A scanning system and methods of scanning

Abstract

The present invention relates to the provision of a scanning system and methods for scanning a discrete location within a body part of a subject. More particularly, the present invention provides a scanning system to scan a suspected injury in a body part of a subject, such as a brain injury. The scanning system provides a diagnostic scanning device that scans the discrete locations within the body part, and a processer configured to carry out calculating the probability of injury in each of the discrete locations; using a statistical map of the probability of injury within the body part; and the diagnostic scanning device scanning each discrete location in order of decreasing probability determined by the processor to obtain results at each discrete location; and displaying the results on a screen as the scanning continues until a clinical assessment of the injury in the body part can be made.

Description

[0001] A SCANNING SYSTEM AND METHODS OF SCANNING

[0002] FIELD OF THE INVENTION

[0003] The present invention relates to a scanning system and methods of scanning a region of a body part of a subject. More particularly, the present invention provides a scanning system to scan a suspected injury in a body part of a subject, such as a brain injury. More particularly the system provides a magnetic resonance (MR) scanning system.

[0004] BACKGROUND

[0005] For acute injuries where treatment is time-sensitive, it is beneficial to reduce the time taken for diagnosis while measuring the most useful information in the process. The underlying cause of the injury means that certain locations are more likely sites for injury than others, and that injury location can be described by a statistical spatial distribution. For example, ischemic stroke is caused by vascular occlusion, which means that the areas of the brain affected by stroke are most likely to follow cerebral arteries.

[0006] Existing methods and systems to accelerate scanning in MR systems include undersampling of k-space -(an array of numbers representing spatial frequencies in the MR image) and the use of compressed sensing reconstruction. However, these techniques are still focussed on measuring full images across the whole body part, and are only useful for the conventional frequency domain I k-space image acquisition methods. They are also not able to be used to change the acquisition process during scanning - meaning low or lower priority areas are necessarily scanned.

[0007] The inventors have developed a magnetic resonance (MR) based scanning device, described in co-pending PCT application PCT / IB2024 / 058197, which acquires MR signals from locations in real space. The contents of the co-pending PCT application PCT / IB2024 / 058197 is hereby incorporated in its entirety. As described in PCT / IB2024 / 058197 the MR scanning device provides a series of acquisition bands, with unique spatial sensitivity patterns at discrete locations that provide coarse spatial resolution in the radial and angular directions of the volume within the cylindrical bore of the MR enclosure. Each acquisition band is measured individually, providing information about MR contrast parameters of tissue in the acquisition band at a discrete location. This is different from the k-space based methods used for spatial localisation used in conventional scanners because the inventors’ scanner operates in real space. It is an object of the present invention to provide a scanning system to overcome some of these known scanning limitations or to at least provide the public with a useful alternative.

[0008] SUMMARY OF INVENTION

[0009] In a first aspect, there is provided a scanning system for scanning one or more discrete locations in a body part of a subject presenting with an injury, the scanning system including a diagnostic scanning device that scans the one or more discrete locations within the body part and a processer, configured to carry out the following steps: a. the processor calculating the probability of injury in each of the discrete locations; using a statistical map of the probability of injury within the body part; b. the diagnostic scanning device scanning each discrete location in order of decreasing probability determined by the processor to obtain results at each discrete location; and c. the scanning system displaying the results on a screen as the scanning continues until a clinical assessment of the injury in the body part can be made from the data collected from the one or more discrete locations.

[0010] In a second aspect, there is provided a scanning system for scanning one or more discrete locations within a body part of a subject presenting with a suspected injury, the scanning system including a diagnostic scanning device that scans the one or more discrete locations within a body part and a processor, configured to carry out the following steps: a. the processor calculating the priority of each discrete location using a statistical map of the probability of injury within the body part; b. the diagnostic scanning device acquiring initial scan data from the one or more discrete locations; c. the processor analysing the acquired scan data to update the probability of injury in that location; d. the processor updating the relative priority of substantially all the locations within the body part using the acquired scan data; e. the processor determining a next discrete location based on the updated priority; f. the diagnostic scanning device collecting the data from the next discrete location; g. the scanning system displaying the results on a screen as the scanning continues; h. repeating steps c, d, e, f and g until a clinical assessment can be made from the data collected from the discrete locations.

[0011] In a third aspect there is provided a scanning system for scanning a body part with a suspected injury the scanning system including a diagnostic scanning device that scans discrete locations within a body part with one or more adjustable sets of scan parameters, and a processer, configured to carry out the following steps: a. the processor calculating the priority of each discrete location and set of scan parameters using a statistical map of the probability of injury within the body part and the sensitivity of each possible scan to the suspected injury; b. the diagnostic scanning device acquiring initial scan data from a discrete location with an initial set of scan parameters; c. the processor analysing the acquired scan data to update the probability of injury in the discrete location; d. the processor updating the relative priority of substantially all of the discrete locations and sets of scan parameters using the acquired scan data; e. the processor determining the next discrete location and set of scan parameters based on the updated priority; f. the diagnostic scanning device collecting the data from the next discrete location with the next set of scan parameters; g. the scanning system displaying the results as the scanning continues; h. repeating steps c, d, e, f and g until a clinical assessment can be made from the data collected from the data collected at the discrete locations.

[0012] In each of the aspects described above, in one example, the processer calculates the probability of injury in each discrete location using a plurality of statistical maps reflecting different injury types.

[0013] In each of the aspects described above, in one example, the scanning and clinical assessment may be completed in less than about 15 mins.

[0014] In each of the aspects described above, in one example, the scanning and clinical assessment may be completed in less than about 10 mins.

[0015] In each of the aspects described above, in one example, the scanning and clinical assessment may be completed in less than about 5 mins.

[0016] In each of the aspects described above, in one example, the diagnostic scanning device may be a magnetic resonance device. In each of the aspects described above, in one example, the system is remote from a primary care health setting.

[0017] In each of the aspects described above, in one example, the body part is a human cranium.

[0018] In another aspect, there is provided a method of determining an injury in a body part using the scanning system as described above.

[0019] In one example of the aspects described above the injury is an ischemic stroke.

[0020] In one example of the aspects described above, the injury is a tumour.

[0021] In one example of the aspects described above, the injury is a haemorrhage or a haematoma.

[0022] In one example of the aspects described above, the method further includes the step of treating the injury.

[0023] In another aspect there is provided a method of scanning one or more discrete locations within a body part of a subject presenting with an injury, the method including the steps of: a. calculating the probability of injury in each discrete location; using a statistical map of the probability of injury within the body part; b. scanning each discrete location in order of decreasing probability with a diagnostic scanning device; and c. displaying the results as the scanning continues until a clinical assessment can be made from the data collected from the one or more discrete locations.

[0024] In another aspect, there is provided a method of scanning one more discrete locations within a body part of a subject presenting with a suspected injury, the method including the steps of: a. calculating the priority of each scan location using a statistical map of the probability of injury within the body part; b. acquiring initial scan data from the one or more discrete locations with a diagnostic scanning device; c. analysing the acquired scan data to update the probability of injury in that location; d. updating the relative priority of substantially all the discrete locations within the body part region using the acquired scan data; e. determining a next discrete location based on the updated priority; f. collecting the data; g. displaying the results as the scanning continues; h. repeating steps c, d, e, f and g until a clinical assessment can be made from the data collected at from the discrete locations.

[0025] In another aspect there is provided a method of scanning one or more discrete locations within a body part of a subject presenting with a suspected injury including the steps of a. calculating the priority of each discrete location and set of scan parameters using a statistical map of the probability of injury within the body part and the sensitivity of each set of scan parameters to the injury; b. acquiring initial scan data from a discrete location within the body part with a set of scan parameters with a diagnostic scanning device; c. analysing the acquired scan data to update the probability of injury in the body part; d. updating the relative priority of substantially all of the possible discrete locations and sets of scan parameters using the acquired scan data; e. determining the next set of scan parameters based on the updated priority; f. collecting the data with the set of scan parameters; g. displaying the results as the scanning continues; h. repeating steps c, d, e, f and g until a clinical assessment can be made from the data collected from the discrete locations.

[0026] In one example of the method aspects described above a plurality of statistical maps reflecting different injury types are used to calculate the probability of injury.

[0027] In one example of the method aspects described above, the scanning and clinical assessment may be completed in less than about 15 mins.

[0028] In one example of the method aspects described above, the scanning and clinical assessment may be completed in less than about 10 mins.

[0029] In one example of the method aspects described above, the scanning and clinical assessment may be completed in less than about 5 mins.

[0030] In one example of the method aspects described above, the diagnostic scanning device may be a magnetic resonance device.

[0031] In one example of the method aspects described above, the method is performed remote from a primary care health setting.

[0032] In one example of the method aspects described above, the body part is a human cranium. In one example of the method aspects described above the injury is an ischemic stroke.

[0033] In one example of the method aspects described above, the injury is a tumour.

[0034] In one example of the method aspects described above, the injury is a haemorrhage or a haematoma.

[0035] In one example of the method aspects described above, the method further includes the step of treating the injury.

[0036] Further aspects and examples of the disclosure will become apparent from the following disclosure and with reference to the accompanying Figures and Examples.

[0037] DRAWING DESCRIPTION

[0038] One or more examples of the disclosure will be described below with reference to the accompanying Figures, in which:

[0039] Figure 1 is a perspective view of an NMR device shown schematically having a subject’s head located within the bore of the device.

[0040] Figure 2 shows a further perspective view of an NMR device with a cut away section of magnets showing schematically the location and proximity of the radiofrequency coils relative to the magnets and the subject’s head located within the bore of the device.

[0041] Figure 3: shows an xy slice through the centre (z=0 mm) of a full ischemic lesion heat map.

[0042] Figures 4a and 4b show two xy slices through the centre (z=0 mm) of two subtype lesion heat maps: Figure 4a shows the middle cerebral artery (MCA) total heatmap and Figure 4b shows the posterior cerebral artery (PCA) heatmap.

[0043] Figure 5 shows the location of a series of acquisition bands with band numbers obtained from the heat maps.

[0044] Figure 6 shows a flowchart of the system steps in the time optimisation algorithm.

[0045] Figure 7 shows the acquisition bands from Figure 5 labelled in scan sequence order after optimisation using the time optimisation algorithm with the total ischemic stroke heatmap shown in Figure 3.

[0046] Figure 8 shows the heatmap coverage as a function of scan number in either band number order (bandid) or optimised order (optimised). Figure 9 shows a flowchart of the system steps in the time optimisation algorithm with injury subtype classification.

[0047] Figure 10 shows the acquisition bands labelled in scan sequence order after optimising with the PCA and MCA heatmaps shown in Figure 4.

[0048] Figure 11 shows the heatmap coverage as a function of scan number in either coil number order (bandid) or optimised order for PCA and MCA injury subtypes (optimised).

[0049] Figure 12 shows a flowchart of the steps in the adaptive scanning system.

[0050] In this specification, where reference has been made to external sources of information, including patent specifications and other documents, this is generally for the purpose of providing a context for the description of the features described. Unless stated otherwise, reference to such sources of information is not to be construed, in any jurisdiction, as an admission that such sources of information are prior art or form part of the common general knowledge in the art.

[0051] DETAILED DESCRIPTION

[0052] Definitions

[0053] As used herein, the term “bore” in connection with the device means the volume within the imaging or scanning device, in particular, the volume that receives a body part.

[0054] As used herein, the term “about” in connection with a referenced numeric indication means the referenced numeric indication plus or minus up to 10% of that referenced numeric indication. For example, the language “about 30” kgs covers the range of 33 kgs to 27 kgs.

[0055] As used herein the term “and / or” means “and” or “or”, both. As used herein “(s)” following a noun means the plural and / or singular forms of the noun. The term “comprising” as used in this specification means, “including” or “consisting at least in part of”. When interpreting statements in this specification which include that term, the features prefaced by that term in each statement all need to be present, but the other features can also be present. Related terms such as “comprise” and “comprised” are to be interpreted in the same manner. The entire disclosures of all applications, patents and publications, cited above and below, if any, are hereby incorporated by reference.

[0056] As used in this specification, the terms “comprises”, “comprising”, “includes”, and “including” are to be construed as being inclusive and open-ended rather than exclusive. Specifically, when used in this specification, including the claims, the terms “comprises”, “comprising”, “includes”, and “including” and variations thereof mean that the specified features, steps, or components are included. The terms are not to be interpreted to exclude the presence of other features, steps, or components.

[0057] The term “approximately” or “approximate” as used herein, means nearly or near to, or about, or close to. Alternatively, “approximately” or “approximate” means estimated, or inexact.

[0058] The term “substantially” as used herein, means for the most part, or mostly, or essentially, or to a great or significant extent.

[0059] The present disclosure is described below with reference to specific examples. However, other examples than those described are equally possible within the scope of the disclosure. The different features and steps of the disclosure may be combined in other combinations than those described.

[0060] A nuclear magnetic resonance (NMR) system is described in co-pending PCT application PCT / IB2024 / 058197. A magnet array 1, suitable for a nuclear magnetic resonance system is shown in Figure 1. A magnet array 1 is configured and shown as receiving a cranium 2 of a subject. The magnet array includes a framework configured to define a bore into which a body part, such as a cranium of a subject is located, when in use. The framework comprises one or more spaced apart rings 3 and 4 to support a plurality of magnets 5 in an array configured into two or more concentric rings, each ring being spaced apart and secured by a plurality of yokes 6 that position the concentric rings in a spaced apart manner and each yoke 6 supporting one or more magnets to form the magnet array 1. The magnet array 1 is combined with a series of radio frequency (RF) coils 7 arranged between the magnet array and the cranium, when in use to form the nuclear magnetic resonance (NMR) system. The radio frequency coils 7 are best shown in Figure 2.

[0061] Traditional NMR systems use a constant (homogeneous) magnetic field Bo provided by magnet arrays, in this example, but also rely on a homogeneous weaker oscillating field B?. This oscillating field is provided by a series of radiofrequency coils, such as coils 7 of the magnet array, which is used to generate Bi magnetic field in order to excite the spins and detect signals from spins.

[0062] The homogeneous Bo and By field in conventional NMR and MRI systems (combined with pulsed gradients) can excite, detect and spatially localise NMR signals from nuclear spins within the NMR system. In contrast, the NMR system as configured and described herein makes measurements in the inhomogeneous field produced by its Bo magnet array. Spins are on-resonance when their resonant frequency, which is dependent on Bo at their location, falls within the excitation frequency bandwidth of the Bi field. Spins that are off-resonance are substantially not sensitive to the excitation from the Bi field. Therefore, measurements in an inhomogeneous Bo field are limited by the Bi excitation frequency and bandwidth, which limits the coverage region: the region where signals can be excited and detected. However, the spatial sensitivity of the described NMR device has been configured through the design of the magnet array and RF coil array to ensure that the Bo and Bi fields still enable coverage of substantially all the volume of target regions within the device.

[0063] Furthermore, the device is configured to minimise mass and improve the efficiency of the magnet array: the strength of the magnetic field inside the bore per weight of magnet array. It is important to configure the magnet array to maximise the strength of the Bo field, because the strength of the NMR signal is dependent on the field strength. However, the weight of the device is an important consideration for the portability of the device, particularly in pre-hospital or in-field environments. Decreasing the weight is desirable because it makes the device easier to transport and use. By design and optimisation of the number, size, position of the magnets in the magnet array (as discussed below in Example 1), the efficiency of the design is improved while maintaining coverage across the target regions.

[0064] The oscillating magnetic field B? is generated by the current flowing through the RF coil and is substantially perpendicular to Bo. This arrangement has been determined to be a substantially optimised arrangement. It is to be understood that each RF coil is positioned independently of the magnetic pairs and what is illustrated in Figures 1 and 2 is to be understood as a simplification of the overall device.

[0065] The magnetic field coverage across the bore is dependent on the inhomogeneous Bo and B? fields produced by the magnet array and RF coils. The inhomogeneous Bo field forms a controlled gradient across the bore, where spins at different locations experience a range of Bo field strengths, which correspond to a range of resonant frequencies. Making multiple NMR measurements with a range of B? frequencies and bandwidths, that correspond to different locations across the bore, allows the device to excite and detect signals from multiple different regions across the inhomogeneous Bo field, increasing coverage. This controlled gradient or controlled inhomogeneous magnetic field Bo is necessary to achieve the functionality of the NMR system.

[0066] As described in co-pending PCT application PCT / IB2024 / 058197 the inhomogeneity of the Bo field was used to select spins from different radial depths inside the bore. This technique enabled NMR signals to be localised to different radii by changing the B? excitation frequency. The contours of the Bo magnetic field strength reflect the volume of spins that are excited at each excitation frequency and that form the acquisition bands. Examples of these acquisition bands are shown in Figure 5. It is envisaged that in real-world application, the number, frequencies, and bandwidth of the acquisition bands will be set based on the limits of the specific spectrometer and RF acquisition hardware, as well as the specific field strength and gradient generated by the magnet array in each band, and the desired resolution.

[0067] Example 1 - NMR Device Construction

[0068] One example of the device is shown in Figure 1. Its general construction will now be described. The device contained a magnet array comprising a ring magnetised in, a ring magnetised out, and one additional ring in the centre. It stretched 1.1x along the y-axis, and the rings were curved to follow the plane of the target region. It produced a 96-115 mT field in the brain volume. The weight was approximately 30 kg, including a mild steel yoke. It had adequate clearance for the 99th percentile largest adult head. It was combined with 12 RF coils in an array, where the coils were equally spaced around the head.

[0069] Each ring was constructed using 18 magnet blocks, preferably 1.5 tesla (T) neodymium iron boron (NIB) - grade N 42, with blocks 35mm x 35mm x 50mm in dimension. The blocks were magnetised along the 35 mm axis with a 1.3 T remanence field sourced from Shanghai Jin Magnet. The yoke pieces were machined from 8 mm thick plates of mild steel having a length dimension of 300 mm x 35 mm. Mild steel was chosen for its high magnetic permeability, to concentrate the magnetic field inside the array. The magnets and yokes were attached to an aluminium frame using stainless steel fasteners. The radio frequency coils were formed from several turns of enamelled copper wire having a diameter of about 150 mm. The magnet array was housed within a plastic shell (not shown in the Figure) and the RF coils were located on the inside of the shell, close to the bore and body part to be measured. The final weight of this magnet array was about 32 kgs, more preferably less than about 25 kgs.

[0070] In addition to the magnet array and RF coil array, additional hardware is required to acquire the NMR signals from spins in the bore. This includes a spectrometer to execute NMR pulse sequences (for example those supplied by Resonint, Wellington NZ) and RF amplifiers for transmit and for receive (for example, those supplied by TOMCO, Stepney, SA Australia). This hardware is similar to the electronics used in other low-field NMR instruments. The spectrometer handles timing within the pulse sequence, produces RF pulses of varying frequencies, powers and duration and acquires the NMR signals. The spectrometer may operate in a multi-channel mode, acquiring RF signals from all RF coils simultaneously, or operate in a multiplexed mode, where only signals from a subset of the RF coils are acquired simultaneously.

[0071] The scanning system is controlled by a processor with a Xilinx Kria SOC ARM CPU with 4 GB RAM and running Linux. It is connected to the scanning device including the spectrometer to set pulse sequence parameters and execute measurements, and to the RF hardware to change acquisition parameters such as frequency and amplifier gain. The system also provides a User Interface with a touch screen for users to operate the scanning system, including selecting scan sequences, starting and stopping the overall scan process, and displaying the scan data to the user and prompting and alerting the user, as required.

[0072] In general terms, the inventors have established that measurements in real space, such as those used by the NMR device described herein, which scans discrete locations in real space in a body part, provide more options regarding the order in which measurements are completed, which means that specific locations can be focused on or prioritised. This is achieved by changing the order or priority in which the acquisition bands are measured based on prior knowledge of the spatial distribution of previously assessed injuries. The sequence of acquisition bands can also be modified or updated during the scanning process, depending on the results of previous acquisition band measurements, to improve classification accuracy and / or reduce uncertainty.

[0073] Example 2 -Acquisition Method, Identification of Target Region and Coverage

[0074] The location at which the scanning device scans is dependent on the spatial distribution of the Bo and B? fields, so it is desirable to ensure that target regions for the particular indication are inside the coverage region and within the volume of the bore. The inhomogeneous Bo and B? fields preclude the use of conventional MRI imaging techniques and pulse sequences for obtaining and localising NMR signals across the bore.

[0075] Conventional NMR experiments would only detect signals from a small region of the bore and would not provide coverage of the target regions.

[0076] To address this, a method for Magnetic Resonance Imaging in static gradient was introduced. The inhomogeneous Bo field was split into a series of bands with different magnetic field strengths along the gradient and this resulted in a series of resonant frequencies. The different bands reflect different radial depths inside the bore. The thickness of these bands in space was determined by the bandwidth of the RF excitation pulses produced by the spectrometer. A series of NMR measurements with the excitation frequency set to the resonant frequency of each individual band were made, which obtained signals from each band. Combining the results from the different bands increased the coverage across the bore.

[0077] Further spatial localisation is given by the RF coil array. Each coil has a sensitivity pattern, due to the inhomogeneous Bi field it produces. Making a series of measurements using the different RF coils in the array allowed the signals to be localised to the specific sector of the bore.

[0078] The spatial localisation technique may be combined with a range of pulse sequences to enable conventional MRI contrasts to be obtained. This includes Ti-weighted, T2-weighted and Diffusion-weighted imaging. Measurements from the different coil and band combinations allow the intensity of signals from different locations inside the body part to be compared, providing information about the position and size of potential lesions.

[0079] In a coverage simulation study, described in co-pending PCT application PCT / IB2024 / 058197 it was found that each simulated acquisition band reflected the volume within the magnetic field that is excited at a specific B? frequency and bandwidth, and by combining the volume of the bands with different excitation frequencies, the overall coverage of the system was obtained. It is envisaged that in real-world application, the number, frequencies, and bandwidth of the acquisition bands will be set based on the limits of the specific spectrometer and RF acquisition hardware, as well as the specific field strength and gradient generated by the magnet array in each band, and the desired resolution.

[0080] The Bo has a strong radial gradient created by the permanent magnet array. In the device described in Example 1, the range of Bo magnetic field strength is ~20 mT, which is too large to be covered by the bandwidth of a single excitation. This means that the scanner requires multiple measurements operating at different RF frequencies to provide coverage of the whole body part. Depending on the requirements of the scanner and the scan sequence, the pulse sequence may be output from any of the RF coils around the magnet, or from a single large transmit volume coil. Using the surface RF coils to transmit the pulse sequence limits excitation to a sector of the bore, providing another method to sequence the scan. The combination of RF frequency and receive coils provides an acquisition band, which measures signals from one radius and sector of the bore.

[0081] As an initial study, the inventors have produced a series of acquisition band sensitivity maps using the simulated Bo field map of the magnet array, and simulations of the Bi field produced by the RF receive coils. The design of the scanner described in Example 1 has 12 RF coils equally distributed around the circumference of the bore, and 16 frequency bands, covering a range from 4.1 to 4.9 MHz in 50 kHz steps. The bandwidth and number of bands were chosen to match the limits of the scanner hardware, which has a limited number of frequencies it can tune and acquire signals from. The outline of the acquisition bands are shown schematically in Figure 5, labelled with the band number. For clarity, the band outlines are shown removing any overlap between the bands. It is envisaged that in real world application, the precise acquisition band parameters may need to be varied for individual patients.

[0082] Measuring an acquisition band detects signals from all tissues in the sensitive region. The signal can be used to produce an average signal for a given band. Using the known sensitivity pattern, the signals from each band can be combined to produce an image, to be similar to conventional MRI image contrast.

[0083] In addition to modifying the scan location, the scan sequence can be modified to measure different MRI contrast properties by changing the pulse sequence and / or changing the pulse sequence parameters. Changing the pulse sequence or pulse sequence parameters (such as diffusion weighting b, TR, or other parameters commonly used in MRI) to detect different MRI contrasts may provide additional information about the condition of the tissue. This is because the sensitivity of each contrast varies for different injuries - the change in signal due to injury lesions - is due to changes in different processes occurring in tissue. For example, it is well known that ischemic tissue is hyperintense on diffusion weighted scans. Increasing the diffusion-weighting b of a scan will create a greater contrast between ischemic and healthy tissue, making it easier to detect ischemic injury. Similarly in conventional MRI, T2 or FLAIR contrast is sensitive to the presence of blood. Adjusting the scan parameters used by the scan sequence to generate this contrast makes it more sensitive to haemorrhage in an acquisition band. Further, collecting scan data with different contrasts may improve the accuracy of classification of the injury as the signal changes from an injury in one contrast may be different from the signal when measured with a different contrast.

[0084] Alternatively with further processing, the signals from each scan location may be processed with a method to classify the probability of an injury being present in the scan location. Alternatively with further processing, the signals may be processed with a multicomponent method that may be used to classify the tissue. This classification may detect the presence of a different brain injury, based on the change in MR contrast. A simple classification based on the change in one MR contrast, may be suitable. For example ischemia may be detected by an increase in DWI signal over a threshold. A classifier may also compare the signals detected in the corresponding region in the opposite hemisphere. The inventors recognise that the same or similar techniques may be applied to other diagnostic devices that scan discrete locations within a body part. As in the NMR example, these devices have sensitivity patterns that project inside a body part to measure signals from locations. While this present application uses a magnetic resonance device as an example, other medical diagnostic devices, such as ultrasound or Near-Infrared spectroscopy (NIRS) based scanners, also have sensitivity patterns. The sensitivity pattern of these devices can not cover the whole body part in a single scan, so different regions must be scanned sequentially, as with the Magnetic Resonance device described above.

[0085] Example 3 Statistical Heatmap Generation

[0086] Some brain injury locations follow a statistical spatial distribution in the head / cranium, due to anatomical structures, the mechanism of injury and the distribution of vulnerable tissues. This means that lesions from some injuries are more likely to occur in some areas of the body. To obtain this spatial distribution, the inventors have implemented a process to compile imaging data from many individual cases. The individual case data was then turned into a statistical “heat map” or a statistical map of the probability of injury to inform the development of the scan sequence algorithms. The values in the heat map reflect the probability that a given voxel is part of a lesion in the data set. Different injuries or conditions may follow their own spatial distribution patterns, which can be used as further information in the optimisation algorithm.

[0087] The distribution is illustrated well by ischemic stroke - when a blockage occurs in the blood vessels in the brain. Occlusion of a large blood vessel affects tissue in the region normally supplied by it, resulting in a clear ischemic lesion in that territory. This has been demonstrated by researchers in literature, for example, Bonkhoffet al. have shown that the location of stroke lesions in the brain is well correlated with the vascular territory, so that occlusion of different vessels creates lesions in different regions. This results in different heatmaps for different stroke injuries, depending on the affected vessels. With increasing amounts of publicly available imaging data for a range of injuries, the inventors recognise that the same techniques can be used to generate statistical heatmaps reflecting the spatial distribution of other injuries, for example haemorrhage, haematoma, or tumours. These other injuries will have different spatial distributions due to their different causes.

[0088] In MRI this ischemic lesion is most clearly visible using Diffusion- Weighted Imaging, which detects the reduction in self-diffusion associated with ischemic tissue. The magnitude of this reduction can be significant (of the order of 50%) which allows the lesions to be easily segmented by thresholding of DWI images. To study the scanning system and optimisation, a heatmap was generated from images from a stroke neuroimaging study (Titan Neuroscience, Australia). Patient images were normalised to the standard MNI space, and the stroke lesions were manually segmented to identify regions of the brain affected by strokes. The segmented lesions were combined and averaged across all patients to produce a statistical heatmap. A representative xy slice through the ischemic stroke heatmaps is shown in Figure 3. These heatmaps are based on the population of strokes that have been imaged in the MRI study, which are predominantly in the MCA territory. While this may be representative of the population, the inventors note that it is also useful to understand the spatial distribution of less common stroke injury types. There is also benefit from including information about the clinical consequences and urgency of the different stroke types.

[0089] Images were manually classified by an expert into subtypes, reflecting a range of stroke subtypes. A series of heatmaps for the individual stroke subtypes were produced using the same method. As an example, the inventors have looked at the MCA and the PCA territories. A representative xy slice through the heatmaps for these stroke subtypes are shown in Figure 4a (PCA) and Figure 4b (MCA).

[0090] Example 4: Time optimisation of scanning sequence

[0091] As described above, the scan sequence requires multiple steps to cover the whole body part. One approach may be to systematically scan the whole body part following a pattern such as from the outside-in (i.e. , decreasing the excitation frequency). However this is a suboptimal solution as it means time may be wasted scanning areas that are less likely to be affected by an injury.

[0092] The inventors have developed a system and method involving a diagnostic scanning device that scans one or more locations in a body part, and a processor to optimise the scanning sequence to improve the time to detection, based on the prior knowledge in the heat maps described above. In one such method, shown in Figure 6, the scanning sequence is optimised using a “greedy” algorithm, hereinafter referred to as the time optimisation algorithm. With the processor taking the dot product of the acquisition band and the heat map, the probability of injury in each acquisition band I scan location can be calculated and the most acquisition bands most likely to detect a lesion can be found and scanned first. The remaining bands are then prioritised and scanned in decreasing order of injury probability. The scan data can be displayed to a user as it is acquired who could change any treatment or stop the remaining steps of the scan in response. As measuring each step can take several seconds, optimising the sequence of scan locations to scan the highest probability locations first reduces the time to detect a lesion if one is present. It is desirable that the scanning and clinical assessment be completed in less than 15 minutes, more preferably less than about 10 minutes, most preferably less than about 5 mins.

[0093] To test this system, the inventors have implemented the method on a simulation of the scanning process. The heat map of Example 3 and the acquisition bands of Example 2 were aligned and the dot product of each acquisition band with the heat map were computed. This gives the probability of an injury in each acquisition band. The optimised sequence of acquisition bands was obtained by sorting the bands in order of decreasing probability. As a baseline, the sequence of acquisition bands was sorted in order of band number, moving from the centre of the magnet out and clockwise around the bore. After optimisation, the sequence is shown in Figure 7, with the labels indicating the order of scanning. It prioritises scanning locations on the centre right and centre left area of the brain, (B034-B037 and B0163-167 respectively), which are the most probable regions shown in Figure 3.

[0094] Figure 8 shows the fraction of heatmap covered as a function of the number of scans. The line labelled bandid shows the coverage increasing following an in-out pattern at each coil, and the line labelled optimised shows the increase in coverage following the optimised pattern. The optimised sequence increases much faster than the in-out pattern, which indicates it should detect an ischemic stroke more quickly. For a scanning device collecting 4 averages with a TR of 1 second, each measurement step takes approximately 4 seconds. In the simulation, after scanning the first 60 bands in 240 seconds, the scanning system has scanned approximately 60% of the probability covered by the acquisition bands in the heatmap. This means that if a lesion is present in the acquisition bands, there is a higher probability it will be found in the first 4 minutes or less.

[0095] One shortcoming of this system is that as discussed in Example 3, the heat map used for optimisation prioritised the most prevalent subtype of ischemic stroke - i.e. it is highly weighted to MCA stroke lesions due to the high prevalence and spatial volume of these strokes. The inventors recognise that it may be beneficial to aim to quickly detect other subtypes, which can be done by reprocessing the acquisition bands against the subtype heatmaps.

[0096] The subtype optimisation method involves the processor selecting the highest probability bands for each subtype up to an arbitrary coverage threshold (e.g. 30% of the subtype heatmap coverage). The processor then orders the scanning of these high probability bands by the priority of each subtype. The processor and scanner then add any remaining bands in order of probability on the total heat map. This series of system steps is shown in Figure 9. The subtype optimisation method was tested on the PCA and MCA heatmaps shown in Figure 4a and 4b. In this test, the processor and scanner prioritised PCA, followed by MCA subtypes, resulting in the sequence shown in Figure 10. This figure shows that the sequence began in the back left and right regions (B134-137 and B071-073), before moving to the centre left and right regions (B147-150 and B052-055). Figure 11 shows the coverage as a function of the number of scans. The optimised sequence shows a sharp increase in heatmap coverage as the scanner scans the highest priority PCA bands, followed by a plateau as it scans lower priority parts of PCA bands. Coverage then increases as the scanner scans the high priority MCA bands before following the same curve as the total heatmap optimisation shown in Figure 8. The inventors have found that the choice of injury and the threshold affect the rate of coverage increase, allowing the scan sequence to be optimised to detect certain subtypes more quickly. It is desirable that the scanning and clinical assessment be completed in less than 15 minutes, more preferably less than about 10 minutes, most preferably less than about 5 mins.

[0097] It is envisaged that in real world application, the heat map and acquisition band calculations may need to be adjusted to account for inter-patient variability in head and / or brain size. The shape of the patients’ head can be used to estimate the variation in brain size and modify the heat map, such that the acquisition band locations in real space still correspond to locations in the heat map. Further, with prior knowledge of the shape of a patient’s head and the discrete scan locations, scan locations that would be outside of the head can be skipped. This can further shorten scan time by avoiding scanning locations that will not contain tissue or injuries.

[0098] Example 5: Adaptive scanning sequence

[0099] In addition to optimising the scan sequence to minimise scan time, the inventors have developed adaptive systems for optimising the sequence of scan locations. In the adaptive system, the sequence of steps is adapted by the processor during the scanning process, based on measurements on earlier acquisition bands. Following a Bayesian approach, the likelihood of different scan locations containing a lesion can be estimated after each measurement. Based on these likelihoods and prior knowledge of the spatial distribution of different injuries, the next step in the scan sequence can be selected to either confirm or rule out these hypotheses. This process can be repeated after the measurement to select the following acquisition band.

[0100] In addition to modifying the scan location, the scan sequence can be modified to measure different MRI contrast properties by changing the pulse sequence and / or changing the pulse sequence parameters. Changing the pulse sequence or pulse sequence parameters (such as diffusion weighting b, repetition time TR, inversion time Tl, or other parameters commonly used in MRI) to detect different MRI contrasts may provide additional information about the condition of tissue. This is because the sensitivity of each contrast to different injuries - the change in signal due to injury lesions - is due to changes in different processes occurring in tissue. Further, collecting this additional information with different contrasts may improve the accuracy of a classifier as the signal changes from an injury in one contrast may be different when measured with a different contrast.

[0101] An adaptive scanning sequence may also be used to re-scan certain acquisition bands based on the existing measurement results. For methods with low SNR such as MRI, it is common to repeat scans and average the received signals to improve the SNR. However, this has the downside of increasing the total scan time, as all scans are repeated. The inventors have recognised that the real-space method of scanning means that areas with low SNR can be prioritised for repeat measurement, based on the SNR of previous measurements. Similarly, the relative likelihood of alternative hypotheses will be modified by even low SNR scans, so in some cases exact repetition of some measurements may be found to be unnecessary.

[0102] When combined with an automated classifier for injured tissue, the adaptive scan sequencing system can provide further improvements to classification accuracy while optimising total scan time. This version of the optimisation system is shown in Figure 12.

[0103] An important factor of an adaptive system is the selection of an algorithm to be used by the processor to determine the priority of each scan location. One approach is an algorithm to calculate mutual information between the unknown lesion presence and the possible scan location and scan parameter. The algorithm calculates a probability and uncertainty of each location containing a lesion. This probability distribution is initialised using information from the statistical heat maps described above. The scan locations are prioritised by the processor according to the decrease in uncertainty across all the locations. The first scan location is scanned by the scanning device to collect the scan data, and the scan data processed by a classifier. Based on the scan data and the heatmaps, the probability, uncertainty and therefore the relative priority of each scan location is updated by the processor. The scan data and classifier results can be displayed to a user on a screen on the scanning device as it is acquired, who could change any treatment or stop the remaining steps of the scan in response. One example of a classifier is an algorithm to identify the presence of an ischemic lesion in an acquisition band. The inventors envisage that a classifier can be based on the change in diffusion-weighted MR signal, due to the reduction in diffusion that occurs in ischemic lesions. Further, the uncertainty and probability of ischemic lesions in each acquisition band can be calculated after each step in the sequence, allowing the priority of each scan parameter set and acquisition band to be updated in response to the acquired data.

[0104] Advantages

[0105] One advantage of the present invention is to provide a system of scanning to optimise the sequence of measurements for these acquisition bands, informed by the prior knowledge of the spatial distribution of injuries or conditions. While the system may be applied to the MR scanning device described above, it is to be appreciated that it is applicable to other diagnostic devices that detect signals from a sensitive region that is moved during the scanning process, such as ultrasound or NIRS scanners.

[0106] The inventors have found the scan sequence optimisations system can provide several advantages for the MR scanner discussed above. These systems decrease the time to diagnosis by ensuring that the scans target the most likely locations for strokes to occur. Minimising time in the scanner improves the patient experience, as patients are not in the enclosed space of the bore and do not have to stay as still for as long.

[0107] Displaying scan data from the most probable scan locations as it is acquired enables a clinical assessment of the injury in the body part to be completed faster. It also allows users to stop a scan in progress to make treatment decisions more quickly if it is clear that a lesion is present.

[0108] An automated classifier for the signals in each scan location is also beneficial, because it may reduce the requirement for an expert to interpret any acquired data. This makes the scanning system more accessible, allowing it to be deployed in locations or situations remote from a primary care health setting, where an expert is not present or available.

[0109] References

[0110] Anna K. Bonkhoff, Tianbo Xu, Amy Nelson, Robert Gray, Ashwani Jha, Jorge Cardoso, Sebastien Ourselin, Geraint Rees, Hans Rolf Jager, Parashkev Nachev, Reclassifying stroke lesion anatomy, Cortex, Volume 145, 2021.

Claims

Claims1 A scanning system for scanning one or more discrete locations within a body part of a subject presenting with an injury, the scanning system including a diagnostic scanning device that scans the one or more discrete locations within the body part and a processer, configured to carry out the following steps: a. the processor calculating the probability of injury in each of the discrete locations; using a statistical map of the probability of injury within the body part; b. the diagnostic scanning device scanning each discrete location in order of decreasing probability determined by the processor to obtain results at each discrete location; and c. the scanning system displaying the results on a screen as the scanning continues until a clinical assessment of the injury in the body part can be made from the data collected from the one or more discrete locations.2 The scanning system as claimed in claim 1 , wherein the processer calculates the probability of injury in each discrete location using a plurality of statistical maps reflecting different injury types.3 The scanning system as claimed in claim 1 or claim 2, wherein the scanning and clinical assessment is completed in less than about 15 mins.4 The scanning system as claimed in any one of claims 1 to 3, wherein the scanning and clinical assessment is completed in less than about 10 mins.5 The scanning system as claimed in any one of claims 1 to 4, wherein the scanning and clinical assessment is completed in less than about 5 mins.6 The scanning system as claimed in any one of claims 1 to 5, wherein the diagnostic device is a magnetic resonance device.7 The scanning system as claimed in any one of claims 1 to 6, wherein the system is remote from a primary care health setting.8 The scanning system as claimed in any one of claims 1 to 7, wherein the body part is a human cranium.9 A scanning system for scanning one or more discrete locations within a body part of a subject presenting with a suspected injury, the scanning system including a diagnostic scanning device that scans the one or more discrete locations within a body part and a processor, configured to carry out the following steps: a. the processor calculating the priority of each discrete location using a statistical map of the probability of injury within the body part; b. the diagnostic scanning device acquiring initial scan data from the one or more discrete locations; c. the processor analysing the acquired scan data to update the probability of injury in that location;d. the processor updating the relative priority of substantially all the locations within the body part using the acquired scan data; e. the processor determining a next discrete location based on the updated priority; f. the diagnostic scanning device collecting the data from the next discrete location; g. the scanning system displaying the results on a screen as the scanning continues; h. repeating steps c, d, e, f and g until a clinical assessment can be made from the data collected from the discrete locations.10 The scanning system as claimed in claim 9, wherein the processer calculates the probability of injury in each scan location using a plurality of statistical maps reflecting different injury types.11 The scanning system as claimed in claim 9 or claim 10, wherein the scanning and clinical assessment is completed in less than about 15 mins.12 The scanning system as claimed in any one of claims 9 to 11 , wherein the scanning and clinical assessment is completed in less than about 10 mins.13 The scanning system as claimed in any one of claims 9 to 12, wherein the scanning and clinical assessment is completed in less than about 5 mins.14 The scanning system as claimed in any one of claims 9 to 13, wherein the diagnostic scanning device is a magnetic resonance device.15 The scanning system as claimed in any one of claims 9 to 14, wherein the system is remote from a primary care health setting.16 The scanning system as claimed in any one of claims 9 to 15, wherein the body part is a human cranium.17 A scanning system for scanning a body part with a suspected injury the scanning system including a diagnostic scanning device that scans discrete locations within a body part with one or more adjustable sets of scan parameters, and a processer, configured to carry out the following steps: a. the processor calculating the priority of each discrete location and set of scan parameters using a statistical map of the probability of injury within the body part and the sensitivity of each possible scan to the suspected injury; b. the diagnostic scanning device acquiring initial scan data from a discrete location with an initial set of scan parameters; c. the processor analysing the acquired scan data to update the probability of injury in the discrete location; d. the processor updating the relative priority of substantially all of the discrete locations and sets of scan parameters using the acquired scan data; e. the processor determining the next discrete location and set of scan parameters based on the updated priority;f. the diagnostic scanning device collecting the data from the next discrete location with the next set of scan parameters; g. the scanning system displaying the results as the scanning continues; h. repeating steps c, d, e, f and g until a clinical assessment can be made from the data collected from the data collected at the discrete locations.18 The scanning system as claimed in claim 17, wherein the processer calculates the probability of injury in each discrete location using a plurality of statistical maps reflecting different injury types.19 The scanning system as claimed in claim 17 or claim 18, wherein the scanning and clinical assessment is completed in less than about 15 mins.20 The scanning system as claimed in any one of claims 17 to 19, wherein the scanning and clinical assessment is completed in less than about 10 mins.21 The scanning system as claimed in any one of claims 17 to 20, wherein the scanning and clinical assessment is completed in less than about 5 mins.22 The scanning system as claimed in any one of claims 17 to 21, wherein the diagnostic scanning device is a magnetic resonance device.23 The scanning system as claimed in any one of claims 17 to 22, wherein the system is remote from a primary care health setting.24 The scanning system as claimed in any one of claims 17 to 23, wherein the body part is a human cranium.25 A method of determining an injury in a body part using the scanning system as claimed in any one of claims 1 to 24.26 The method as claimed in claim 25, wherein the injury is an ischemic stroke.27 The method as claimed in claim 25, wherein the injury is a tumour.28 The method as claimed in claim 25, wherein the injury is a haemorrhage or haematoma.29 The method as claimed in any one of claims 25 to 28, further including the step of treating the injury.30 A method of scanning one or more discrete locations within a body part of a subject presenting with an injury, the method including the steps of: a. calculating the probability of injury in each discrete location; using a statistical map of the probability of injury within the body part; b. scanning each discrete location in order of decreasing probability with a diagnostic scanning device; and c. displaying the results as the scanning continues until a clinical assessment can be made from the data collected from the one or more discrete locations.31 A method of scanning one more discrete locations within a body part of a subject presenting with a suspected injury, the method including the steps of:a. calculating the priority of each scan location using a statistical map of the probability of injury within the body part; b. acquiring initial scan data from the one or more discrete locations with a diagnostic scanning device; c. analysing the acquired scan data to update the probability of injury in that location; d. updating the relative priority of substantially all the discrete locations within the body part region using the acquired scan data; e. determining a next discrete location based on the updated priority; f. collecting the data; g. displaying the results as the scanning continues; h. repeating steps c, d, e, f and g until a clinical assessment can be made from the data collected at from the discrete locations.32 A method of scanning one or more discrete locations within a body part of a subject presenting with a suspected injury including the steps of a. calculating the priority of each discrete location and set of scan parameters using a statistical map of the probability of injury within the body part and the sensitivity of each set of scan parameters to the injury; b. acquiring initial scan data from a discrete location within the body part with a set of scan parameters with a diagnostic scanning device; c. analysing the acquired scan data to update the probability of injury in the body part; d. updating the relative priority of substantially all of the possible discrete locations and sets of scan parameters using the acquired scan data; e. determining the next set of scan parameters based on the updated priority; f. collecting the data with the set of scan parameters; g. displaying the results as the scanning continues; h. repeating steps c, d, e, f and g until a clinical assessment can be made from the data collected from the discrete locations.33 The method as claimed in any one of claims 30 to 32, wherein a plurality of statistical maps reflecting different injury types are used to calculate the probability of injury.34 The method as claimed in claim 30 to 33, wherein the scanning and clinical assessment is completed in less than about 15 mins.35 The method as claimed in any one of claims 30 to 34, wherein the scanning and clinical assessment is completed in less than about 10 mins.36 The method as claimed in any one of claims 30 to 35, wherein the scanning and clinical assessment is completed in less than about 5 mins.37 The method as claimed in any one of claims 30 to 36, wherein the diagnostic scanning device is a magnetic resonance device.38 The method as claimed in any one of claims 30 to 37, wherein the method is performed remote from a primary care health setting. 39 The method as claimed in any one of claims 30 to 38, wherein the body part is a human cranium.40 The method as claimed in claim 39, wherein the injury is an ischemic stroke.41 The method as claimed in claim 39, wherein the injury is a tumour.42 The method as claimed in claim 39, wherein the injury is a haemorrhage or hae- matoma.43 The method as claimed in any one of claims 40 to 42, further including the step of treating the injury.

Images