode.

[0026] During the first operating mode, the eye-tracking system primarily detects the observer's gaze or eye movement. The observer's gaze is directed specifically at an image display unit, such as a monitor, on which the image is displayed.

[0027] During the first operating mode, the control unit can identify the observer's gaze direction based on the eye-tracking data. Accordingly, it can be assumed that the gaze direction is directed towards the ROI. Therefore, during the first operating mode, the position of the priority area, or the position of the filter element, is controlled such that the priority area coincides with a region around a point towards which the observer's gaze is directed. In particular, the priority area can be mapped by means of the image such that the observer's gaze direction is directed towards the image of the priority area, specifically that a center of the image of the priority area coincides with a point towards which the observer's gaze is directed, for example, a fixation point.By controlling the priority area according to the observer's viewing direction, it is possible during the first operating mode to ensure that the priority area corresponds as closely as possible to the ROI, which is thereby mapped with the highest possible quality.

[0028] The eye movement sequence corresponds, for example, to a series of eye movements by the observer that indicate or suggest that the observer's gaze is moved away from the ROI, particularly the image of the ROI, especially on the display unit, in a predefined manner. This can correspond, for example, to a situation in which the observer's gaze shifts from the display unit to another display, for instance, in the vicinity of the display unit, particularly briefly and / or temporarily. This other display unit may show, for example, further information relevant to the treatment of the object, such as, in the case of medical treatment, an ECG, a reference image, a pulse waveform, or an oxygen saturation range, or the like.It can also be a situation in which the observer deliberately and consciously focuses their gaze on the object and / or the tool, for example a position of the catheter or the catheter tip, in order to guide it in certain cases by directly looking at the tool instead of indirectly observing the tool via the image.

[0029] Once the eye movement pattern has been identified, the eye tracking system may still be able to record the observer's eye movements, depending on its detection range. In other situations, the eye tracking system may no longer be able to record the observer's eye movements, for example, because the observer's gaze has moved out of the system's detection range according to the eye movement pattern.

[0030] For example, during the second operating mode, image analysis of the image can be used to identify a specific area of ​​the tool, particularly the designated area, such as the catheter tip, within the image data that defines the image. The control unit can access and execute a model, such as a software algorithm or a computer program stored on the storage medium, to determine the position of the tool area within the image. This model can be a trained model.

[0031] Once the tool's area has been identified from the image data, the control unit can adjust the filter element, in particular, so that the priority area is mapped onto a region surrounding the designated area of ​​the tool. Accordingly, it is assumed that the designated area of ​​the tool corresponds to or lies within the ROI, especially at its center.

[0032] This allows for redundancy to be created for controlling the filter element based on the observer's eye movements, through image analysis and corresponding control of the image element.

[0033] As an alternative to control based on image analysis, the control unit can leave the position of the filter element unchanged during the second operating mode, i.e., not control it and, in particular, not remove it from the beam path.

[0034] Using a device based on the improved concept, the radiation dose for the object and the observer can be reduced, firstly, by providing the filter element to attenuate the portion of the ionizing radiation. The dose is further reduced, however, by allowing the filter element to be positioned within the beam path of the ionizing radiation in both the first and second operating modes, particularly regardless of whether the eye-tracking system can detect the observer's gaze or not, and regardless of whether the position of the filter element is controlled based on eye movement or not.

[0035] In particular, the filter element can remain in the beam path at all times during the first and second operating modes. According to the improved concept, removing the filter element from the beam path to maintain image quality of the ROI is no longer necessary.

[0036] In particular, the priority area remains correctly positioned and correlated with the ROI or presumed ROI even during the second operating mode, since control in the second operating mode can be performed based on the image analysis results.

[0037] Since the specified eye movement sequence has been identified, it can be assumed that extending the filter element is not necessary to continue ensuring good image quality of the ROI.

[0038] Since removing the filter element from the beam path would significantly increase the dose for the object and / or observer, this contributes to dose reduction.

[0039] Another advantage is that a second observer, who continues to monitor the ROI, especially while the observer is looking at another image display unit or the catheter, can still see the ROI correctly positioned with high image quality if the position of the filter element is controlled based on image analysis. Furthermore, the priority area remains correctly positioned without delay even when switching from the second operating mode to the first, as the filter element does not need to be moved back into the beam path.

[0040] Furthermore, artifacts that would be disturbing to the observer are avoided by extending and retracting the filter element into and out of the beam path.

[0041] Even if the eye-tracking system malfunctions, i.e., if the eye-tracking system loses the observer's gaze or incorrectly identifies the predefined eye movement sequence and switches to the second operating mode accordingly, but the observer, for example, still has their gaze directed at the ROI, it remains recognizable with good image quality due to the control of the filter element based on image analysis.

[0042] For example, the control unit can be configured to always control the position of the filter element in the second operating mode based on an image analysis, regardless of the reason for switching to the second operating mode. In particular, besides identifying the predefined eye-tracking sequence, there may be other criteria or situations that cause the control unit to switch to the second operating mode.

[0043] According to at least one embodiment of the device for monitoring the treatment of the object, the optical unit includes a collimator device to convert the ionizing radiation at least partially into a parallel beam, i.e., to parallelize it. In particular, the filter element, when located in the beam path, is arranged between the collimator device and the object.

[0044] This allows for a more specific and precise definition of the priority area.

[0045] According to at least one embodiment, the control unit is configured to control the position of the filter element depending on the eye movement whenever the control unit and / or the device are in the first operating mode.

[0046] According to at least one embodiment, the device includes an image display unit, wherein the imaging unit is coupled to the image display unit in order to transmit the image data or the image or other data representing the image to the image display unit in order to enable a visual representation of the image on the image display unit, which may, for example, include a monitor or display.

[0047] According to at least one embodiment, the control unit is configured to control the position of the filter element whenever the control unit and / or the device are in the second operating mode, depending on the image analysis of the image.

[0048] According to at least one embodiment, the control unit is configured to detect, based on eye movement, whether the observer's gaze direction lies within a predefined viewing direction range. The control unit is configured to maintain the first operating mode or switch from the second operating mode to the first operating mode if the gaze direction lies within the viewing direction range.

[0049] In the first operating mode, the filter element is again controlled primarily by the observer's eye movement.

[0050] The viewing direction range can, in particular, correspond to the view of the observer directed towards the image display unit.

[0051] This ensures that, if possible, the first operating mode is activated, particularly after the second operating mode has been activated and the first operating mode can be reactivated. Depending on the training status or other characteristics of the image analysis model, or other individual circumstances, identifying the ROI using the eye-tracking system may be more reliable or accurate than identifying the ROI using image analysis.

[0052] According to at least one embodiment, the control unit is configured to remove the filter element from the beam path of the ionizing radiation if the second operating mode has been activated for at least a predetermined maximum duration, in particular if the position of the filter element remains unchanged during the second operating mode.

[0053] The maximum duration can, for example, range from 1 s to 10 s, for example 3 s to 7 s or approximately 5 s.

[0054] This ensures, for example, that in the event of a faulty or unwanted activation of the second operating mode, the ROI is permanently displayed with suboptimal image quality.

[0055] According to at least one embodiment, the control unit is configured to switch from the second operating mode to the first operating mode if the second operating mode has been activated for a predetermined period of time, in particular continuously or without interruption.

[0056] According to at least one embodiment, by identifying the eye movement sequence, it is recognized that the observer's gaze direction or the observer's gaze lies within a predetermined further gaze direction range.

[0057] The extended viewing range can, for example, correspond to the observer's view of another image display unit. In particular, the image displayed on this additional display unit is not the same image generated by the imaging unit, but rather, for example, further information relating to the handling of the object.

[0058] In particular, if the eye movement sequence has been identified, the observer can consciously and, for example, briefly or temporarily direct their gaze to the further image display unit in order to acquire the further information and, for example, then direct their gaze back to the image display unit, especially the ROI.

[0059] In such a situation, it would be particularly undesirable to remove the filter element from the beam path in order to ensure the high image quality of the ROI, since in such a case the observer would not be looking at the ROI at all.

[0060] According to at least one embodiment, the device includes the additional image display unit.

[0061] According to at least one embodiment, the control unit is configured to switch from the first operating mode to the second operating mode if the eye-tracking system cannot detect the observer's eye movement.

[0062] For example, a malfunction or interference with the gaze tracking system, such as the presence of another object between the system and the observer, can cause the system to lose visual contact with the observer. Similarly, if the observer's gaze is no longer within the system's detection range, the system may fail to record their eye movements.

[0063] In particular, in such a case, according to the embodiment, the system switches to the second operating mode, although the eye movement sequence is not necessarily recognized.

[0064] Even in the event of an error or other reason that prevents the eye-tracking system from recording the observer's eye movement, the position of the filter element is controlled in the second operating mode based on image analysis, so that here too the filter element does not have to be removed from the beam path to ensure continued high image quality, especially of the ROI.

[0065] According to at least one embodiment, the device includes a computing unit, in particular coupled to the control unit and the imaging unit, which is configured to perform image analysis in order to determine the position of the ROI. The control unit is configured to control the position of the filter element during the second operating mode depending on the position of the ROI.

[0066] As described above, it can be assumed that during the first operating mode, the ROI coincides, or at least partially coincides, with an area of ​​the image displayed on the image display unit toward which the observer's gaze is directed. In the second operating mode, it can be assumed that the ROI corresponds to the designated area of ​​the tool.

[0067] In particular, image analysis is used to determine the position of the distinguished area of ​​the tool and define it as the target position for the priority area, in order to control the filter element by means of the control unit in such a way that the priority area corresponds to the distinguished area of ​​the tool, i.e. the presumed ROI.

[0068] According to at least one embodiment, the control unit is configured to identify a predefined further eye movement sequence depending on the eye movement. The control unit is configured to determine at least one state variable of the device and / or a change in that at least one state variable. Depending on the at least one state variable or a change in that at least one state variable, the control unit is configured to switch from the first operating mode to a second operating mode once the further eye movement sequence has been identified.

[0069] As explained above, the eye movement sequence can, for example, correspond to a situation in which the observer consciously and briefly averts their gaze from the image display unit, and this can be determined with sufficient reliability based solely on the identified eye movement sequence.

[0070] If the subsequent eye movement sequence is identified, the situation may be the same. However, it may also be the case that, in this instance, it cannot be assumed with sufficient probability that the observer is only briefly and temporarily averting their gaze from the image display unit. Therefore, the state variable of the device can additionally be used to decide whether to switch to the second operating mode.

[0071] The latter can correspond, for example, to a situation where an observer looks at a control element for changing the state variable, and then looks back at the image display unit. According to such embodiments, the unnecessary extension of the filter element can also be avoided here.

[0072] According to at least one embodiment, the at least one state variable contains a relative position and / or a relative orientation of a component of the device with respect to another component of the device.

[0073] According to at least one embodiment, the at least one state variable includes a relative position and / or a relative orientation of at least one component of the device with respect to the object or the tool.

[0074] The components can, for example, correspond to the optical unit or the imaging unit.

[0075] According to at least one embodiment, the at least one state variable contains an operating parameter for operating the device, for example the radiation source, the optical unit, the filter element, the imaging unit, the eye-tracking system or the control unit.

[0076] According to at least one embodiment, the at least one state variable includes a distance between the radiation source and the imaging unit, in particular a source-to-image distance (SID).

[0077] According to at least one embodiment, the at least one state variable includes an angulation of the device, wherein the device is in particular designed as a C-arm device.

[0078] According to at least one embodiment, the device includes a camera system configured to generate camera data that depicts the observer. The control unit is configured to switch from the first operating mode to a second operating mode, depending on the camera data and, for example, depending on at least one state variable or a change in at least one state variable, once the further eye movement sequence has been identified.

[0079] The control unit or processing unit is, for example, configured to determine the position or pose of the observer or a part of the observer's body, such as the observer's head, relative to the device, based on the camera data. Depending on this information, the control unit can then switch to the second operating mode.

[0080] According to another independent aspect of the improved concept, a training device for an image analysis model is specified for monitoring the treatment of an object. The training device comprises an optical unit configured to direct ionizing radiation onto the object. The training device includes a filter element that can be positioned in the beam path of the ionizing radiation to attenuate a portion of the radiation and thereby define a priority area, as well as an imaging unit configured to detect portions of the ionizing radiation passing through the object in order to generate an image of a portion of the object.The training device includes an eye-tracking system configured to record an observer's eye movements, and a control unit configured to determine a target position for the priority area based on these eye movements. The training device also includes a computing unit configured to determine a parameter set for the model based on a correlation between the image and the target position.

[0081] The model, in particular the specific set of parameters, can be used in particular by a device for monitoring the treatment of an object, in particular by its control unit and / or computing unit, to control the position of the filter element based on the image analysis, in particular to perform the image analysis.

[0082] Further embodiments of the training device follow directly from the various configurations of the device for monitoring the treatment of the object and vice versa.

[0083] The training device may, in particular, be physically the same device as the device for monitoring the treatment of an object.

[0084] According to another independent aspect of the improved concept, a method for training an image analysis model during the monitoring of an object's treatment is described. In this method, ionizing radiation is directed at the object. A portion of the ionizing radiation is attenuated, particularly by means of a filter element that can be inserted into the beam path of the ionizing radiation, to define a priority area. An image of a portion of the object is generated based on the fractions of ionizing radiation passing through the object, particularly by means of an imaging unit. The eye movement of an observer, particularly an observer of the image on an image display unit, is detected, particularly by means of an eye-tracking system, and a target position for the priority area is determined based on the eye movement, particularly by means of a control unit.A parameter set for the model for image analysis, in particular for the analysis of the image, is determined by means of a correlation or comparison of the image, in particular the image data, with the target position, in particular by means of a computing unit or control unit.

[0085] The filter element can be controlled, in particular by means of the control unit, according to the target position, so that the priority area is moved to the target position.

[0086] During runtime, i.e., while a procedure for monitoring the processing of the object is being carried out, the model serves, in particular, to determine the target position based on the analysis of the image and to specify this as the target position for the priority area for controlling the filter element. During the training procedure, however, the target position is specified independently of the model, i.e., from an external source, by deriving the target position from the observer's eye movements. This is therefore primarily a form of supervised learning, specifically supervised machine learning, with which the model is trained in a process for training the model.

[0087] The specific set of parameters can, for example, be seen as the result of the procedure for training the model.

[0088] The parameter set can then be used at runtime for productive or actual image analysis of the image and accordingly to determine the target position for the priority area.

[0089] The described steps of the model training process can be repeated or performed iteratively, particularly until the parameter set meets predefined quality criteria, meaning the model can determine the target position with a specified accuracy based on the current parameter set. In each iteration, the representation of the object can change, for example, by moving a tool used to manipulate the object, which may be located within the object itself. This allows for the generation of numerous training datasets, which can then be used to train the model iteratively.

[0090] In particular, the model serves to identify the position of the distinguished area of ​​the tool from a given image. It is assumed that the observer focuses their gaze on the distinguished area of ​​the tool in the image on the display unit.

[0091] One advantage of this method for training the model is that the eye-tracking system can generate a large amount of high-quality training data. This enables effective model training.

[0092] The model training process can be performed online, meaning that the training data is acquired during runtime, i.e., while the actual processing of an object is being monitored. During runtime, the position of the filter element or the priority area can be controlled, particularly based on the target position determined by eye movement. The training itself—the correlation or comparison of the image or image data with the respective target position—can also be performed online, either during runtime or later after the monitoring is complete.

[0093] Online training inherently allows for the detailed consideration of system-, application- and user-specific, i.e. observer-specific, influences, leading to a high quality of training.

[0094] The computing unit can, for example, consist of software elements and / or hardware elements, such as a microprocessor. The computing unit can be, for example, a computer or part of a computer system.

[0095] According to at least one embodiment of the method for training the model, a further target position is determined based on the model and the specified parameter set, particularly using the computing unit. This further target position is compared with the target position, particularly using the computing unit, and a quality indicator for the specified parameter set is determined based on the result of the comparison of the target position with the further target position, particularly using the computing unit.

[0096] In such an embodiment, a productive application of the model is tested using the specified parameter set by determining the further target position based on the image using image analysis.

[0097] The quality indicator can, in particular, quantify a deviation of the target position from the further target position and thus express how well the model can determine the target position based on image analysis using the currently used parameter set.

[0098] In particular, the steps for training the model are repeated iteratively until the quality indicator corresponds to a predefined quality target value or lies within a predefined range for the quality indicator.

[0099] Further embodiments of the method for training the model follow directly from the various configurations of the device for monitoring the treatment of the object and the training device, and vice versa. In particular, the training device according to the improved concept is configured or programmed to perform a method for training according to the improved concept, or the training device performs a method for training according to the improved concept.

[0100] According to at least one embodiment of the device for monitoring the treatment of an object, the control unit is configured to control the position of the priority area during the second operating mode using a trained model. This model was trained based on control data generated during the first operating mode to control the position of the priority area in relation to eye movements. In particular, the model was trained using a method for training an image analysis model during the monitoring of an object's treatment according to the improved concept.

[0101] According to a further independent aspect of the improved concept, a method for monitoring the treatment of an object is described, wherein ionizing radiation, in particular by means of an optical unit, is directed at the object and a portion of the ionizing radiation is attenuated, in particular by means of a filter element, to define a priority area. An image of a portion of the object is generated based on the fractions of the ionizing radiation passing through the object, in particular by means of an imaging unit, and the eye movement of an observer, in particular of the image, is detected, in particular by means of an eye-tracking system.

[0102] During a first operating mode, particularly monitoring, the position of the priority area is controlled based on eye movement, specifically by means of a control unit. Depending on the eye movement, a predefined eye movement sequence is identified, specifically by means of the control unit, and the system switches from the first operating mode to a second operating mode, specifically by means of the control unit, once the eye movement sequence has been identified. During the second operating mode, the position of the priority area is controlled based on an image analysis of the image, specifically by means of the control unit.

[0103] Controlling the priority area is equivalent to controlling the filter element.

[0104] According to at least one embodiment of the method for monitoring the treatment of the object, a target position for the priority area is determined, particularly by means of the control unit, depending on the eye movement, especially during the first operating mode, in order to control the position of the priority area as a function of the eye movement. The position of the filter element for attenuating the portion of the radiation is controlled such that the position of the priority area at least approximately coincides with the target position.

[0105] According to at least one embodiment of the monitoring method, the position of the priority area is controlled during the second operating mode using a trained model. The model is trained based on control data, in particular by means of a computing unit, wherein the control data is generated during the first operating mode to control the position of the priority area based on eye movement, in particular by means of the control unit.

[0106] Training the model based on control data can be done, in particular, based on the target position for the priority area. Specifically, training can be performed by correlating the target position with image data of the model.

[0107] The training can, in particular, include a comparison of the target position with a training target position, whereby the training target position is determined by the computing unit depending on the image data.

[0108] According to at least one embodiment of the method for monitoring the treatment of the object, the model is trained using a method for training a model for image analysis according to the improved concept.

[0109] According to at least one embodiment of the monitoring method, the position of the ROI is determined or estimated by image analysis. The position of the ROI is defined as a further target position for the priority area in order to control the position of the priority area during the second operating mode.

[0110] In particular, to determine the further target position, the position of the tool's area within the image data is identified and determined. The position of the filter element is then controlled so that the position of the priority area corresponds to the further target position.

[0111] According to at least one embodiment of the monitoring method, the eye movement determines whether the observer's gaze direction lies within a predetermined viewing direction range. If the gaze direction is within the viewing direction range, the first operating mode is maintained, or the system switches from the second operating mode to the first operating mode.

[0112] According to at least one embodiment of the monitoring method, identifying the eye movement sequence reveals whether, or that, the observer's gaze direction lies within a predetermined further gaze direction range.

[0113] The various configurations of the device directly lead to further configurations of the method for monitoring the treatment of the object according to the improved concept, and vice versa. In particular, a device according to the improved concept is, for example, configured or programmed to carry out a method for monitoring according to the improved concept, or the device carries out such a method.

[0114] According to another independent aspect of the improved concept, a computer program with instructions is specified, wherein, when the computer program is executed by a device according to the improved concept, in particular by a computing unit of the device, the instructions cause the device to perform a method for training the model for image analysis according to the improved concept and / or a method for monitoring the treatment of the object.

[0115] According to another independent aspect of the improved concept, a computer-readable storage medium is specified on which a computer program is stored according to the improved concept.

[0116] The invention is explained in more detail below with reference to specific embodiments and associated schematic drawings. In the figures, identical or functionally equivalent elements are designated with the same reference numerals. The description of identical or functionally equivalent elements is repeated in different figures, if necessary, but not necessarily.

[0117] The figures show Fig. 1 a schematic representation of an exemplary embodiment of a device according to the improved concept; Fig. 2 a schematic representation of image display units for use with a further exemplary embodiment of a device according to the improved concept; Fig. 3 a flowchart of an exemplary embodiment of a method according to the improved concept; and Fig. 4 a flowchart of another exemplary embodiment of a method according to the improved concept.

[0118] In Fig. Figure 1 shows an exemplary embodiment of a device 1 for monitoring the treatment of an object 4 according to the improved concept. In this non-limiting example, the object 4 is a patient 4 on whom medical treatment is performed using a tool 27. The tool 27 can, in particular, be a catheter 27.

[0119] The device 1 includes, for example, a radiation source 9, which is configured, for example, to generate ionizing radiation 3, such as X-rays 3. The device 1 also includes an optical unit 2, which includes, for example, a collimator to at least partially align the X-rays from the radiation source 9. The optical unit 2 may also include, for example, a lens arrangement and / or other optical elements, for example, for beam shaping.

[0120] The optical unit 2 allows the X-ray radiation 3 to be directed at the patient 4, in particular at an area of ​​the patient 4 where the catheter 27 is located inside the patient 4's body. The device 1 also includes an imaging unit 6, which in particular includes an X-ray-sensitive element that can detect X-ray radiation 3 passing through the patient 4's body and, based on the detected X-ray radiation 3, can generate an image of an area of ​​the patient 4.

[0121] The device 1 includes, for example, an image display unit 7, such as a monitor or a display, on which a representation 8 of the image can be visually displayed. For this purpose, image data containing the image is transferred from the imaging unit 6 to the image display unit 7.

[0122] An observer 25, for example a doctor who guides the catheter 27 to perform the treatment of patient 4, can look at the representation 8 of the image on the image display unit 7, in particular to be able to guide the catheter 27 using X-ray guidance.

[0123] The device also includes an eye-tracking system, which, for example, comprises an eye tracker 26. The eye tracker 26 can, for example, include one or more active optical sensors, in particular infrared sensors, and / or a camera system to detect and determine the eye movements of the observer 25 and, in particular, the gaze direction of the observer 25. The device 1 has a control unit 10, which is coupled, in particular, to the eye tracker 26 and the imaging unit 6. Furthermore, the device 1 has a computing unit 11, which is coupled to the control unit 10 and / or the eye tracker 26.

[0124] Furthermore, the device 1 or the optical unit 2 comprises a filter element 5, for example, a ROI filter, which can be arranged in the beam path of the X-ray radiation 3, in particular between the radiation source 9 and the patient 4. The filter element 5 attenuates the intensity of the X-ray radiation 3 in a sub-region of the beam path in order to keep the radiation dose for the patient 4 and for the observer 25 as low as possible. In a central region of the filter element 5, the filter element has a transparent area, for example, an opening, through which the X-ray radiation 3 can pass unattenuated or substantially unattenuated to strike a priority area 13 on the patient 4. The position of the filter element 5, and thus the position of the priority area 13, can be controlled by the control unit 10.

[0125] The image quality generated by the imaging unit 6 is, for example, better the higher the intensity of the respective X-ray radiation 3 in the corresponding area. In particular, the noise component in the image data decreases as the local X-ray intensity increases. Accordingly, particularly high image quality can be achieved in the priority area 13, while lower image quality is attainable in the areas where the X-ray radiation 3 is attenuated by the filter element 5.

[0126] Accordingly, it is advantageous to control the priority area 13 by means of the control unit 10 such that the priority area 13 coincides as closely as possible with a region of interest (ROI) 12. The ROI 12 corresponds in particular to an area of ​​the catheter 27 that is of particular importance for carrying out the treatment of patient 4. This can be, in particular, the tip of the catheter 27. The observer 25, or another observer who performs the treatment and guides the catheter 27, observes, for example, the catheter tip in the image on the image display unit 7 in order to ensure precise guidance of the catheter 27.

[0127] Therefore, the ROI 12 typically corresponds to an area on the patient 4 that corresponds to a viewing direction or fixation point of the observer 25 on the image 8. The observer 25 will usually or mostly look at the catheter tip in the image 8 during the treatment.

[0128] During operation of the device 1, the eye tracker 26 can, for example, capture the eye movements of the observer 25 to identify a position of the ROI 12 in the representation 8 and thus also on the body of the patient 4, for example, independently or in combination with the control unit 10 and / or the processing unit 11. For example, the eye tracker 26 can transmit eye-tracking data to the control unit, and the control unit 10 can determine the position of the ROI 12 based on the eye-tracking data. The position of the ROI 12 determined in this way can serve as the target position for the priority area 13 in a first operating mode of the control unit 10 or the device 1. Accordingly, the control unit 10 can control the filter element 5 and thus the priority area 13 in such a way that the priority area 13 corresponds to or follows the ROI. This ensures that the ROI 12 is always depicted with the highest possible image quality.At the same time, the total radiation dose for the patient 4 and the observer 25 is reduced by attenuating the X-ray radiation 3, accepting a lower image quality outside the ROI 12.

[0129] In Fig. Figure 2 shows the image display unit 7 with the image 8 and the ROI 12 again. Two further image display units 14 and 15 are also shown, which may display additional information relevant to the treatment of patient 4. For example, an ECG of patient 4, a reference image, or similar information may be displayed on these additional image display units 14 and 15.

[0130] During the treatment of patient 4, it may happen that the observer 25 turns his gaze away from the representation 8 of the image and from the image display unit 7, for example to look at one of the other image display units 14, 15 or directly at patient 4 or the catheter 27.

[0131] The Eye Tracker 26 is specifically designed to identify one or more eye movement sequences in the observer's 25 eye movements that correspond to one of the described situations in which the observer 25 directs their gaze from the image display unit 7 to one of the other image display units 14, 15, or to the catheter 27. For this purpose, the Eye Tracker 26, for example in combination with the processing unit 11 and / or the control unit 10, can employ machine learning methods, such as artificial neural networks or K-means classifications, or conventional methods for eye trajectory tracking.

[0132] For example, observer 25 can consciously look away from the image display unit 7 and then, shortly afterwards, return their gaze to the previous position in the display 8. The eye tracker 26 can, for example, identify a relatively rapid eye movement of observer 25 in a predefined direction, which corresponds, for example, to a direction from the image display unit 7 to one of the other image display units 14, 15 or to the catheter 27, in order to identify the eye movement patterns.

[0133] Optionally, the Eye Tracker 26 can also use predefined information, such as the identity of the observer 25, the type of treatment, the availability of additional image display units 14, 15 and how they are arranged, the monitor layout of the image display units 7, 14, 15, whether an ECG and / or a reference image is used, or similar information, to ensure more reliable detection of eye movements.

[0134] If one of the predefined eye movement sequences has been identified by the eye tracker 26, the control unit 10, or rather the device 1, can switch to a second operating mode. In this second operating mode, for example, the filter element 5 and its position are no longer controlled based on the eye tracking data from the eye tracker 26 and the target position determined therefrom. Instead, the processing unit 11 can be configured to perform an image analysis based on the image data generated by the imaging unit 6 in order to identify the position of the catheter tip or another distinguished area of ​​the catheter 27 and to specify this as the target position for the priority area 13.

[0135] The control unit 10 is configured to control the position of the filter element 5, and thus the priority area 13, during the second operating mode, depending on the target position for the priority area 13 determined by image analysis. Accordingly, even if the observer 25 is not currently looking at the representation 8 of the image and / or if the observer 25's gaze cannot be detected by the eye tracker 26 for this or other reasons, the position of the priority area 13 can be controlled according to the presumed or probable ROI.

[0136] In particular, in such a case, i.e., in the second operating mode, the filter element 5 is not completely removed from the path of the X-ray radiation 3. Accordingly, the radiation dose for the patient 4 and the observer 25 is reduced.

[0137] Another advantage is that, as a consequence, the filter element 5 does not need to be moved back into the X-ray beam path 3 when the observer 25 looks again at the image 8. This prevents disturbances such as image artifacts when the filter element 5 is retracted.

[0138] The image analysis of the image data can be performed by the computing unit 11, for example, using a model, in particular a trained model, especially by employing machine learning methods, such as supervised machine learning. Specifically, a Bayesian classifier, a naive Bayesian classifier, a nearest-neighbor classifier, discriminant analysis, and / or an artificial neural network for supervised learning can be used.

[0139] In Fig. Figure 3 is an exemplary embodiment of a method for monitoring the treatment of an object, for example by means of a device made of Fig. 1, as well as an exemplary embodiment of a method for training a model of image analysis according to the improved concept, in particular in the form of a flowchart.

[0140] Regarding the individual steps, reference is also made to the explanations concerning Fig. 1 and Fig. 2. In step 16 of the procedure, the ionizing radiation 3 is directed at the patient 4 by means of the radiation source 9 and the optical unit 2 as well as the filter element 5, and the portion of the ionizing radiation 3 that does not pass through the opening of the filter element 5 is attenuated accordingly. In step 17, based on the portions of the ionizing radiation 3 that pass through the patient 4, the image of the area of ​​the patient 4 on which the radiation 3 strikes is generated by means of the imaging unit 6.

[0141] In step 18, the eye movement of the observer 25 is recorded, in particular by means of the eye tracker 26, and the position of the priority area 13 is determined as with respect to Fig. 1 described, controlled depending on eye movements.

[0142] In step 19, one of the predefined eye movement sequences is identified, indicating that the observer's gaze 25 is directed away from the representation 8 of the image. Now, in step 22, the second operating mode is activated using the control unit 10.

[0143] In step 23, the processing unit 11 performs an image analysis of the image data. Using this image analysis, a position of a specific area of ​​the catheter 27, particularly the catheter tip, is identified and set as the target position for the priority area 13. In step 24, the control unit 10 adjusts the position of the filter element 5 so that the position of the priority area 13 corresponds to the target position, thus enabling high-quality imaging of the catheter tip.

[0144] Image analysis can be performed, for example, using a predefined set of parameters. This parameter set may originate from a training procedure based on the improved concept. In particular, in step 20, the computing unit can compare the position of the ROI 12, determined by the eye tracker 26 in the first operating mode in step 18, with the position of the catheter tip determined from the image analysis. This allows the model parameters, i.e., the parameter set, for localizing the catheter tip based on the image data to be adjusted or updated.

[0145] Target positions determined using the Eye Tracker 26 during the initial operating mode are used as training data for optimizing or training the model. A quality score for the current parameter set of the model is determined by comparing the position of the catheter tip, as calculated by the model, with the position of the ROI from the eye-tracking data. This quality score quantifies the deviation between the two positions. The greater the deviation, the lower the quality of the given parameter set.

[0146] In step 21, it is checked whether the quality indicator corresponds to a predefined target value for quality. The steps of determining the position of the catheter tip using the model, comparing it with the ROI position from the eye-tracking data, and adjusting the parameter set are repeated iteratively, for example, until it is determined in step 21 that the quality indicator corresponds to a predefined target value for quality. If this is the case, the parameter set determined in this way can be passed to step 23 of the procedure at runtime, and the model can be used at runtime during the second operating mode to serve as the basis for controlling filter element 5.

[0147] The training procedure according to the improved concept can therefore be advantageously carried out at least partially online, i.e. during the running time of the procedure, to monitor the treatment of patient 4 and thus inherently take into account the specific features of the concrete system design as well as user-specific details in order to generate the parameter set.

[0148] In Fig. Figure 4 is another exemplary embodiment of a method for monitoring the treatment of an object, for example by means of a device made of Fig. 1, shown. Regarding the individual steps, reference is also made to the explanations concerning Fig. 1 and Fig. 2 referred.

[0149] In step M1, the device 1 or the control unit 10 is operated in the first operating mode, as described above. Fig. 1 to Fig. 3 described. In particular, the eye movement of the observer 25 is recorded by means of the eye tracker 26 and the position of the priority area 13 is determined as described in section 3. Fig. 1 described, controlled depending on eye movements.

[0150] In step S1, it is checked whether one of the predefined eye movement sequences or another predefined eye movement sequence can be identified.

[0151] The further eye movement sequence differs, for example, from the eye movement sequences in that, if the further eye movement sequence is present, it cannot necessarily be concluded that the observer 25 only briefly averted his gaze from the image display unit 7 in order to return his gaze shortly afterwards to the previous position in the representation 8.

[0152] If neither one of the predefined eye movement sequences nor the subsequent eye movement sequence is identified, step S2 checks whether the gaze of observer 25 can still be detected by the eye tracker 26. If so, the first operating mode is maintained. Otherwise, step M2 switches to the second operating mode, as described above. Fig. 1 to Fig. 3 executed.

[0153] If one of the eye movement sequences is identified in step S1, the system also switches to the second operating mode in step M2.

[0154] If the further eye movement sequence is identified in step S1, then in step S3 it is determined whether a predefined state variable of device 1, for example a SID or an angulation, changes or has changed. If this is the case, then in step M2 the system switches to the second operating mode.

[0155] Otherwise, proceed with step S2 as described above.

[0156] If the further eye movement sequence has been identified and the change in the state variable is also present, it can be assumed, for example, that the observer 25 only briefly averted his gaze from the image display unit 7.

[0157] The improved design eliminates the need to remove the filter element from the path of the ionizing radiation in situations where, for example, the observer intentionally looks away from the display unit or where the eye tracker can no longer detect the observer's eye movements for other reasons. This ensures sufficiently high image quality in the priority area, thus reducing the radiation dose for both the object and the observer.

[0158] For example, the filter element can remain permanently in the beam path during the second operating mode. Alternatively, after a certain time following activation of the second operating mode, the filter element can be completely removed from the beam path if it is not detected that the observer is looking back at the image display unit.

[0159] Both when using an eye tracker and during image analysis, situations can arise where the priority area cannot be positioned correctly. According to the improved concept, these two control approaches complement each other, thus preventing the filter element from being moved out of the beam path.

[0160] This increases the overall effectiveness of the filter element (dose reduction). Furthermore, it avoids artifacts that would be distracting to the observer, caused by the filter element moving in and out of the beam path. Using a training procedure based on the improved concept, the image analysis model can be specifically tailored to the concrete situation in a procedure for monitoring treatment according to the improved concept. The eye tracker information, i.e., the gaze-tracking data, provides excellent training data for the model. In particular, the model's quality can be improved during each runtime.

[0161] For training, algorithms based on artificial intelligence or machine learning can be used to optimize the control of the filter element based on image analysis. This training can be performed online, particularly using data provided by the eye tracker. A further advantage is that the eye tracker's tracking of eye movements is generally very accurate, providing correspondingly good input data for training.

Claims

[1] Device for monitoring the treatment of an object (4) comprising the device (1) - an optical unit (2) designed to direct ionizing radiation (3) onto the object (4); - a filter element (5) which can be arranged in a beam path of the ionizing radiation (3) in order to attenuate a portion of the ionizing radiation (3); - an imaging unit (6) configured to detect components of the ionizing radiation (3) passing through the object (4) in order to produce an image (8) of part of the object (4); - an eye-tracking system (26) designed to detect the eye movements of an observer (25); and a control unit (10) designed to control the position of the filter element (5) depending on the eye movement during a first operating mode; characterized by , that the control unit (10) is set up to - to identify a predefined eye movement sequence depending on the eye movement; - to switch from the first operating mode to a second operating mode once the eye-tracking sequence has been identified; and - during the second operating mode, to control the position of the filter element (5) depending on an image analysis of the image (8). [2] Device according to claim 1, characterized by , that the control unit (10) is configured to, - to recognize, depending on the eye movement, whether the observer's gaze direction (25) lies within a given gaze direction range; and - to maintain the first operating mode or to switch from the second operating mode to the first operating mode if the viewing direction is within the viewing direction range. [3] Device according to one of claims 1 or 2 characterized by , that by identifying the eye movement sequence it is recognized that the gaze direction of the observer (25) lies within a given further gaze direction range. [4] Device according to any one of claims 1 to 3, characterized by , that the control unit (10) is configured to switch from the first operating mode to the second operating mode when the eye-tracking system (26) cannot detect the eye movement of the observer (25). [5] Device according to any one of claims 1 to 4, characterized by , that the device (1) includes a computing unit (11) configured to perform image analysis in order to determine the position of an area of ​​interest (12); and the control unit (10) configured to control the position of the filter element (5) during the second operating mode depending on the position of the area of ​​interest (12). [6] Device according to any one of claims 1 to 5, characterized by , that the control unit (10) is configured to, - to repeatedly capture eye movements and compare the repeatedly captured eye movements with the predefined eye movement sequence; and - to identify the predefined eye movement sequence based on a result of the comparison. [7] Device according to any one of claims 1 to 6, characterized by , that the control unit (10) is configured to, - to identify a predefined further eye movement sequence depending on the eye movement; - to determine at least one state variable of the device (1) or a change in the at least one state variable; - to switch from the first operating mode to a second operating mode depending on at least one state variable or on the change of at least one state variable, once the further eye movement sequence has been identified. [8] Training device for a model for image analysis during monitoring of the treatment of an object (4), characterized by , that the training device (1) an optical unit (2) configured to direct ionizing radiation (3) onto the object (4); a filter element (5) which can be arranged in a beam path of the ionizing radiation (3) to attenuate a portion of the ionizing radiation (3) and thereby define a priority area (13); an imaging unit (6) configured to detect components of the ionizing radiation (3) passing through the object (4) in order to produce an image (8) of part of the object (4); a gaze detection system (26) designed to detect the gaze movement of an observer (25); a control unit (10) configured to determine a target position for the priority area (13) depending on eye movement; and a computing unit, set up to determine a parameter set for the model based on a correlation of the image (8) with the target position. [9] Method for training a model for image analysis during monitoring of the treatment of an object (4), characterized by , that ionizing radiation (3) is directed at the object (4), in particular by means of an optical unit (2); a portion of the ionizing radiation (3) is attenuated, in particular by means of a filter element (5) to define a priority area (13); an image (8) of part of the object (4) based on the components of the ionizing radiation (3) passing through the object (4) is produced by means of an imaging unit (6); an observer’s eye movement (25) is recorded by means of an eye-tracking system (26); a target position for the priority area (13) is determined, in particular by means of a control unit (10), depending on the eye movement; and a parameter set for the model is determined by means of a computing unit (11) based on a correlation of the image (8) with the target position. [10] Method according to claim 9, characterized by , that by means of the computing unit (11): based on the model using the specified parameter set based on the image (8), a further target position for the priority area (13) is determined; the further target position is compared with the target position; and A quality indicator is determined for the specific parameter set based on a result of the comparison. [11] Device according to any one of claims 1 to 7, characterized by , that the control unit (10) is configured to control the position of the priority area (13) during the second operating mode using a trained model; wherein the model was trained depending on control data generated during the first operating mode to control the position of the priority area (13) depending on the eye movement, in particular by means of a method according to one of claims 9 or 10 trained model. [12] Method for monitoring the treatment of an object (4) wherein ionizing radiation (3) is directed at the object (4), in particular by means of an optical unit (2); a portion of the ionizing radiation (3) is attenuated, in particular by means of a filter element (5) to define a priority area (13); an image (8) of part of the object (4) based on the components of the ionizing radiation (3) passing through the object (4) is produced by means of an imaging unit (6); an observer's eye movement (25) is recorded by means of an eye-tracking system (26); and During a first operating mode, a position of the priority area (13) is controlled depending on the eye movement by means of a control unit (10); characterized by , that Depending on the eye movement, a predefined eye movement sequence is identified by means of the control unit (10); from the first operating mode to a second operating mode by means of the control unit (10) when the eye movement sequence has been identified; and During the second operating mode, the position of the priority area (13) is controlled depending on an image analysis of the image (8) by means of the control unit (10). [13] Method according to claim 12, characterized by , that to control the position of the priority area (13) depending on the eye movement, a target position for the priority area (13) is determined depending on the eye movement; and a position of a filter element (5) to attenuate the part of the radiation (3) is controlled such that the position of the priority area (13) corresponds to the target position. [14] Method according to claim 13, characterized by , that the position of the priority area (13) during the second operating mode is controlled by a trained model; the model is trained depending on control data generated during the first operating mode to control the position of the priority area (13) depending on eye movement. [15] Method according to claim 14, characterized by that the model is trained using a method according to one of claims 9 or 10. [16] Method according to any one of claims 12 to 15, characterized by, that by means of image analysis using a computing unit (11) a position of an area of ​​interest (12) is determined; the position of the area of ​​interest (12) is defined as a further target position for the priority area (14) using the computing unit (11) in order to control the position of the priority area during the second operating mode. [17] Computer program product with commands, which, when implemented by a device (1) according to one of claims 1 to 7 or 11, cause the device (1) to carry out a method according to one of claims 12 to 16; or which, when performed by a training device (1) according to claim 8, cause the training device (1) to perform a method according to one of claims 9 or 10.

Images