Monitoring an fMRI scan

Abstract

The invention relates to a method 10 for monitoring the performance of a predefined test task performed by a subject during an fMRI examination. The method comprises the steps of acquiring (11) a time series of fMRI images and, during the acquisition (12), providing the subject with a sensory stimulus according to the predefined test task. The method comprises the steps of determining (13) after the sensory stimulus an activity measure indicative of brain activity in at least one predefined region of interest in the brain of the subject imaged by the time series of images and generating (14) at least one signal when the activity measure indicates an insufficient performance of the predefined test task and / or an insufficient compliance with the predefined test task. The signal is generated during the acquisition of the time series and / or before removing the subject from the magnetic resonance imaging system. Further aspects of the invention relate to a computer program product, a computer system and an MRI system.

Description

[Technical field]

[0001] The present invention relates to the field of functional magnetic resonance imaging, and more particularly, the present invention relates to devices and methods for monitoring the performance of a given task by a subject, such as the subject's compliance with the intended task, during a functional imaging procedure. [Background technology]

[0002] Magnetic resonance imaging (MRI) is an amazing tool for non-invasively imaging internal body structures in the human or animal body. In MRI, a strong magnetic field is used to align the nuclear spins of atoms in the body, after which the return signals emitted due to the relaxation of these spins can be examined in a position-dependent manner, for example, using appropriate magnetic field gradients and purposefully designed radio frequency excitation pulses. Many techniques exist in the art for preparing and / or processing the observed signals, such as to create images sensitive to various material properties in the body.

[0003] One type of these techniques involves creating a detectable image contrast indicative of blood oxygen, which allows detection of hemodynamic responses in active regions of the brain, and thus determining which regions of the brain are active. These MRI techniques, which rely on, for example, the blood oxygen level-dependent (BOLD) effect, cerebral blood flow, or similar approaches, are commonly referred to as functional magnetic resonance imaging (fMRI).

[0004] However, one of the difficulties in fMRI relates to the poor image contrast generated by this BOLD effect (or its alternatives). That is, for example, signal components not related to the BOLD effect are typically much more obvious (by several orders of magnitude stronger) than the actual signal components of interest. This is typically overcome by applying an appropriate statistical analysis (or alternative data modeling and / or mining approaches) of an extended time series of MRI data. During the collection of this time series, a brain function is triggered (typically repeatedly) in order to identify changes in the brain image in relation to other images collected in the absence of this function trigger. Thus, conventional approaches entail an analysis in which statistically significant deviations between activation and control states are detected.

[0005] Functional Magnetic Resonance Imaging (fMRI) is a versatile approach to investigate the functioning of the human brain, for example for research purposes as well as for diagnosing brain-related neurological, psychiatric or other diseases and disorders. Specific sensory, cognitive, emotional, behavioral, motivational, memory and motor functions of the brain can be visualized and studied in detail using fMRI. Thus, the involvement of specific areas of the brain in the neocortex and limbic system in behavioral (but not necessarily limited to direct and / or active) tasks can be studied in detail.

[0006] To visualize these brain functions using fMRI, it is necessary to induce a contrast difference between an image acquired while a certain function of the brain is active, for example by having the imaged subject perform a certain task, and a reference image acquired when this certain function is absent. It will be clear that the reference may relate to a condition in which the brain activity of interest is not literally absent, but is activated to a lesser extent, or the reference image is acquired under more specific control conditions, for example a similar task different from the one of interest, possibly performed in a very specific manner. Many more complex or slightly different approaches are conceivable, for example (to name only a few known variants) instead of directly contrasting a single task of interest against a control condition, contrasting more than two conditions against each other, or taking into account task parameters, such as variable intensities or magnitudes associated with each repetition of the task performed. The task or specific function to be considered can be selected from a very wide range of possibilities, including, for example, "tasks" that involve the mere observation of sensory stimuli, the controlled movement of a muscle(s) or simply the intention to move, specific mental calculations or thought processes, emotional cues (overt or unconscious), and / or many other possibilities.

[0007] Functional MRI can be applied for many different purposes, including neuroscience research and clinical applications, such as planning neurological surgery. One of the first practical applications of fMRI was the mapping of the primary motor cortex, which must be avoided in surgery by a task consisting of finger tapping movements in active condition blocks (with rest as a control condition) to avoid paralysis or deterioration of motor functions.

[0008] However, most fMRI studies, especially those for clinical applications, use simple trial designs, such as blocked trial designs with alternating periods (blocks) of task and control conditions. Each block consists of multiple trials or at least extended periods of expected (task-induced) activity in the targeted brain region, allowing a large signal contrast (relative to the control condition) to be built up and maintained across multiple image acquisitions in that block. This allows the signal-to-noise ratio of the contrast of interest to be improved, and the influence of transient effects near the beginning (and / or end) of the trial condition to be reduced, e.g., reduced dependency on accurate assessment of activation-flow coupling (AFC) responses. Similarly, multiple successive repetitions of task and control blocks are included in the trial paradigm to further improve robustness and sensitivity.

[0009] Regardless of the specific task paradigm and fMRI imaging protocol, cooperation of the imaged subject is typically required in performing the designated task diligently. Even if a specific task only relies on the brain's unconscious or autonomous response, cooperation of the imaged subject is still likely required to avoid movement artifacts that can prohibit the detection of subtle image contrasts of interest. Non-compliance with the designated task can be manifested in various ways, such as the test subject's random response or even non-response to task-related cues. This can easily occur because the subject is distracted or even falls asleep.

[0010] It is clearly understood that problems due to non-compliance with the assigned tasks may be even more significant, more pronounced, and / or more frequent in fMRI studies of psychiatric patients and / or patients suffering from severe neurological disorders or trauma, such as epilepsy, stroke, meningioma, glioma, etc. Other factors that correlate with such diseases or conditions include a reduced ability to perform the tasks due to age or physical disability.

[0011] Considering that consenting patients are generally highly motivated to comply with the testing procedures, patients suffering from adverse psychiatric or neurological conditions may have, among other potentially perplexing problems, a reduced ability to understand a wide range of possible task instructions, reduced attention while performing a task, and / or physical difficulties receiving sensory cues, concentrating on cues, and responding accurately and / or on time. Furthermore, in subjects who are highly motivated to comply with the testing procedures, this compliance may affect their ability to perform as desired, which may result in false positive fMRI results that are of significance or difficulty for interpretation.

[0012] To alleviate the problems associated with poor patient (or test subject) compliance, several approaches are known in the art. For example, fMRI mapping approaches known in the art rely on spontaneous brain activity while the subject is resting, often referred to as resting-state functional connectivity fMRI. Correlation patterns between fluctuating activity in different brain regions can be used to access functional connectivity and activity in the resting state, i.e., a state in which the subject's participation is not required, even when the subject is sedated or asleep. However, limitations are quite obvious. It is difficult to identify specific brain regions associated with specific functions, which is important for many clinical (and research) purposes.

[0013] US Patent Application Publication No. 2015 / 272469 describes an approach for decomposing fMRI data acquired from a subject while performing a functional task into task-related and non-task signals using a model related to the functional task. This allows the advantages of traditional event-task-related fMRI (e.g., in terms of specificity to the function of interest) to be combined with those of resting-state functional connectivity fMRI (rs-fcfMRI) by generating task activity maps from the task signals and non-task activity maps from the non-task signals. Both maps can then be combined with appropriate weighting. This has the advantage that it allows relying on spontaneous activation mapping even when compliance is insufficient to obtain useful task-related results. However, this does not solve the problem of loss of task-related information when compliance is insufficient, even if at least some information can still be gleaned from the dataset.

[0014] US Patent Application Publication No. 2019 / 159715A1 discloses a system in which brain activity in response to a particular stimulus is determined. A follow-up task is constructed to account for the actual measured brain activity. In other words, the original brain activity measurement is essentially a calibration to determine optimal task settings.

[0015] WO 2010 / 053976 A2 discloses administering a neuropharmaceutical to a patient to improve the efficiency of neurofeedback training: fMRI is taken of the patient while they perform a task under the influence of the drug, and the patient is shown the resulting brain activity as neurofeedback.

[0016] U.S. Patent Application Publication No. 2014 / 206982 discloses calibrating positron emission tomography (PET) results based on fMRI scans obtained after introducing a PET tracer into a subject to measure the subject's brain activation.

[0017] TR Steger and EF Jackson in 'Real-time motion detection of functional MRI data', J. Appl. Clin. Med. Phys, vol. 5, no. 2, p. 64-70 (2004) disclose a method for quantifying motion in real time during pre-operative fMRI procedures. Summary of the Invention [Problem to be solved by the invention]

[0018] It is an object of embodiments of the present invention to provide good and efficient methods and means for monitoring the performance of a task by a subject being imaged during a functional Magnetic Resonance Imaging (fMRI) procedure.

[0019] It is an advantage of embodiments of the present invention that it is possible to monitor fMRI task performance in substantially real time, for example alerting an operator to intervene if poor task compliance is detected.

[0020] It is an advantage of embodiments of the present invention that a subject being imaged can be notified of poor task performance so that the subject redirects his / her attention to the task being performed during the fMRI examination.

[0021] It is an advantage of embodiments of the present invention that fMRI time series can be annotated with information regarding the subject's task performance / compliance, e.g. during (or shortly after) acquisition of the time series, such that this annotation information can be taken into account in the analysis of the fMRI time series data. For example, a time period or block of the time series can be marked to indicate poor task performance during this period or block, e.g., and thus can be ignored (or treated in another appropriate manner, e.g. given a lower weight) in further processing and analysis of that time series.

[0022] It is an advantage of embodiments of the present invention that an operator can be alerted when a subject being imaged performs a task poorly and / or measures of task performance and / or compliance can be associated with the fMRI data, for example to weight the data according to the detected level of performance and / or compliance and / or to take that level into account in statistical analysis of the fMRI data.

[0023] It is an advantage of embodiments of the present invention that a blocked trial design with alternating periods of at least two different conditions, of which there is at least one task condition (and e.g. at least one control condition), provides good sensitivity, e.g. detectable signals in single image voxels or regions of multiple voxels (or changes in detectable signals when comparing between at least two of these periods), from which real-time or at least fast (e.g. near real-time or substantially real-time) determination of task performance can be made during fMRI imaging runs. By using extended "blocks", e.g. substantial time periods, during which the condition of interest is held, e.g. by repeating multiple trials in each block, good signal quality (contrast, signal to noise ratio) is achieved that allows for (substantially) real-time detection of performance / compliance or lack thereof, e.g. quantification of (near) current (momentary) performance and / or compliance of the task being performed. Multiple trials of the same "block" may be identical events or at least similar to each other in the sense that they each correspond to the same "condition" ascribed to that block. Furthermore, the blocked trial design advantageously reduces dependence on timing-specific parameters that are unknown a priori or only approximately known, such as the resulting hemodynamic response function and / or activation-flow coupling response, as determined from the fMRI data by statistical analysis.

[0024] It is an advantage of embodiments of the present invention that it is possible to detect an adverse condition(s) in the subject being imaged while the fMRI examination is still in progress (or has been very recently completed), so that it is possible to correct such adverse condition and / or to repeat and / or extend the fMRI imaging run to improve data quality that may otherwise be compromised by the adverse condition, such as by not being repeated or extended in time. Examples of such adverse conditions include a distracted state of mind, a drowsy or sleeping subject, loss of consciousness, a reduced level of consciousness or awareness, an agitated state, a confused state, and / or any other state of mind that is unsuitable or less suitable for performing the intended task.

[0025] It is an advantage of embodiments of the present invention that poor compliance with a specified fMRI task, e.g. due to a lack of (complete) understanding of the relevant task instructions, can be detected and mitigated, e.g. by indicating the need for further assistance from the operator.

[0026] It is an advantage of embodiments of the present invention that changes in maintained attention and compliance with task demands during an fMRI scan can be detected, identified and / or quantified.

[0027] It is an advantage of embodiments of the present invention that the methods and / or devices according to the embodiments are particularly suitable for monitoring (and / or improving) the efficiency of fMRI examinations of psychiatric or neurological patients, such as subjects suffering from psychiatric disorders, neurological disorders, or trauma. For example, such disorders and / or trauma cause a decrease in the ability to understand a range of tasks, to remember task instructions, to maintain a sufficient attention span, and / or to perform specified mental and / or physical actions. However, such disruptions may also occur (albeit likely with a lower incidence and / or severity) when studying healthy subjects, so the embodiments are not necessarily limited to being suitable for psychiatric and / or neurological patients.

[0028] It is an advantage of embodiments of the present invention that compliance with (and / or performance on) task requirements can be monitored even for tasks that do not have a directly observable outcome, such as cognitive tasks (e.g., silent word generation, mental imagery, autobiographical memory, etc.). For such tasks, for example, it can be difficult to ascertain during a scanning session whether an imaged subject has successfully (or is) performing a specified task, or whether they have performed the task at all. This difficulty can be overcome or mitigated by methods and / or devices according to embodiments of the present invention.

[0029] Devices and methods according to embodiments of the present invention accomplish these objectives. [Means for solving the problem]

[0030] In a first aspect, the present invention relates to a method for monitoring the performance of a predefined test task performed by a subject during a functional magnetic resonance imaging examination, the method comprising the step of acquiring a time series of functional magnetic resonance images of the brain of the subject using a magnetic resonance imaging system, the method comprising the step of providing at least one sensory stimulus to the subject via a sensory stimulus source during the acquisition of the time series in order to prompt the subject to perform the predefined test task and / or to induce an intended brain activity in accordance with the predefined test task, the method comprising the steps of determining, after the step of providing the sensory stimulus, at least one activity measure indicative of brain activity in at least one predefined region of interest in the brain of the subject imaged by the image time series, and generating at least one signal when the at least one determined activity measure indicates an insufficient performance of the predefined test task and / or an insufficient compliance with the predefined test task, the signal being generated during the acquisition of the time series and / or before removing the subject from the magnetic resonance imaging system.

[0031] In a method according to an embodiment of the present invention, the at least one signal includes an alert signal for notifying an operator of the magnetic resonance imaging system.

[0032] In methods according to embodiments of the present invention, at least one signal includes instructions to the magnetic resonance imaging system to extend, stop, terminate, and / or resume acquisition of the time series.

[0033] In methods according to embodiments of the present invention, the at least one signal includes a performance feedback signal provided to the subject.

[0034] In a method according to an embodiment of the present invention, at least one signal comprises annotation information for storing together with the acquired time series of functional magnetic resonance images such that when analysing the time series after the end of the acquisition, the evolution over the period of the time series of the determined activity measures can be taken into account.

[0035] In methods according to embodiments of the present invention, at least one signal includes a measure of overall attention or distraction and / or compliance across the time series for use in planning future functional magnetic resonance imaging studies having the same test task design and / or involving the same subject population.

[0036] In a method according to an embodiment of the invention, a plurality of sensory stimuli are provided to a subject at different time points during the acquisition of a time series, activity measures are iteratively determined at and in relation to different time points during the acquisition, and a signal is generated if and at the time (i.e. near, shortly after) a most recently determined activity measure indicates insufficient performance and / or compliance.

[0037] In a method according to an embodiment of the present invention, the steps of acquiring the time series and providing the sensory stimuli are performed according to a blocked functional magnetic resonance imaging task paradigm that includes time blocks of at least one active state and at least one control state, and at least one of the plurality of sensory stimuli is included in each time block corresponding to the active state.

[0038] In a method according to an embodiment of the invention, determining the activity measure comprises comparing at least one first image of the time series acquired during an estimated absence of effect of the sensory stimulation on the brain with at least one second image of the time series acquired during an estimated presence of a direct or indirect effect of the sensory stimulation on the brain.

[0039] In a method according to an embodiment of the present invention, providing the sensory stimulation comprises displaying a visual cue, emitting an auditory cue, providing a tactile sensation, and / or providing other types of sensory stimulation, and activity measures are determined for at least one region of interest including a sensory cortex, a primary visual cortex, a primary auditory cortex, a primary somatosensory cortex, portions thereof, and / or combinations thereof, in which neural activation is caused by the sensory stimulation.

[0040] In a method according to an embodiment of the present invention, the step of providing a sensory stimulus is configured to directly or indirectly induce a brain response in a content-specific higher sensory brain region, and an activity measure is determined for at least one region of interest comprising said content-specific higher sensory brain region or a portion thereof.

[0041] In a method according to an embodiment of the present invention, a predetermined test task is configured to evoke visual and / or auditory mental imagery and activity measures are determined for at least one region of interest including higher order perceptual and / or memory regions, content specific higher order sensory regions, one or more primary sensory brain regions, portions thereof and / or combinations thereof.

[0042] In a method according to an embodiment of the present invention, the predetermined test task is configured to induce a motor imagery task, and the at least one region of interest includes the motor cortex, a cerebellar region, parts thereof, and / or combinations thereof.

[0043] In a method according to an embodiment of the present invention, the predetermined test task comprises a spatial navigation imagery task and the at least one region of interest comprises the hippocampus or a portion thereof.

[0044] In a method according to an embodiment of the present invention, the predetermined test task comprises a mental scene construction task and the at least one region of interest comprises the default mode network, the hippocampus, the ventromedial prefrontal cortex, portions thereof, and / or combinations thereof.

[0045] In a method according to an embodiment of the present invention, the predetermined test task comprises an emotional scene construction task and the at least one region of interest comprises the amygdala or a part thereof.

[0046] Methods according to embodiments of the invention include providing instructions and / or trial or pilot runs of the predetermined test task or related tasks prior to acquiring a time series in order to train the subject in performing the predetermined test task, test understanding and / or performance of the predetermined test task, calibrate the expected magnitude of the activity measure, and / or localize a region or regions of interest for use in determining the activity measure.

[0047] A method according to an embodiment of the invention comprises the step of acquiring additional information via different imaging modalities and / or physiological parameter sensors during the acquisition of the time series, which additional information is taken into account in the step of determining the activity measure.

[0048] In methods according to embodiments of the present invention, determining the activity measure comprises performing an attention network activation analysis (e.g. detecting activation or deactivation of dorsal and / or ventral attention networks, resting state networks and / or default mode networks), detecting activation and / or deactivation in at least one region of interest that is elicited in a given task, and / or detecting activation and / or deactivation in at least one region of interest that has no known association with the given task.

[0049] In a method according to an embodiment of the present invention, the at least one region of interest with no known association with the predetermined task includes the anterior cingulate cortex (ACC), the basal ganglia, the ventromedial prefrontal cortex (vmPFC), and / or other brain regions whose brain activity indicates cognitive fatigue.

[0050] In a method according to an embodiment of the present invention, at least one region of interest having no known association with the predetermined task includes a disorder-specific and / or pathology-specific brain region, in which neural activation is indicative of hallucinations or other episodes of the disorder or pathology.

[0051] In a method according to an embodiment of the present invention, the step of detecting activation and / or inactivation in at least one region of interest having no known relevance to a given task is not necessarily limited to being determined after the step of providing a sensory stimulus, for example, such task-unrelated activation or inactivation may be checked continuously (i.e., repeatedly) during image acquisition by functional magnetic resonance imaging, without necessarily being correlated with the time of providing the at least one sensory stimulus.

[0052] A method according to an embodiment of the invention comprises the further step of detecting, during acquisition of the time series, further activity measures indicative of sleep state and / or sleepiness.

[0053] A method according to an embodiment of the present invention comprises the steps of detecting movement of a subject during acquisition of a time series and generating a signal or a further signal if the detected movement exceeds a predefined threshold.

[0054] A method according to an embodiment of the invention comprises performing a functional magnetic resonance imaging analysis on an acquired time series of magnetic resonance imaging images, which is distinct from, for example, determining at least one activity measure, which is performed, for example, after acquisition of the time series has finished and / or after the subject has been removed from the magnetic resonance imaging system. This fMRI analysis is a voxel-based analysis and is not constrained to, for example, one or more particular regions of interest, for example, by fitting a general linear model to each voxel of the imaged volume that are relevant (e.g., including grey matter) and independent of each other (even though image smoothing is used, for example, to control voxel correlation).

[0055] In a second aspect, the present invention relates to a computer program product comprising machine-executable instructions that are executed by a processor, the execution of which causes the processor to perform a method according to an embodiment of the first aspect of the present invention.

[0056] In a third aspect, the present invention relates to a computer system for monitoring the performance of a predefined test task performed by a subject during a functional magnetic resonance imaging examination, comprising an input for receiving, in substantially real time, a time series of functional magnetic resonance images of the subject's brain acquired with a magnetic resonance imaging system, a first output for providing at least one sensory stimulus to the subject via a sensory stimulus source during acquisition of said time series in order to prompt the subject to perform the predefined test task and / or to induce an intended brain activity in accordance with the predefined test task, a processor for determining, after providing the sensory stimulus, based on the images received by said input, at least one activity measure indicative of brain activity in at least one predefined region of interest in the subject's brain imaged by the time series of images, and for generating at least one signal when said at least one determined activity measure is indicative of an insufficient performance of the predefined test task and / or an insufficient compliance with the predefined test task, and a second output for outputting said at least one signal. The processor is configured to generate the signal during acquisition of the time series (ie, prior to completion of the time series) and / or immediately after completion of acquisition of the time series.

[0057] In a fourth aspect, the present invention relates to a magnetic resonance imaging system comprising a computer system according to an embodiment of the third aspect of the invention.

[0058] The independent and dependent claims set out particular and preferred features of the invention. Features from the dependent claims may be combined with features of the independent claims and with features of other dependent claims as appropriate and not necessarily only as explicitly set out in the claims. [Brief description of the drawings]

[0059] [Figure 1] 1 illustrates a magnetic resonance imaging system according to an embodiment of the present invention. [Diagram 2] 1 illustrates an exemplary method according to an embodiment of the present invention. [Diagram 3] 4 illustrates a further exemplary method according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0060] The drawings are schematic and non-limiting, the elements in the drawings are not necessarily drawn to scale, and the invention is not necessarily limited to the particular embodiment(s) of the invention shown in the drawings.

[0061] Notwithstanding the exemplary embodiments described herein below, the present invention is limited only by the claims appended hereto, which are hereby expressly incorporated into this Detailed Description, with each claim, and each combination of claims permitted by the substructure defined by the claims, forming a separate embodiment of the present invention.

[0062] The term "comprising" as used in the claims is not limited to the subsequently listed features, elements or steps and does not exclude additional features, elements or steps. Thus, the term specifies the presence of a stated feature without excluding the further or additional presence of one or more features.

[0063] In this detailed description, numerous specific details are presented. It is possible for embodiments of the present invention to be practiced without these specific details. Moreover, well-known features, elements and / or steps are not necessarily described in detail for the sake of clarity and conciseness of the present disclosure.

[0064] When reference is made in this disclosure to "real-time" or "near real-time", the skilled person will understand that real-time or immediate processing is obviously constrained by technical capabilities as a result of which processing operations necessarily require at least some time for calculation. However, the use of the term "real-time" indicates that the processing being performed is time-critical in nature and is preferably performed quickly to allow actions to be taken in response to the determined results, even if the results are still meaningful. In particular, in an embodiment of the present invention, as a result of the activity measures being determined during the acquisition of a time series of fMRI images, it is possible to take actions in response to the determined activity measures while the time series has not yet been fully acquired. Thus, preferably (although not strictly necessary), the processing takes less time than the length of time (or, for example, twice the length) of one block in the fMRI paradigm / protocol being used (if, of course, the protocol under consideration is not a non-blocked event protocol). For example, this allows corrective actions to be taken before the next block in the fMRI protocol design (e.g., the next block of the same event type) is scheduled to start. Thus, (near) real-time refers to processing times of (merely by way of example) less than 2 minutes, less than 1 minute, preferably less than 30 seconds, or even more preferably less than 10 seconds. Obviously, processing times substantially shorter than that, e.g. less than 1 second, are even more preferred if they can be achieved without excessive impact in terms of cost (e.g. cost of required computing hardware, operating power, etc.). However, at least some of the embodiments of the invention include alerting an operator of the results based on the (near) real-time calculations, so that the operator can then act accordingly.This means that the value added by reducing processing times to (say) less than a few seconds is somewhat limited (e.g., unless an automated response is provided instead of or in addition to alerting an operator), given the time required for an operator to take responsive action when necessary.

[0065] References to "tasks" in this disclosure can be interpreted quite broadly. Such tasks typically involve specific responses in the subject's brain, responses that can be localized, for example, with respect to time (at the moment the task is performed or at a moment that has a specific temporal relationship to the moment the task is performed) and space (e.g., to one or more specific brain regions). However, a task may not necessarily involve an active behavior of the subject, and may be limited, for example, to the passive observation by the subject of one or more stimuli (e.g., triggered via any sensory organ or a combination of multiple senses) and / or subconscious and / or unconscious responses to such stimuli, such as, for example, emotional responses, mood changes, sentiments, memory functions, etc. Obviously, this does not exclude tasks that actually involve active behavior (reactions) of the subject as defined by a specific task paradigm. Such active task-guided (prescribed) behaviors may be externally observable, such as finger movements, speech or verbalization, among many other possibilities, but may also simply be internal processes, such as thought processes, mental calculations, intentional memory processes, imaged or simulated physical actions or scenarios, among many other possibilities, e.g., intentional and conscious processes (as opposed to the subconscious and / or unconscious responses described above). It is noted that tasks that do not involve an externally observable response of the subject may be particularly difficult to monitor, i.e., difficult to determine whether the task is actually being performed as intended. This difficulty can be advantageously overcome by embodiments of the present invention.

[0066] In a first aspect, the present invention relates to a method for monitoring performance of (and / or compliance with) a predetermined test task performed by an imaged subject during a functional magnetic resonance imaging examination. In particular, a method according to an embodiment of the invention is a computer-implemented method, e.g. executed by a computer program product when executed on a suitable computing platform (although embodiments implemented wholly or partly in computing hardware, e.g. digital processing hardware, are not excluded).

[0067] The method is carried out by interacting with a magnetic resonance imaging system. Figure 1 shows an example of such a magnetic resonance imaging system 100, for example an MRI system generally suitable for implementing steps of a method according to an embodiment of the invention. The MRI system 100 comprises a magnet 104, for example a superconducting cylindrical magnet through which a bore 105 is defined. Variations and / or alternatives to this type of magnet are possible, for example the magnet can also be a separate cylindrical magnet or an open magnet (for example with magnet parts arranged around, for example above and below, an open space that receives a subject to be imaged). The magnet 104 includes a number of superconducting coils, for example in a cryostat. Inside the bore 105 an imaging region 108 is defined in which the subject is imaged. In this imaging region 108 the magnetic field of the magnet 104 is typically strong and uniform, i.e. strong and uniform enough to perform MR imaging. For open magnets or other alternative types of magnet assemblies, a bore cavity may not be defined, but the imaging region 108 is still clearly identifiable, e.g., as characterized by suitable magnetic field strength and uniformity. In its application, the magnetic resonance data acquired by the system typically represents the contents of a space defined by the imaging region 108 or a portion thereof, such as a particular region of interest defined within the imaging region.

[0068] A subject 110, which may be a human subject, for example a patient to be diagnosed or examined, may be placed inside the imaging region 108 in use. For this purpose, the system is provided with a subject support 112, for example a movable patient couch. By controlling the subject support, a body part or region of interest of the subject on the support may be moved to a suitable position inside the imaging region 108.

[0069] The system further comprises a number of magnetic field gradient coils 103 configured to spatially encode magnetic spins within the imaging region 108 (e.g., by superimposing a localized magnetic field variation on the strong magnetic field of the magnet). For example, the magnetic field gradient coils comprise at least three coils (or sets of coils) that allow spatial encoding in three orthogonal spatial directions, i.e., in any direction in space. The magnetic field gradient coils 103 are controlled as a function of time, but also controlled to select the direction of spatial encoding, e.g., to produce gradients that are pulsed, graded, or otherwise varying as a function of time. For example, the magnetic field gradient coil power supply (or power supplies) are controlled by a controller, processor, or computer to achieve a magnetic field (gradient) according to a desired spatial and / or time function. The system typically comprises a controller 118 for controlling the magnetic field gradient coils 103.

[0070] The system typically includes a radio frequency (RF) coil 115 (RF antenna device) that functions as a radio frequency receiver and transmitter. For example, the RF coil transmits a signal that disrupts the alignment of (nuclear) spins with the magnetic field (of atoms and molecules in the imaging region 108), and the RF coil receives a return signal that is generated by precession and relaxation that brings the disrupted nuclear spins back into alignment with the magnetic field. The RF coil is a composition that includes multiple separate coil elements and / or support devices. The RF coil may be integrated with a bore system or is a separate component that is placed in close proximity to or in contact with the subject being imaged. For example, the RF coil(s) are integrated into a head coil assembly for imaging the brain of the subject 110.

[0071] The RF coil 115 comprises or is connected to an RF transceiver 116. Obviously, the system may also comprise separate RF transmitting and receiving antennas and corresponding (separate) receiver and transmitter systems, or may comprise multiple antenna coils and transceivers, and / or any combination of receiver and / or transmitter systems. For example, multiple antenna coils are connected to separate receiving and / or transmitting channels, which are controlled separately, in unison, or in any combination deemed suitable for the intended purpose. Several parallel channels, such as SENSE (sensitivity encoding) or ASSET (array coil spatial sensitivity encoding), are useful, for example, for parallel imaging techniques.

[0072] The MRI system 100 also includes a sensory stimulus source 120 that provides sensory stimuli to the subject during the MRI imaging operation. Such sensory stimuli include signals that can be provided to the subject through any (or a combination of) of the body's physical senses, such as touch, smell, taste, hearing, and vision. Typically, the sensory stimulus source includes a light or display (which may include a projection screen and / or an optical system that relays the displayed information to the interior of the bore volume) that provides visual stimuli to the subject while positioned in the imaging region, and / or a speaker or sound source (which may include a relay system using induced pneumatic or contact transmission of sound into the bore volume, such as headphones) that provides auditory stimuli to the subject.

[0073] The sensory stimulation source 120 is used to trigger activity in the brain region of interest in the subject when performing the method according to the embodiment. The subject thus receives an instruction to perform a "task" signaled by the sensory stimulation source. As already mentioned, this task can be passive or active (or a combination thereof) in nature. The timing is controlled with respect to the timing of the acquired MRI data, so that the acquired images can be analyzed by taking into account the stimulation given at the time of image acquisition (or at least approximately known time delay before the time of image acquisition).

[0074] The RF transceiver 116, the gradient coil controller 118 and the sensory stimulation source 120 are connected to a hardware interface 121 of a computer system 122. The hardware interface is, for example, a digital network interface or a data bus interface. The hardware interface is configured to exchange data and / or commands between the computer system and other components of the system. The hardware interface is also configured to exchange data and / or commands between different components of the computer system. The hardware interface is not necessarily a homogenous interface infrastructure, but may, for example, include or consist of different types of data exchange media, computer networks, communication buses, shared memory access structures, etc.

[0075] The MRI system 100 is particularly configured to acquire fMRI images. In other words, a time series of images can be acquired, and each image in the sequence is reconstructed, for example, by the computer system 122, based on signals detected by an RF transceiver and an antenna coil(s) connected thereto. For fMRI imaging, the RF coil(s) are integrated into a head coil assembly to achieve good signal quality, particularly for imaging the subject's brain.

[0076] An embodiment in which the computer system comprises multiple computers interconnected by a suitable network infrastructure is not necessarily excluded. For example, the processing performed by the computer system may be distributed across a computing cluster or a cloud computing platform. Also, although a conventional computer architecture is referred to below, it will be clear to the skilled person that this is not necessarily applicable to all embodiments. For example, the computer system may comprise or even consist of dedicated hardware implementing the processing functions described below in this specification, for example with application specific integrated circuits and / or field programmable gate array hardware. It will also be understood that when a single computer system is referred to, different functions of this computer system may be assigned to multiple dedicated computer systems capable of exchanging data, for example.

[0077] The computer system 122 includes at least one processor 123 that is operatively connected to a hardware interface, such as to allow the gradient coils, the sensory stimulation source, and the RF transceiver to be controlled by the processor and / or to allow data received by the RF transceiver to be processed by the processor.

[0078] The computer system 122 includes a user interface 124 for interacting with, for example, an operator operating the MRI system. The processor is also coupled to the user interface, for example, to allow a user to select actions to be performed by the MRI system, parameters to apply, review data indicative of the operation of the system, and / or images to be acquired by (or derived from) the system. The user interface may include a display, a keyboard, a pointer device, a touch interface, and / or other similar components as are common in human-computer interaction systems. The user interface may be configured to render images, for example, via a display.

[0079] Computer system 122 includes memory 125. The memory may include or be composed of different types of memory, all of which are accessible (directly or indirectly) to the processor. This may include, for example, random access memory, cache memory, non-volatile memory (such as flash memory), hard drives, solid-state drives, and / or other storage devices. Memory comprises a non-transitory computer-readable medium.

[0080] The memory includes machine-executable instructions for enabling the processor to perform various data processing tasks and / or for controlling other components of the MRI system. The machine-executable instructions enable the processor to perform at least a portion of a method according to an embodiment of the present invention.

[0081] Methods according to embodiments of the invention may be embodied in a source program, an executable program (e.g., bytecode or intermediate language code, object code, and / or machine code), a computer-interpretable script, and / or any other entity including an instruction set that can be executed directly or indirectly by a computer system (e.g., by a simplified program or code, such as an interpreter, a just-in-time compiler, a compiler, etc., and / or a supporting code library, framework, operating system, etc.), such as by a processor of the computer system 122. It is pointed out that the embodiments are not limited to implementation as computer code, and for example, the methods may be executed (fully or partially) by dedicated hardware specifically designed for the intended method, for example as an ASIC and / or by a hardware configuration that can be loaded into a programmable hardware configuration, such as an FPGA.

[0082] The memory 125 also contains representations of pulse sequence commands that enable the processor to control the MRI system in a corresponding manner, for example by controlling the gradient coils and the RF transceiver in accordance with these commands, and the memory also contains magnetic resonance imaging data acquired by controlling the MRI system with the pulse sequence commands.

[0083] The memory 125 includes representations of sensory stimulation commands (or data) for controlling the sensory stimulation source 120. These sensory stimulation commands are associated with or integrated with the pulse sequence commands so that, upon execution, a well-defined relationship in timing is established between the actions performed by the gradient coils and RF transceiver (and / or further components of the MRI system) and the cues given to the subject via the sensory stimulation source. For example, the sensory stimulation commands command or induce a change between an active state, in which a particular task is performed, and a resting state, in which no task or other tasks for reference / control are performed, for example, during the acquisition of the MRI data. Metadata is associated with the acquired MRI data to indicate the status of the sensory stimulation source, e.g., an indication of the information, or type of information, given to the subject at a time corresponding to the time of the acquisition of the MRI data.

[0084] The computer system 122 is also configured to reconstruct slice images from the MRI data acquired from the RF transceiver. Methods for controlling an MRI system with appropriate pulse sequences and reconstructing slice images from signals acquired by the RF transceiver in response thereto are well known in the art and will not be discussed in detail in this disclosure. However, it should be noted that the pulse sequences typically include commands or data for controlling the MRI system to acquire image data (reconstructed by corresponding processing of acquired signals) that are sensitive to changes in the human brain due to brain activity. These changes depend on the brain's hemodynamic response in response to brain activity, such as, for example, increased blood flow in active regions of the brain, changes in blood consumption rate and / or blood resource supply, and / or signal contrast that depends on blood oxygen concentration.

[0085] The computer system 122 comprises a detection and / or analysis module for analyzing the fMRI images substantially in (near) real time to monitor the current attention level of the imaged subject, e.g., by implementation via suitable software and / or hardware, as described in detail herein below in the description of the method according to an embodiment of the present invention.

[0086] The computer system 122 comprises, for example by implementing, a response and / or alert module for providing information to an operator of the system and / or to a subject, based on the output of the detection and / or analysis module, for example to correct a loss of attention of the subject during the fMRI examination, for example by operator intervention and / or by drawing the subject's attention to the loss of attention to the fMRI task being performed, as will also be further described herein below in connection with methods according to embodiments.

[0087] With reference to Fig. 2 (and / or Fig. 3), a method 10 for monitoring the execution of a predetermined test task performed by a subject imaged during a functional magnetic resonance imaging examination according to an embodiment of the present invention is shown. It should be noted that in Fig. 2 and Fig. 3, a vertical axis T is shown to illustrate the temporal relationship of the different steps of the method relative to one another, e.g. as a flow of time T. This temporal relationship is merely exemplary, but is provided to help a person skilled in the art understand the general flow of events in a possible method according to an embodiment, especially when multiple parallel streams of events are involved.

[0088] The method 10 comprises acquiring 11 a time series of functional magnetic resonance images of the subject's brain using a magnetic resonance imaging system, such as MRI system 100, as described herein above. Acquiring the time series of images comprises controlling the MRI system according to a pulse sequence and acquiring signals by an RF transceiver. The time series thus comprises a series of images acquired during subsequent time blocks T0, T1, ..., TN (or substantially acquired at a series of instants in time T0, T1, ..., TN).

[0089] Acquiring the time series also includes processing the collected signals to reconstruct an image, e.g., a tomographic image. The reconstructed image may be a volumetric (3D) tomographic image, one or more planar (cross-sectional, 2D) images, or a combination thereof. It is known in the art to reconstruct the signals substantially in real time, i.e., while the time series is still being acquired (before the time series is finished), although the instantaneous reconstruction may (but not necessarily) be performed at a lower resolution, in a smaller area (e.g., only one or a few slices), and / or with a lower image quality than the final reconstruction. Such a final reconstruction may, for example, be performed at a point in time after the time series has been acquired in its entirety, e.g., resulting in a diagnosis based on information contained in the time series as a whole and at the image quality and resolution of the highest achievable reconstruction.

[0090] Depending on the fMRI imaging technique used, the acquired signals may represent substantially the same time instant for multiple different locations of the brain (e.g., using appropriate 3D imaging sequences and / or accelerated imaging techniques) or may represent slightly different time instants for multiple different locations of the brain (e.g., using conventional echo-planar imaging techniques), but it is also known in the art to correct small time offsets, if necessary (e.g., using slice timing correction methods), to obtain an image that represents a substantially single time instant throughout the imaged volume. It can therefore be envisaged that the acquired time series will include (2D, preferably 3D) images that are substantially (or at least approximately) indicative of the imaged brain at a corresponding series of time points, for example by including appropriate pre-processing steps on the signals / images, if necessary or desired.

[0091] Given the state of the art in the fields of MRI and fMRI, the present disclosure provides a method for detecting brain activity (e.g., T2 * We will not go into details regarding suitable pulse sequences, image reconstruction methods, and / or fMRI analysis techniques for producing images representative of brain activity, such as statistical maps of the assumed brain activity, from MRI images with suitable sensitivity to such brain activity, such as weighted images. The skilled person is familiar with suitable methods known in the art. However, it should be noted that techniques are known both for performing MRI image reconstruction in (near) real time and for detecting brain activity based on such images in (near) real time (e.g. via evaluation of the BOLD response).

[0092] As an example, acquired fMRI data in a time series includes a number of MRI images at a number of different echo times after application of an excitation RF pulse for each time sample point in the time series.

[0093] For example, reference is made to EP 3660530A1 for a suitable (near) real-time fMRI processing method, without necessarily limiting embodiments of the present invention to that particular approach. Considering the requirements of the method according to embodiments of the present invention, more simple and / or straightforward approaches to determining activity measures 13 can be considered as well, e.g. advantageously reducing memory and / or processing requirements to determine the activity measures sufficiently fast, e.g. near real-time, as will be discussed in more detail further below in this specification. Obviously, this does not exclude more complex approaches, even if computing resources are available. For example, more conventional fMRI analysis approaches, such as, for example, commonly applied statistical parametric mapping methods, are applied to the partial time series up to the time at which the activity measures are determined (i.e. when the calculations are performed), and the method is repeated at a later time for the time series up to a later time if other values ​​for the activity measures are desired at that time. Even if such conventional processing is applied, it is also clear that some intermediate results from previous calculations (for shorter partial time series, at earlier time points) can be easily reused (for longer partial time series, at later time points) to simplify the calculations, at least to some extent. Examples of such reusable intermediate results include slice time correction, motion compensation realignment, co-registration to a reference image, normalization, image smoothing, image segmentation, etc., for images of a time series already collected for an earlier partial sequence. Even the evaluation of (but not limited to) general linear models can be simplified if results or intermediate results of earlier evaluations for partial time series are available.However, it should be noted that for reasons of efficiency and careful use of computing resources, methods specifically designed to achieve fast or near real-time fMRI brain activity measures are preferred to conventional processing flows, such as those exemplified by the algorithms provided in the Statistical Parametric Mapping toolbox (well known in the art and see, e.g., SPM12 revision 7771, publicly available from the Wellcome Centre for Human Neuroimaging, London, UK).

[0094] Method 10 comprises providing sensory stimuli 12 to the subject, e.g., via sensory stimulation source 120, during acquisition of the time series of fMRI images to prompt the subject to perform a predetermined test task (or an element or step thereof) or to directly induce intended brain activity in the subject's brain according to the predetermined test task by the stimulation (see note above, a passive or unconscious brain response to the stimulation can also be considered as a "task" according to at least some embodiments of the present invention). "During acquisition" refers to at least one image of the time series being acquired while an effect of the stimulation on the brain is estimated and / or intended, e.g., immediately after providing the stimulation (either directly or due to an intended response of the subject by the task), and at least one other image of the time series being acquired while no effect of the stimulation can be estimated, e.g., before or at a sufficient time after the stimulation (e.g., particularly if the subject has not been instructed to maintain the prompted activity for a period of time extending after this "sufficient" time). Thus, as is generally known in the art, the effect of the stimulation (direct or indirect) can be evaluated by comparing these at least two images. It is noted that typically a large number of images are processed to accurately determine the activity of interest in the brain, for example, at least 10 images without effect are compared with at least 10 images with effect. It is also noted that the sensory stimulation is typically repeated to repeatedly induce the brain activity of interest so as to improve the sensitivity by joint analysis. Furthermore, the stimulation is repeated during the first time block, while the stimulation is absent during the second (next) time block, and this type of block is alternated for the next few times to further improve robustness.

[0095] As known in the art, a task is selected (e.g., by an operator) from a number of different paradigms designed to generate activity in a specific brain region or regions of interest, e.g., for diagnostic purposes or to plan treatment / surgery. These different task paradigms also take into account different subject populations / groups, and can be tailored, e.g., for patients with a specific disorder or condition. During an fMRI session, the subject (e.g., patient) is then instructed to perform this specific task, with one or more stimuli intended to carry these instructions to directly generate specific brain activity patterns and / or to act as cues to synchronize the timing of the subject's task-related activity with the timing of fMRI image acquisition. For more complex tasks, the subject receives detailed instructions in advance, so that simple cues (stimuli) may be sufficient during the execution of the task.

[0096] Providing a stimulus may include displaying a visual cue, such as an instruction or a pictogram containing an instruction, an image intended to evoke a particular brain response (e.g., a thought, memory, or emotion), or a simple cue of a degree of change in lighting (e.g., a flash, etc.). Providing a stimulus may include emitting an auditory cue, such as a sound, a tone, an auditory verbal instruction, and / or a verbal or non-verbal sound intended to evoke a particular brain response. Providing a stimulus may also include any other cue that can be sensed by the subject's brain in any other recognizable manner, such as touch, or the release of a molecule that can be detected by the sense of smell.

[0097] The method considers individual differences in the ability to perform the fMRI task among multiple subjects, and in particular the underlying disease and / or pathology. For example, this prior information is taken into account when selecting an appropriate fMRI paradigm / protocol. Furthermore, the method comprises a step 22 of providing instructions (e.g. automated or operator-dependent) to the subject before acquiring the fMRI time series and / or before a trial run of the task or related tasks. These instructions vary depending on the subject (e.g. underlying disease / pathology), even for the same task. Furthermore, the method comprises a step of providing the subject with a stimulus(s) before acquiring the fMRI time series, in order to practice the task and / or to test whether the understanding of the task instructions is adequate. For example, a non-imaging-related cognitive-based experimental procedure is performed, which depends on the psychiatric disorder. In such a trial run, sensory stimuli can be applied without yet performing a brain scan. Furthermore, the task performed during this trial run may (e.g., slightly) differ from the task performed during the imaging run, for example by receiving feedback (e.g., audio or motion) from the subject to test whether the task is performed correctly (although the actual task performed during the imaging run does not necessarily require such feedback). Another difference may be that the trial task may involve presenting a particular stimulus, whereas the task for the actual imaging may require the subject to simply imagine (or recall) the same or a similar object, setting and / or scene. Furthermore, such differences between actual perception and memory or imagination advantageously allow the subject's response to be tested / confirmed, which may also serve calibration purposes.For example, during trial execution (or a portion thereof), separate fMRI image time series are acquired (from those described above) to detect specific brain regions for use as regions of interest in monitored real-time task compliance as discussed further herein below, and / or to calibrate the magnitude of activation / deactivation (e.g., in specific regions of interest) predicted by this monitoring.

[0098] The method 10 comprises a step 13 of determining, after providing the stimulation but before acquisition of the time series is terminated, an activity measure indicative of brain activity in a given region of interest in the subject's brain volume imaged by the time series of fMRI images by comparing at least one first image of the time series acquired during an estimated no effect of the stimulation on the brain with at least one second image of the time series acquired during an estimated effect of the stimulation on the brain. For example, the first image is acquired before the stimulation is applied, whereas the second image is acquired during or after providing the stimulation, such as immediately after, which may for example be less than one minute after the presentation, such as less than 30 seconds, such as less than 10 seconds. It should be noted that while direct effects of stimulation on the brain are typically temporary, indirect effects can be sustained and thus detectable for substantially longer periods of time, e.g. in combination with an instruction to the subject to initiate a specific mental or (limited) physical activity when given together with the stimulation (e.g. extending for minutes after the fact, without any restriction to maintaining neutral activity for much longer periods, although real concerns may arise when a block of task-induced activity extends for too long a period of time). It should be noted that physical activity is obviously limited by scanner bore constraints and the typical requirement (or strong preference) to prevent movement during acquisition of the time series, but some limited physical activity, e.g. finger tapping movements, is still possible without interfering with the test.

[0099] Thus, while a patient performs a task guided or synchronized by one or more stimuli and an MRI system collects a time series of MRI images, a method according to an embodiment provides for (near) real-time analysis of (at least some of) the MRI images to determine whether the subject is performing the task properly and / or is distracted.

[0100] Comparing at least the first and second images may be broadly interpreted as including any operation involving image information from the first and second images, for example to determine a qualitative or quantitative measure indicating that brain activity was present in the second image but not present or present to a lesser extent in the first image, or vice versa, for example, that there is less brain activity in the second image compared to the first image. This includes image subtraction, image segmentation, or calculation of similar contrast measures. This also includes appropriate pre-processing and / or processing steps, for example to compensate for motion between the images, correct timing, register the images to a known coordinate system, for example so that the region of interest is easily found in the images, image smoothing, etc. This also includes evaluation of a model, such as a general linear model, for assigning strength and / or confidence to the brain activity in the region of interest, especially when more than two images are considered.

[0101] Step 13 of determining activity measures may optionally be augmented with additional imaging and / or sensor information, such as, for example, by acquiring positron emission tomography (PET) images, electroencephalogram (EEG), ultrasound (US) images, and / or other measurements and / or images simultaneously with acquiring the fMRI time series. Different sources of information (e.g., different imaging modalities) allow different parameters (or the same parameters using different methods) that may be indicative of loss of attention and / or task compliance to be tested, thereby allowing a more robust signal indicative of such poor task performance to be generated.

[0102] However, step 13 of determining the activity measure differs (although not necessarily) from known approaches of conventional fMRI processing. First, while in conventional fMRI imaging approaches a relatively large number of images are collected to determine a robust measure of brain activity for each image voxel of the brain, the activity measure determined in step 13 according to the embodiment is limited to a single scalar value, or a few scalar values, indicating activity in a single voxel in one particular brain region (region of interest) or in a few brain regions (e.g., when the region of interest may consist of multiple regions of interest that may be disjoint). In other words, while conventional fMRI analysis aims to visualize and / or quantify brain regions that are involved in or (de)activated by a task and that are only approximately known a priori (and therefore need to be imaged), the activity measure according to the embodiment is calculated in a priori known (i.e. predetermined) brain regions that are only slightly involved in performing the task and preferably performing the task carefully and diligently, not necessarily in the regions that the fMRI examination aims to detect (but is not limited to). However, these primary regions of interest in the test and the regions used in assessing the activity measures are not necessarily completely separate or distinct: for example, the aim of the test is to precisely define the brain region involved in the task, while a larger region in which this brain region is likely to be located can be used as the region of interest for determining the activity measure. For example, this region of interest could correspond to a gyrus, several adjacent gyrus, or even an entire cerebral lobe, in which case the specific point or location involved in the task would be localized.

[0103] Secondly, the number of images based on which the activity measure is determined is relatively small, e.g. only two. While a larger number of images in the sequence is preferred to perform an accurate fMRI analysis, the activity measure 13 is preferably determined from only a few images (i.e. fewer than the entire time series) in order to be more sensitive to (approximately) instantaneous changes in task compliance or performance. The fact that the assessment of the activity measure can be limited to a specific predefined region(s) of interest and can be based on only a few images contributes to efficient processing and thus to the desirable near real-time assessment of the activity measure.

[0104] However, this does not mean that it is not possible to take into account a larger number of images. For example, the images based on which the activity measure is determined could include all (or many or most) of the images in the time series collected up to the time the activity measure is calculated, and regressors could, for example, be used in a model fitted to the time series to determine changes in activity indicative of a decrease (or increase) in compliance or performance, e.g., a decrease (or increase) in attention. Multiple such regressors could be included to model more complex attention effects, e.g., other than linear. Furthermore, if the activity measure is repeatedly evaluated (13), previous values ​​of the activity measure are taken into account when evaluating the model at a later time point, e.g., as covariates to detect changes in relation to previously evaluated activity measures.

[0105] As a third potential difference in relation to conventional fMRI analysis, it should be noted that the detection of sufficiently strong brain activity (or a sufficiently strong decrease thereof) in a region of interest that is at least slightly involved in performing a task does not necessarily require statistical analysis to the extent that it is commonly used in fMRI analysis. It should be noted that one of the problems with conventional fMRI analysis arises from the need to adequately control the number of false positives in light of the large number of voxels considered (e.g. corresponding to the entire gray matter in the brain). In particular, an appropriate correction of the statistical threshold is required, such as the Bonferroni correction (although more complex alternatives are also commonly used). However, this is not an issue, or at least is less of an issue, when a single region of interest or a relatively small number of regions of interest are considered, for example in the case of (near) real-time analysis according to the embodiment.

[0106] Although it is important to determine which brain regions are involved in an fMRI task in a robust and statistically reliable manner, for the purpose of monitoring task performance, the need to simply detect whether a particular brain region(s) is (or is) activated by the task is deemed less important. For example, a false positive merely attracts the operator's attention without many further consequences for a subject who is mistakenly considered distracted, whereas a false positive in the final analysis of the fMRI time series could potentially lead to (for example) an incorrect medical diagnosis. Thus, the step 13 of determining the activity measure is kept fairly simple and thus efficient to perform in (near) real-time. For example, the region(s) of interest are defined to avoid significant contrast with nearby structures, making it possible to avoid (but not limited to) a step of image registration prior to analysis. Similarly, for example, processing steps such as image smoothing and / or slice timing correction, which are conventionally applied to enable statistically robust estimation (e.g. based on the properties of Gaussian random fields), are not (strictly) necessary for the purpose of real-time monitoring.

[0107] For example, during an fMRI examination session (e.g., while still acquiring time series), the fMRI signal may be used to map transverse relaxation (T2 * For example, a predefined model of fMRI data variability may be used to generate a T2 * To generate the map, a transverse relaxation map is taken into consideration, for example the first image described above, and a reference T2 * The map is compared (e.g., the first image is based on data collected at a time point preceding the stimulation). The blood-oxygen-level-dependent (BOLD) response of the region(s) of interest during the current activity time window (e.g., immediately after providing the stimulation) is then determined using the results of this comparison. The activity measure may then be based on (or correspond to) this assessed BOLD response.

[0108] The second image(s) are preferably images representative of the current state during the examination, e.g. recently acquired in relation to the action 13 of determining the activity measure, while the first image(s) are not necessarily limited to reference images or images acquired in the most recent period of inactivity (e.g. "resting block"). Since such reference images are considered to include the general background activity of the brain, it is advantageous to combine several such images, e.g. (but not limited to) by averaging. For example, many images, or even all of them, acquired up until the time the activity measure is determined and suitable to be used as representing the control ("resting", "baseline") condition are taken into account, even if, e.g., not all of these reference images belong to the most recent "resting block". Thus, for each reference time window (e.g. "resting block"), T2 * The baseline T2 generated by the time the map was generated and activity measures were calculated * The maps are combined to obtain a relatively robust reference image (the "first image").

[0109] It will also be apparent to one skilled in the art that the data collected for a time block corresponding to a reference or control condition may be filtered to avoid transient effects of the active condition time block, for example by excluding a portion of the data collected during the initial time period of the time block.

[0110] For example, in a simple paradigm R1A1R2A2R3A3R4A4R5A5, which includes a series of alternating active (A) and resting (B) state blocks, when calculating an activity measure representative of task compliance during the execution of active state block A4, the first image(s) are based on data collected in resting state blocks R1, R2, R3, and R4, while the second image(s) are limited to data collected during active state block A4. However, using data collected in previous active state blocks is not necessarily excluded, e.g. a weighting approach is applied to assign higher importance to the most recently collected data without discarding earlier data altogether, so that the activity measure can more accurately track changes of attention (e.g. less sensitive to noise) during the test. It should be noted that the embodiments are also not necessarily limited to blocked fMRI protocols, e.g. the method according to the embodiments also applies mutatis mutandis to event-related protocols.

[0111] By determining activity measures indicative of neural activity in one or more regions of interest in the subject's brain in (near) real-time, methods according to embodiments allow for monitoring of the fMRI study, particularly while acquiring the fMRI time series, allowing an operator (or an automated monitoring system configured to take appropriate action) to ascertain whether the task is being performed properly and / or whether there are any distractions. This is particularly useful in potentially impaired brains of neurological or psychiatric patients, and where performance of the fMRI task is not readily apparent by the patient's external response (e.g., mental tasks as opposed to finger tapping or other easily detectable tasks).

[0112] The step 13 of determining an activity measure indicative of brain activity in a given region of interest may also include performing an (almost) real-time attentional (e.g. resting state or default mode) activation analysis, e.g., where the assessed activities (e.g. BOLD values) of multiple different regions of interest are considered in relation to each other. Thus, the measure of brain activity (or a component thereof) may indicate the relationship between the activities in multiple different regions, instead of simply indicating the activity in a single corresponding region (or a list of activities in multiple different regions).

[0113] For example, to attend to attention, recall instructions, and process a given fMRI task, specific resting-state networks are engaged, e.g., default mode networks are activated during rest and therefore deactivated during task performance. Other plausible networks for task-related cognition include the fronto-parietal network, dorsal and / or ventral attention networks (FPN, DAN, VAN, respectively). Network activation analysis involves the calculation of one or more measures of connectivity and / or coherence that represent the joint (coherent) activity of different given regions in the brain. For example, functional connectivity measures may be determined based on (but not limited to) graph analysis, where the nodes of the graph represent different brain regions and the edges of the graph represent the connections between these regions. A simpler (exemplary) approach is based on a measure of correlation or association between time series representing multiple different brain regions, e.g., by averaging acquired fMRI signals over voxels associated with brain regions to obtain a matrix X=(X1,X2,...,X n ) T where each random variable X i represents the time series of detected activity in the corresponding brain region.

[0114] For example, during an fMRI task scanning session, activity measures are determined (possibly iteratively) to ascertain in substantially real time whether there is activation and / or inactivation in brain regions relevant to the task, such as one or more regions of interest that are predicted to be either activated or inactivated based on prior knowledge, e.g., from one or more hypotheses underlying the task design and / or from a normative database of predicted activations and inactivations in a number of different brain regions for a number of different task paradigms. If the predicted activation and / or inactivation is not detected, i.e., is not reflected by the determined activity measures, an operator alert may be triggered or other action may be taken, as described herein below.

[0115] Furthermore, the regions of interest for which activity measures are determined are not necessarily limited to brain regions relevant to the task. For example, the method of the present invention may alternatively or additionally identify activations in irrelevant and / or non-task related regions that may indicate, for example, mental instability. In other words, in addition to identifying whether a predicted activation or deactivation is present in a particular region of interest, the activity measures may also identify unexpected activation or deactivation in regions of interest that have no known association with the fMRI task or that have a known association with being distracted or contemplative.

[0116] Similarly, attentional neural network activation analysis is performed to identify attentional networks that are predicted to be more or less activated from one or more hypotheses of the activation state of the task and / or based on prior knowledge, such as a database of predicted activations and deactivations for particular attentional networks in particular task paradigms.

[0117] A combination of several of these examples may be used. For example, the predicted activation and / or deactivation in one or more regions of interest related to the task is affirmed in the check. If the conditions of this check are not met, a warning or other signal is generated, whereas a second check is performed to confirm the activation of unrelated and / or non-task related regions, where such activation is not predicted based on one or more hypotheses and / or prior knowledge of the task. If this second check is positive, a warning or other signal (which may be the same as the above-mentioned signal or warning or may be different to provide additional disambiguating information) is generated. If this second signal is negative (i.e., no unexpected activation was detected), a third check confirms the predicted increase or decrease in the activation of the attention network based on one or more hypotheses and / or prior knowledge of the task. If this third check is positive, e.g., if the detected activation of the attention network follows the prediction, no further action needs to be taken, e.g., until the check is (optionally) repeated at a later time during the fMRI examination. This example of a combination of three stages of validation is, of course, merely illustrative and a person skilled in the art will be able to think of many similar or equivalent approaches without exerting any inventive effort.

[0118] For example, for a task that uses a sensory input (one or more stimuli), the region(s) of interest include sensory regions (and / or primary visual, auditory and / or somatosensory cortices, portions thereof, and / or combinations thereof) that are presumed to be (sufficiently) activated or at least not deactivated by the stimuli, e.g., by being presumed to be active when the task is performed, the negation of this presumption indicating insufficient attention and / or performance. The region(s) of interest include content-specific higher sensory brain regions (or portions thereof and / or combinations thereof), such as the fusiform "face" region or the parahippocampal "location" (spatial location) region, that are presumed to be activated by the stimuli.

[0119] Similarly, for tasks that use visual and / or auditory mental imagery, the region(s) of interest include higher-order sensory and / or memory regions predicted to be (sufficiently) activated by the task. The region(s) of interest include primary sensory regions predicted to be isolated and / or deactivated by the task. The region(s) of interest include content-specific higher-order sensory regions, such as the fusiform face area, predicted to be (re)activated by the task.

[0120] Further, for such mental imagery tasks, the method may include performing a pilot fMRI sequence scan to localize the region(s) of interest and / or to calibrate predicted levels of activation (or deactivation / isolation), where the pilot fMRI run is performed with direct stimulation, e.g., an explicit version of the task as opposed to mental imagery.

[0121] Similarly, in the case of a motor imagery task, activation of motor and / or cerebellar regions (or parts thereof, or a combination thereof) is expected; in the case of a spatial navigation imagery task, activation of the hippocampus (or parts thereof) is expected; in the case of a mental scene construction task, activation of the default mode network (DMN), hippocampus, and / or ventromedial prefrontal cortex (or parts / combinations thereof) is expected; in the case of an emotional scene construction task, emotional content induces the expected activation (or change in activity relative to resting state) in the amygdala (or parts thereof).

[0122] Checking for unexpected activation or deactivation in one or more regions of interest that have no known association with the fMRI task or that have a known association with distraction or contemplation also includes (near) real-time analysis of activity indicative of cognitive fatigue in regions of interest including the anterior cingulate cortex (ACC), basal ganglia, ventromedial prefrontal cortex, portions thereof, and / or combinations thereof.

[0123] The check for unexpected activation or deactivation in regions of interest that have no known association with the fMRI task at hand is also adapted to subject-specific disorders or pathologies. For example, in the case of patients prone to suffer from hallucinations, brain regions activated by such hallucinations (even if not included in the task at hand, either by direct reliance on them or by being triggered through the applied stimuli) are monitored during the examination to detect whether the patient suffers from such a hallucinatory episode. This method thus performs an additional check for sensory or conscious experiences that are characteristic for the disorder at hand. For example, in the case of schizophrenic patients, brain regions involved in hearing voices are monitored during the scan. When such events are detected, they are processed as discussed previously herein and similarly to be discussed below, for example by generating a signal (but not necessarily indistinguishable from the signals in the case of more generally applicable detections, such as, for example, loss of attention) to alert the operator. Such disorder-specific region(s) of interest may, for example, in the case of hallucinatory experiences, include parts of the primary visual / auditory / somatosensory cortex(s) and / or associated higher brain processing (unless there is conflict with expected brain activity due to the task at hand), but may also be more individually defined by their known (and specific) involvement with the patient's disorder or condition.

[0124] Further, optionally, the method also comprises the step 15 of detecting the subject's motion during acquisition of the fMRI time series, e.g. detecting motion in (near) real time. This motion detection is performed by analysis of acquired fMRI images (e.g. the most recently acquired image in the time series). Methods for estimating motion based on, e.g. image registration of multiple images in a time series are well known in the art. For example, for example, at a time T i For each image in the time series collected at i(For example, at time T i-1 It should be noted that the assessment of motion is more robust if multiple previous images are taken into account (even though this is not strictly required), e.g. to assess the progression of motion over time.

[0125] However, other options are not excluded, such as motion detection based on observation by a camera (to name just a few examples) and / or by a tracker, e.g. an adhesive skin tag that can be tracked in a coordinate space with a known spatial relationship to the coordinate system of the fMRI scanner. When excessive motion is detected, e.g. the magnitude of the motion vector (displacement vector) or any component of the motion vector (or more generally, a determined motion parameter, e.g. any element of an affine transformation matrix) exceeds a predefined threshold (different types of parameters are compared to different thresholds to take into account different effects of scale), a signal is generated, e.g. to warn an operator. This is shown as an arrow to step 14 of generating a signal, as discussed below in this specification, but it should be noted that different signals are generated when motion is detected and when a decrease in the subject's attention is detected. However, in a particularly simple embodiment, a simple warning sound or light could be triggered to draw the operator's attention, regardless of whether the detected problem is due to excessive motion or to a loss of the subject's attention. However, a more informative warning would be preferable.

[0126] Furthermore, the step 13 of determining activity measures indicative of brain activity in a given region of interest includes a step of detecting activity measures indicative of a sleep state. However, it should be noted that this is also more broadly applicable and not necessarily limited to a sleep state being detected through an imaging procedure, i.e. (only) after applying stimulation. In particular, when a subject goes to sleep, obviously as a consequence, images acquired after stimulation will no longer represent the brain functions intended to be imaged / visualized, but will also modify the brain activity in images supposedly not affected by the stimulation, e.g. in "resting state blocks", and as a consequence, those images will no longer be a good source of information to define a reference or control state against which to contrast the brain activity of interest.

[0127] However, this also means that a sleep state can be detected in images assigned to an active state condition and also in images assigned to a control or reference ("rest") state condition. In other words, the method also includes a further (optional) step 16 of detecting further activity measures indicative of a sleep state during acquisition of the time series, for example based on the most recently acquired images (or a subset of recently acquired images). The result of this sleep detection is processed to generate a signal, for example an alert signal to an operator. For the sake of simplicity, this is shown as a link to step 14 of generating a signal, but it should be understood that the actions taken upon detection of a sleep state are (or may not be) different from the actions taken upon detection of a distracted mood (for example poor task performance).

[0128] It will be clear that although it is possible to detect a sleep state independently of previously provided stimuli, as already discussed herein, it is not necessarily precluded to limit the detection of a sleep state only to other components of the activity measure (e.g., for simplicity, by focusing the processing performed in real time only on the moment when other activity measures are also determined / calculated, e.g., such that the processing can be focused on the same image being evaluated). For example, a signal generated in response to detection of a sleep state may include providing a stimulus to the subject to wake him / her up (although not limited thereto and not necessarily the same type of stimulus used to cue the task).

[0129] For example, to detect sleep states, MRI signals (e.g., T2 *A change in voxel values ​​(e.g., voxel values ​​of voxels) is detected, for example, the one or more regions of interest include the VLPO or a sub-region thereof. It is noted that detection of a sleep state also includes (additionally or alternatively) detection of sleep onset (e.g., a transient state before actually entering sleep). As another example, the image time series is used to determine intermediate physiological parameters indicative of sleep or sleep onset, such as respiration, heart rate, and / or plethysmogram. It should also be understood that the scanner is controlled to alternate (but not necessarily 1:1 interleaved) between image acquisition for the fMRI image time series and dedicated image or signal acquisition for the specific purpose of determining such related physiological parameters, e.g., respiration and / or heart rate. It is noted that a short transient period of non-wakeful brain and physiological activity may be detected prior to the (actual) sleep state(s), which is referred to as a drowsy state. Thus, this transient state can be advantageously detected before the actual onset of sleep, for example by less than 30 seconds. Accordingly, corrective measures may also be taken in due time, such as generating a signal to bring the condition to the operator's attention and / or gently warning the subject (preferably without startling him / her).

[0130] For example, the method comprises monitoring movement via (near) real-time image registration of fMRI image time series, which is useful information for the operator (and / or automatically correcting and / or discarding images acquired when the subject was moving too much). This movement pattern is further analyzed (e.g., by time-frequency analysis, e.g., by Fourier analysis) to detect breathing patterns, whereas, for example, a slowing of breathing may indicate a sleep state or its onset.

[0131] During the acquisition of the time series, step 16 of detecting further activity measures indicative of a sleep state (states indicative of sleep and / or its onset) is not necessarily limited to detecting this further activity measure based on the acquired MRI image(s). For example, sensors are used to monitor one or more physiological parameters that may be indicative of this sleep state, such as sensors for acquiring (preferably repeatedly during the examination) an electroencephalogram (EEG), an electrocardiogram (ECG), a photoplethysmogram (PPG), a ballistocardiogram (CBG), an electromyogram (EMG) (e.g. of muscles directly or indirectly involved in the respiratory movement) and / or to monitor respiratory and / or cardiac activity (e.g. respiratory rate, heart rate) using a radar (or lidar) system or a camera system for optical and / or infrared vital signs monitoring.

[0132] The method also includes tracking the movement of the pupils of the subject's eye(s), for example using MRI scanner compatible eye tracking methods known in the art. For example, such eye tracking is used to ensure that the subject is paying attention to a given stimulus(s) (which of course are limited to visual) and / or is instead distracted by his / her environment. Lack of eye movement mobility also indicates a loss of attention, for example a subject at risk of falling asleep. When eye tracking indicates a high probability that attention to the task has been lost, a signal is generated, for example to alert an operator, as discussed further below in the present specification in connection with signals generated based on real-time analysis of fMRI data.

[0133] The method 10 further comprises a step 14 of generating a signal, e.g. an alert signal, when the determined activity measures indicate insufficient task performance and / or compliance by the subject. This signal is generated before the end of the acquisition of the fMRI time series (e.g., at least as long as the activity measures indeed indicate a need for an alert) and may result in an action being taken, e.g., by an operator, during further acquisition of the time series. This signal may be considered as a (conditional) output of the method, but is not necessarily an "output" of the system or device performing the method, e.g., the signal may be entirely internal to the system, e.g., activating a subroutine to extend, stop, terminate, and / or resume the fMRI examination. It will also be apparent that the signal may include multiple separate signals of similar or different nature (e.g., a value stored in memory, a message displayed on a screen, an auditory cue, transmission to a computer network, etc.). Such multiple different signals disambiguate between multiple different reasons why a (conditional) output may be triggered, e.g., detected movement, undetected but predicted task-related brain activation, unpredicted brain activation in non-task related brain regions, high probability sleep / sleep onset, etc.

[0134] Methods according to embodiments of the invention provide (near) real-time analysis of MRI images to check if the subject is performing the task properly and / or is distracted, and if necessary, provide feedback to the subject, such as alerting the operator or taking other corrective action such as modifying the imaging protocol (e.g., extending the time series or stopping the examination) and / or automatically informing the subject of the detected poor performance. When alerted, the operator reports or gives feedback to the patient, such as by explaining the task again or in more detail. The operator can also switch to an easier task to perform (if other options exist), repeat the examination, extend the length of the time series to compensate for the poor quality of the response, and / or take other appropriate actions.

[0135] The activity measure may, for example, be indicative of the BOLD response in a brain region(s) of interest, and thus indicative of the activity state of the brain structure in the region(s) of interest, where the relevant voxel(s) value(s), such as, for example but not limited to, the average of the voxel values ​​in the region(s) of interest, is calculated using a corresponding reference (e.g., resting state) T2 * value (or the corresponding average voxel value) higher (or possibly, alternatively, lower, depending on the region of interest and the task being considered). * Indicates the value.

[0136] For example, an alert signal may be provided to an operator via the user interface 124 of the computer system 122. In this manner, the operator's attention may be drawn to the fact that the subject is distracted, asleep, or otherwise not performing the task as intended. The operator may then take action to alleviate this problem before the time series is fully collected. For example, the operator may ask the subject to focus on the task at hand, check the subject's mood, and / or extend the length of the time series to compensate for the poor quality of data collected due to poor task performance.

[0137] Other possible actions include terminating the fMRI test in order to switch to another test that uses, for example, a simpler task that may be easier for the subject to perform.

[0138] The alert signal is not necessarily limited to simply providing an informative prompt to the operator: in methods according to embodiments of the present invention, the alert signal includes a cue to the subject to redirect attention to the task at hand, which is provided to the subject, for example, via the sensory stimulation source 120 of the MRI system 100.

[0139] In yet another example, the signal may include an instruction to the computer system 122 controlling the MRI system 100 to extend the time series (e.g., to include additional time samples after Tn) and / or to terminate or stop acquisition of the time series, e.g., to allow the operator to provide further briefing and / or instructions to the subject being imaged before continuing or resuming the examination.

[0140] The signal also includes annotation information stored (e.g., metadata) with the fMRI data, such that the determined activity measure, e.g., an indicator of the subject's attention as a function of time, can be taken into account when processing the fMRI data after the test is completed. For example, the activity measure, or a value derived therefrom, is used as a covariate, e.g., a covariate in a general linear model applied to the time series, when performing an fMRI analysis of the collected imaging data. Alternatively or additionally, one or more portions of the time series where the annotation information indicates a low level of attention, poor task performance, and / or poor compliance are excluded (e.g., discarded) when performing the fMRI analysis, e.g., the analysis is limited to periods when the task was performed properly.

[0141] Thus, the method includes storing annotation information with the MRI time series after the time series acquisition is completed, indicating the progression of attention and / or compliance with the task during the study (e.g., from which covariates can be constructed). The method also includes calculating and recording a measure of overall attention (or distraction) and / or compliance for the completed time series, such as the percentage of time the subject was (was not) fully paying attention. For example, such a measure of overall attention is useful for planning future fMRI studies of the same type (e.g., task design) and / or for the same subject population (e.g., patients with a particular disorder). If low compliance is observed for the task and / or patient population, the future study may be extended, for example, to collect more data, a note may be made for design changes or replacements, and / or an assessment of the time required for the study may be adjusted to aid in scheduling.

[0142] The signal generating step 14 includes generating a number of signals, e.g. a combination of the examples already given herein. The term "signal" may be interpreted broadly, and a signal or a component of a signal includes, e.g. digital data stored in a memory or transmitted via a temporary or non-transient data carrier, information provided in a user interface, a body signal conveyed via a stimulation system, and / or any other useful form of data representation capable of being interpreted by a computer system (the same or a different computer system from the one where the signal is generated) and / or a human (e.g. an operator or subject) as is evident from the type of information conveyed by the signal.

[0143] For example, as illustrated in Figure 3, sensory stimuli (same or different) are repeatedly presented to a subject during acquisition of a time series of fMRI images (12), and an activity measure is repeatedly determined, e.g., each time based on the most recent (e.g., most recent) image acquired in the time series (13). In other words, the activity measure is updated to determine current (now, substantially real) task performance and / or compliance, and thus alert signals are generated at different points during the fMRI examination where task compliance and / or performance is suspected to be insufficient (14).

[0144] As is known in the art, a sensory stimulus (or a number of different stimuli) are repeated in time blocks (so-called active blocks) that are interleaved with other time blocks (so-called rest or rest blocks) in which either no such sensory stimulus is provided or one or more stimuli of a different type are provided that are intended to evoke a different brain response, e.g., a control or baseline condition.

[0145] In this way, the fMRI time series is divided into interleaved resting and active state periods of the subject, which are facilitated (e.g., synchronized) via the stimulation provided to the subject. It should be noted that the "resting state" corresponds to any suitable control or reference condition that differs from the "active state" in that it allows the brain function of interest to be studied by the contrast between the "active state" and the "resting state". In other words, the terms "active" and "resting" should not be interpreted narrowly. It is clear that the number of different types of blocks (e.g., two for "active" and "resting") can be easily extended if necessary, e.g., several different (sub)tasks and / or reference conditions can be compared with each other. It should also be noted that the stimulation events are not necessarily organized as such blocks, even though this may have some advantages. Alternatively, isolated events can be triggered at different times without being grouped together as blocks, e.g., as commonly referred to as event-related protocols / paradigms.

[0146] Methods according to embodiments of the invention (optionally) comprise performing 20 an fMRI analysis on the acquired time series of MRI images after the time series has been fully acquired using methods known in the art. This fMRI analysis is substantially different from the method or methods used in step 13 for determining activity measures in (near) real time during the fMRI time series acquisition. For example, this fMRI analysis is a voxel-based analysis, e.g., performed without being constrained to one or more specific regions of interest, or to determine brain activation in a region or regions of interest other than those assessed during the real time analysis described above.

[0147] In a second aspect, the present invention relates to a computer program product comprising machine-executable instructions that are executed by a processor, the execution of which causes the processor to perform a method according to an embodiment of the first aspect of the present invention, for example the processor being (or being included in) the exemplary computer system 122 of the exemplary MRI system 100 (see FIG. 1 and the above description).

[0148] In a third aspect, the present invention relates to a computer system, such as the computer system 122 described herein above, for monitoring the performance of a predetermined test task performed by a subject during a functional magnetic resonance imaging examination.

[0149] The computer system comprises an input 131 for receiving, in substantially real time, a time series of functional magnetic resonance images of the subject's brain acquired using a magnetic resonance imaging system, for example, the input being included in a hardware interface 121 of the computer system 122.

[0150] The computer system has a first output 132 for providing at least one sensory stimulus via a sensory stimulus source to the subject during acquisition of the time series to prompt the subject to perform a predetermined test task and / or to induce intended brain activity in accordance with the predetermined test task.

[0151] The computer system comprises a processor 123 for determining, based on images received by the input after providing a sensory stimulus, at least one activity measure indicative of brain activity in at least one predetermined region of interest in the subject's brain imaged by the time series of images, and for generating at least one signal when the at least one determined activity measure is indicative of poor performance of a predetermined test task and / or poor compliance with the predetermined test task.

[0152] The computer system includes a second output 133 for outputting at least one signal. The second output 133 may be included in the hardware interface 121, the user interface 124, the memory 125, or a combination thereof.

[0153] The processor 123 is further configured to generate at least one signal during and / or immediately after completion of the acquisition of the time series, for example within 10 minutes, within 5 minutes, preferably within 1 minute of the acquisition of the time series.

[0154] In a fourth aspect, the present invention relates to a magnetic resonance imaging system comprising a computer system according to an embodiment of the invention, such as, for example, the magnetic resonance imaging system 100 described herein above.

[0155] The computer system and / or magnetic resonance imaging system according to the embodiments is configured in particular to perform the method according to the embodiment of the first aspect of the invention, such that details of other features of the MRI system and / or computer system according to the embodiments of the invention or of the features described herein will become clear in the light of the description provided further herein with respect to the method according to the embodiments of the invention.

Claims

1. 1. A method for monitoring performance of a predetermined test task performed by a subject during a functional magnetic resonance imaging examination, the method comprising: acquiring a time series of functional magnetic resonance images of the subject's brain using a magnetic resonance imaging system; providing at least one sensory stimulus to the subject via a sensory stimulus source during the acquisition of the time series to prompt the subject to perform the predefined test task and / or to induce intended brain activity in accordance with the predefined test task; determining, after the step of providing the sensory stimulus, at least one activity measure indicative of brain activity in at least one predetermined region of interest in the subject's brain imaged by the time series of images; generating at least one signal when at least one determined activity measure indicates insufficient performance of and / or insufficient compliance with the predefined test task, said signal being generated during the acquisition of the time series and / or prior to removing the subject from the magnetic resonance imaging system, said at least one signal comprising instructions to stop, terminate and / or resume the acquisition; The method comprising:

2. The at least one signal is an alert signal for notifying an operator of the magnetic resonance imaging system; Instructing the magnetic resonance imaging system to extend the acquisition; and / or The method of claim 1 , further comprising providing a performance feedback signal to the subject.

3. The at least one signal is and / or annotation information for storing together with the acquired time series of functional magnetic resonance images such that the evolution of the determined at least one activity measure over the duration of the time series can be taken into account when analyzing the time series after the acquisition. Measures of overall attention or distractibility and / or compliance across the time series for use in planning future functional magnetic resonance imaging studies having the same or similar test task design and / or involving the same subject population. The method of claim 1 or 2, comprising:

4. 4. The method of claim 1, wherein a plurality of sensory stimuli are provided to the subject at different time points during the acquisition of the time series, the activity measures are iteratively determined at different time points during the acquisition and in relation to different time points during the acquisition, and the signal is generated when and at a time when a most recently determined activity measure indicates insufficient performance and / or compliance.

5. 5. The method of claim 1 , wherein determining the activity measure comprises comparing at least one first image of the time series acquired during an estimated absence of an effect of the sensory stimulation on the brain with at least one second image of the time series acquired during an estimated presence of a direct or indirect effect of the sensory stimulation on the brain.

6. 6. The method of claim 1, wherein the step of providing the sensory stimulation comprises displaying a visual cue, emitting an auditory cue, providing a tactile sensation, and / or providing another type of sensory stimulation, and the activity measure is determined for at least one region of interest comprising a sensory cortex, a primary visual cortex, an auditory cortex, a somatosensory cortex, portions thereof, and / or combinations thereof, in which neural activation is caused by the sensory stimulation.

7. 7. The method of any one of claims 1 to 6, comprising providing instructions and / or trial or pilot runs of the predefined test task or a related task prior to acquiring the time series, in order to train the subject in performing the predefined test task, test understanding and / or performance of the predefined test task, calibrate the expected magnitude of the activity measure used in determining the activity measure, and / or localize the region or regions of interest for use in determining the activity measure.

8. 8. The method according to claim 1 , further comprising acquiring additional information during the acquisition of the time series via different imaging modalities and / or physiological parameter sensors, the additional information being taken into account in determining the activity measure.

9. 9. The method of claim 1, wherein determining the activity measure comprises performing an attention network activation analysis, detecting activation and / or deactivation in at least one region of interest that is elicited in the predetermined task, and / or detecting activation and / or deactivation in at least one region of interest that has no known association with the predetermined task.

10. 10. The method of claim 9, wherein the at least one region of interest without known association with the predetermined task comprises the anterior cingulate cortex, the basal ganglia, the ventromedial prefrontal cortex, and / or other brain regions in which brain activity is indicative of cognitive fatigue, and / or the at least one region of interest without known association with the predetermined task comprises a disorder-specific and / or pathology-specific brain region in which neural activation is indicative of hallucinations or other episodes of the disorder or pathology.

11. 11. The method of claim 1, comprising detecting further activity measures indicative of sleep state and / or sleepiness during the acquisition of the time series.

12. 12. The method of claim 1, comprising detecting movement of the subject during acquisition of the time series, and generating the signal or a further signal if the detected movement exceeds a predefined threshold.

13. 13. A computer program comprising machine executable instructions for execution by a processor of an fMRI system comprising an MRI system and a sensory stimulation source, the execution of the machine executable instructions causing the processor to perform a method according to any one of claims 1 to 12.

14. 1. A computer system for monitoring performance of a predetermined test task performed by a subject during a functional magnetic resonance imaging examination, the computer system comprising: an input for receiving, in substantially real time, a time series of functional magnetic resonance images of the subject's brain acquired using a magnetic resonance imaging system; a first output for providing at least one sensory stimulus to the subject via a sensory stimulus source during the acquisition of the time series to prompt the subject to perform the predetermined test task and / or to induce intended brain activity in accordance with the predetermined test task; a processor for determining, based on images received by said input after providing said sensory stimuli, at least one activity measure indicative of brain activity in at least one predefined region of interest in the subject's brain imaged by said time series of images, and for generating at least one signal when at least one determined activity measure is indicative of insufficient performance of and / or insufficient compliance with said predefined test task; a second output for outputting said at least one signal; Equipped with The processor generates the signal during the acquisition of the time series and / or immediately after completion of the acquisition of the time series.

15. A magnetic resonance imaging system comprising the computer system of claim 14.

Images