The present invention relates to an apparatus and method for measurement of inflammation or permeability of the bowel wall.

The bowel wall has two main functions, it provides a barrier between the content of the bowel and the body, allowing certain substances to pass through into the body whilst blocking others. It works to propel and mix the contents of the bowel. The bowel wall is affected by many gastrointestinal diseases, including coeliac disease, Crohn’s disease, inflammatory bowel disease (IBD), liver cirrhosis, irritable bowel syndrome (IBS), obesity, diabetes and Human Immunodeficiency Viruses (HIV).

The cells in the bowel wall are held together by proteins. If these proteins become damaged or malfunction, which may be caused by inflammation or oedema, the barrier becomes impaired, the permeability of the bowel wall is increased, and the bowel wall is said to be 'leaky'. Once the wall permeability is increased, a process called increased bacterial translocation can occur. In the healthy bowel, the passage of a small amount of bacteria from the bowel content through the wall is a normal immunological process which allows for the contents of the bowel to be screened for harmful substances. If the amount of transfer is increased, then the chance of bacteria spreading throughout the body is increased, leading to an increased risk of bacterial infection. The bowel wall permeability may therefore provide an indicator to the health of the bowel.

At present, two methods are used to measure bowel permeability:

• direct visualisation using confocal laser endoscopy (CLE). This involves placing a flexible endoscope down the oesophagus and into the small bowel where a fluorescent dye is injected, allowing the cell lining of the bowel wall to be visualised and the points where dye leaks can be quantified. The sample area of the bowel wall is limited and the patient must be in deep sedation.

• Lactulose:Mannitol urine test. This involves measuring the amount of two different sized sugars, lactulose and mannitol, that are present in urine. The two sugars are absorbed by the bowel but are poorly metabolised meaning the majority of the sugar is passed straight out of the body in the urine. An increase in the Lactulose:Mannitol ratio (LMR) in the urine would indicate an increase in the permeability of the bowel wall.

However, these methods are invasive and/or are currently unstandardized. CLE requires the use of specialised clinics and facilities and is time consuming for the practitioner.

There has now been devised an improved method and associated apparatus for measurement of inflammation or permeability of the bowel wall, which overcomes or substantially mitigates disadvantages associated with the prior art.

According to a first aspect of the invention, there is provided a method of measuring the inflammation or permeability of a bowel wall comprising: obtaining nuclear magnetic resonance measurements, in the form of an image, of at least a bowel including the bowel wall; isolating a portion of the bowel wall within the image; determining, from the image, a nuclear magnetic resonance measurement for the isolated bowel wall; and calculating the inflammation or permeability of the isolated bowel wall based on its nuclear magnetic resonance measurement.

The method according to the invention is advantageous principally because a measurement of the inflammation or permeability of an isolated bowel wall is calculated, based on its nuclear magnetic resonance measurement, which provides advantages over prior art methods of measuring the permeability of the bowel wall. For example, the method according to the invention is less invasive than some prior art methods and does not use a contrast agent, and may provide a more direct assessment of the bowel wall compared to some prior art methods. Furthermore, the assessment of inflammation or oedema of an isolated bowel wall based on its nuclear magnetic resonance measurement, rather than a measurement of the thickness of the bowel wall, may provide a more accurate result, for example by detecting changes before there is any increase in bowel wall thickness.

The permeability being measured may relate to the passage of substances through the biological tissue, or may relate to the passage of one or more specific substance through the biological tissue. The substance may include bacteria. The permeability may relate to the rate of passage through the biological tissue.

A measurement of inflammation or increased permeability of the bowel wall may be an indication of a problem with the health of the patient, for example that the patient is ill or that the patient has a propensity to other illnesses, such as bacterial translocation.

The nuclear magnetic resonance measurements may be obtained using conventional magnetic resonance imaging (MRI) techniques, and hence the image may be an MRI image.

MRI is based on the magnetisation properties of atomic nuclei. A uniform, external magnetic field is employed to align the protons that are normally randomly oriented within the water nuclei of the biological tissue. This alignment (or magnetization) is next perturbed or disrupted by introduction of an external Radio Frequency (RF) energy. The nuclei return to their resting alignment through various relaxation processes and in so doing emit RF energy. After a certain period following the initial RF is applied, the emitted signals are measured. Fourier transformation is used to convert the frequency information contained in the signal from each location in the imaged plane to corresponding intensity levels, which are then displayed as shades of grey in a matrix arrangement of pixels in the form of an image. By varying the sequence of RF pulses applied & collected, different types of images are created.

Biological tissue can be characterised by two different relaxation times - T1 and T2. T1 (longitudinal relaxation time) is the time constant which determines the rate at which excited protons return to equilibrium. It is a measure of the time taken for spinning protons to realign with the external magnetic field. T2 (transverse relaxation time) is the time constant which determines the rate at which excited protons go out of phase with each other, or lose phase coherence with respect to each other.

Nuclear magnetic resonance may therefore be used to measure T1 and/or T2 measurements of the biological tissue. In presently preferred embodiment, nuclear magnetic resonance imaging is used to measure T2.

The image may be shown to the user and the user may provide input based on the image, such as identification of relevant parts of the image, which may facilitate the isolation of a portion of the bowel wall within the image. Hence, the isolation of a portion of the bowel wall within the image may be semi-automated. In addition, the image may be shown to the user in order to facilitate other diagnostics, and this may occur before, during or at the end of the method according to the invention.

The method of the present invention does not need an NMR contrast agent.

Hence, the biological tissue may be devoid of contrast agent when the nuclear magnetic resonance measurements are being obtained. Many prior art MRI assessments of the bowel rely on the use of gadolinium as a contrast agent. However, the use of gadolinium contrast agents in healthy volunteers has been limited due to an FDA safety alert regarding its use in humans. This restriction may inhibit the ability to use prior art MRI investigative tools to look at differences in bowel wall structure between healthy controls and patients. The method may comprise obtaining images with different NMR properties acquired at different times, with correction for some movement of the body with respect to the images. The correction may be applied to relative displacement between the images with respect to the bowel wall. The correction may be applied to local movement of the biological tissue.

The image may be of the bowel wall for which inflammation or permeability is being measured and surrounding biological tissue. The surrounding biological tissue may consist of an entire body of a human or animal, or part of a body.

The isolation of the bowel wall may comprise creating a mask corresponding to the position of biological tissue within the image and excluding any portion of the image outside of the mask to create an isolated image. The selection of the portion of the image within the mask may comprise selecting a portion of the image with intensity values within a given range or above/below a given threshold.

The isolation of the bowel wall may comprise edge detection to determine one or more boundary of the mask. The mask may be determined by selecting the largest area bounded by the one or more boundary. Two or more masks may be combined to provide a single mask. A mask may be used to select a portion of the image corresponding to the bowel wall. The mask may be configured to exclude one or more of: fat; muscle; bowel content.

The isolation of the biological tissue may comprise creating a first mask of the intestine and intestinal wall. The intestinal wall may be isolated from the intestine and intestinal wall image. The intestinal wall may be located using edge detection. Edge dilation may be performed on the edge-detected image. Discontinuities in the mask may be removed by linking one or more discontinuous portions thereof.

Once a portion of the biological issue within the image has been isolated, a nuclear magnetic resonance measurement is determined, from the image, for the isolated biological tissue. The nuclear magnetic resonance measurement that is determined, from the image, for the isolated bowel wall may correspond to the intensity of a nuclear magnetic resonance signal recorded by the image. The nuclear magnetic resonance measurement may be determined for the entire isolated bowel wall. Alternatively, a nuclear magnetic resonance measurement may be determined for each of a plurality of discrete sampling areas of the isolated bowel wall. Each nuclear magnetic resonance measurement may be an average for the isolated bowel wall, or an average for the bowel wall in the associated sampling area. The sampling areas may be arranged along the bowel wall, for example, such that heterogeneity along the wall may be measured.

The nuclear magnetic resonance measurement for the isolated bowel wall, or for each of a plurality of discrete sampling areas of the isolated bowel wall, may be a non-dimensional measurement, eg the nuclear magnetic resonance measurement may be an intensity of a point or an area, or an average intensity of a plurality of points or areas, but not a dimensional measurement, such as length, width or thickness.

The one or more nuclear magnetic resonance measurement determined for the isolated biological tissue may be a T1 measurement and/or a T2 measurement. In presently preferred embodiments, the T2 measurement is used.

The measurement of the inflammation or permeability of the bowel wall calculated by the present invention may be an indirect measurement, and may be based on a correlation between the nuclear magnetic resonance measurement and the inflammation or permeability of the biological tissue. An algorithm may be used to calculate the measurement of the inflammation or permeability of the isolated biological tissue based on its nuclear magnetic resonance measurement, and the algorithm may reflect the correlation between the nuclear magnetic resonance measurement and the inflammation or permeability of the biological tissue. The correlation and/or algorithm may be determined by experiment. The calculation of the measurement of the inflammation or permeability of the bowel wall may be an estimation. The measurement of the inflammation or permeability of the bowel wall calculated by the present invention may be a measurement on a binary scale, e.g. healthy permeability or unhealthy permeability, or inflammation present or inflammation not present. Alternatively, the measurement of the inflammation or permeability of the bowel wall calculated by the present invention may be on a scale comprises at least three different possible measurements, for example a numerical scale. The numerical scale may correlate with a physical parameter of inflammation or permeability, e.g. rate of transfer through the biological tissue, and/or a diagnostically-relevant indication of inflammation or permeability, e.g. inflammation or permeability indicative of certain conditions.

According to a further aspect of the invention, there is provided apparatus for measuring the inflammation or permeability of a bowel wall comprising: apparatus for obtaining nuclear magnetic resonance measurements, in the form of an image, of at least a bowel including the bowel wall; and a computer processor configured to isolate a portion of the bowel wall within the image, to determine, from the image, a nuclear magnetic resonance measurement for the isolated bowel wall, and to calculate a measurement of the inflammation or permeability of the isolated bowel wall based on its nuclear magnetic resonance measurement.

The apparatus for obtaining nuclear magnetic resonance measurements may be an MRI apparatus. The computer processor may be part of the MRI apparatus, e.g. with software installed to provide the apparatus according to the invention. Alternatively, the computer processor may be provided separately, e.g. in a separate device, such as a computer, e.g. with software installed to provide the apparatus according to the invention. According to a further aspect of the invention, there is provided a computer program stored in a computer readable carrier, the computer program being adapted to be loaded onto a computer processor, such that the computer processor is configured to receive nuclear magnetic resonance measurements, in the form of an image of at least a bowel including the bowel wall, isolate a portion of the bowel wall within the image, determine, from the image, a nuclear magnetic resonance measurement for the isolated bowel wall, and calculate a measurement of the inflammation or permeability of the isolated bowel wall based on its nuclear magnetic resonance measurement.

The present invention may be used with existing nuclear magnetic resonance measurements, in the form of an image of the bowel wall. Hence, according to a further aspect of the invention, there is provided a method of calculating a measurement of the inflammation or permeability of bowel wall, comprising: receiving nuclear magnetic resonance measurements, in the form of an image including the bowel wall; isolating a portion of the biological issue within the image; determining, from the image, a nuclear magnetic resonance measurement for the isolated bowel wall; and calculating a measurement of the inflammation or permeability of the isolated bowel wall based on its nuclear magnetic resonance measurement.

The present method/apparatus provides a quantitative method of determining the inflammation or permeability of the bowel wall. Therefore, the method may be used to diagnose a wide variety of bowel diseases or disorders. The method can be used to analyse large areas of the bowel simultaneously, thereby allowing mapping of the bowel wall functionality.

The method is minimally invasive, typically only requiring the injection of an antispasmodic drug, and does not require the use of sedation of the patient. The method does not require the use of potentially dangerous biomarkers and/or contrast agents.

The method provides a high degree of automation, thereby providing quick, reliable and efficient processing. The method further corrects for movement of the patient, thus allowing a more comfortable experience for the patient. According to a further aspect of the invention, there is provided a method of investigating one or more bowel disorder, such as inflammatory bowel disease (IBD), Crohn’s disease, diverticulitis, coeliac disease, HIV, liver cirrhosis, diabetes, irritable bowel syndrome (IBS) or mechanisms by which increased gut inflammation or permeability can contribute to inflammation or dysfunction of other organs in the body such as involvement of joints, kidneys, lungs, heart, liver or other organs in autoimmune disorders, using a method or apparatus according to the invention, as defined above.

Practicable embodiments of the present will now be described with reference to the accompanying drawings, in which

Figure 1 shows a flowchart of a permeability measurement method; Figures 2a-d show a schematic representation of a first motion correction technique;

Figures 3a-d show a schematic representation of a second motion correction technique;

Figure 4 shows a flowchart of a first process to generate a bowel mask; Figures 5a-d show a sequence of images to remove noise from an image; Figure 6 shows a flowchart of a second process to generate a bowel mask Figures 7a-d show a sequence of images to remove fat from an image; Figures 8a-d show a sequence of images to remove muscle from an image;

Figures 9a-d show a sequence of images to remove visceral fat from an image;

Figure 10 shows a flowchart of a second process to generate a bowel wall mask from the bowel mask

Figures 11a-c show a first sequence of images to generate a bowel wall mask;

Figures 12a-e show a second sequence of images to generate a bowel wall mask;

Figures 13a-c show a third sequence of images to generate a bowel wall mask; Figures 14a-e show a second sequence of images to generate a bowel wall mask;

Figure 15 shows an image identifying a misidentification;

Figure 16 shows a graph of T2 values during a placebo or dose of Indomethacin;

Figure 17a shows a graph of median T2 values versus LMR; and Figure 17b shows a graph of change in T2 values versus change in LMR.

The present disclosure provides a method of identifying a human or animal bowel, and more specifically the wall of the bowel. This allows measurement of the permeability of the bowel wall. Variations in changes of permeability of the bowel wall may indicate the presence of one or more bowel disorder, for example, inflammatory bowel disease (IBD), Crohn’s disease, diverticulitis, coeliac disease, HIV, liver cirrhosis, diabetes, irritable bowel syndrome (IBS) or mechanisms by which increased gut inflammation or permeability can contribute to inflammation or dysfunction of other organs in the body such as involvement of joints, kidneys, lungs, heart, liver or other organs in autoimmune disorders,. In other embodiments, the method may be used in a research environment, for example, to determine the uptake of a pharmaceutical drug or nutrient etc.

The method of performing the technique will now be described with reference to figure 1.

In a first step 4, images of the bowel area are captured using Magnetic Resonance Imaging (MRI). This is achieved using MRI apparatus in a fashion that is customary in the art. The images may capture the entire bowel of a subject (i.e. from the stomach to the anus), or may only capture a portion of the bowel that is relevant to the diagnosis (e.g. the lower bowel).

The subject may be administered an antispasmodic drug to prevent/reduce peristaltic movement of the bowel during capture of the images. Additionally, the subject may hold their breath. This allows clear/consistent images of the bowel. One or more fast T1 weighted image captures are performed (e.g. using the “Fast low angle shot” (FLASH) MRI sequence). T1 (longitudinal relaxation time) is the time constant which determines the rate at which excited protons return to equilibrium. It is a measure of the time taken for spinning protons to realign with the external magnetic field.

One or more fast T2 image acquisitions are performed (e.g. using “Spin-Echo prepared Balanced Turbo Field Echo” (BTFE) MRI sequence). T2 (transverse relaxation time) is the time constant which determines the rate at which excited protons reach equilibrium or go out of phase with each other. It is a measure of the time taken for spinning protons to lose phase coherence among the nuclei spinning perpendicular to the main field.

Other techniques to capture T1 and/or T2 images may be used, as will be known to the person skilled in the art.

A series of images are collected sequentially with the pulse sequences adapted for different T 1 or T2 weighting in a way known to a person skilled in the art. The images may be spaced at one or more predetermined time intervals. Alternatively, images may be captured between each breath of the user. The series may comprise a series of T1 and/or T2 images.

In second step 6, the captured images are sent to and read by a computer system. The computer system may be an integral part of the MRI apparatus, may be connected (e.g. via LAN) to MRI apparatus, or may be remote from the MRI apparatus. Additionally, or alternatively, the MRI apparatus may store one or more captured image for later retrieval and processing. The computer comprises an image processing system. The image processing system may comprise software and/or hardware configured to process images.

In a third step 8, the motion correction is performed on the captured images by the image processing system. This maintains consistency between captured images, ensuring like features (e.g. the bowel wall) remain in position relative to the frame of the image. This ensures the respective images are processed correctly.

A first processing step corrected for gross movement of subject between each of the captured images. The gross movement may be caused by respiration and/or movement of the subject, thus resulting in a translation of the subject relative to a frame (i.e. the scan area) of the image. This results in like features being relatively displaced (e.g. via a translation, scale, shear or rotation) between captured images.

A “similarity” algorithm determines like features between respective images. The algorithm then determines a geometric/affine transformation (e.g. the reverse of translation, scale, shear or rotation displacement) of like features between the images. The transformation is then applied accordingly to one or more image, such that like features are matched.

As shown in figures 2a-d, a first image (figure 2a) and a second image (figure 2b) are captured. The image 20b in figure 2b is translated and distorted relative to the image 20a of figure 2a. The algorithm determines the transformation between image 20a and image 20b. As shown in figures 2c and 2d, the transformation 22 is applied to figure 2c, such that the image 20b now resembles the image 20a of figure 2b.

A second processing step corrects for localised movement within the biological tissue (i.e. movement not caused by gross relative movement between the subject and scan area). This may include peristaltic movement of the bowel and/or matter therein. Such distortions are typically smaller than those produced by gross movement.

An algorithm performs non-linear regression between respective images to counter-act any variations in the images. For example, the algorithm estimates a displacement field, which is configured to minimise the difference in intensity between two images. This displacement field is then applied to one or more of the images, such, that the like features in the images are now similar.

As shown in figures 3a-d, a first image (figure 3a) and a second image (figure 3b) are captured. The image 24b in figure 3b is locally distorted relative to the image 24a of figure 3a (whilst the gross position of the image 24a remains substantially constant). The algorithm determines the transformation between image 22a and image 24b. As shown in figures 3c and 3d, the transformation 26 is applied to image 3c, such that the image 24b now resembles the image 24a of figure 3b.

The motion correction/compensation is applied to the series of images, such that like features are spatially consistent between each image. The motion correction steps may require no human input/intervention, and thus the correction can be performed automatically.

In order to determine the permeability of the bowel wall, the portion of image comprising the wall must be semi automatically identified and/or selected, to avoid including structures not from the bowel wall such as vasculature or tissues adjacent to the bowel wall. Once identified, the appropriate measurements (e.g.

T1 and/or T2 values) can be determined.

In some embodiments, the wall may be identified manually. For example, an operator may manually “trace” over the portion of the image containing the bowel wall. The computer system can then determine the T1 and T2 values in the regions traced by the user accordingly. However, such an identification technique is time consuming and requires manual intervention.

Therefore, according to a preferred embodiment, the identification of the relevant portion of the image is performed automatically (i.e. by the image processing system). As shown in step 10, an isolated image of the bowel wall is generated. This is achieved by creating a “mask” representing the area of the image occupied by the bowel wall. The mask can then be applied to the image, and any portion of the image outside of the areas of the mask are disregarded, leaving only the areas of image occupied by the wall. The T1 and T2 values are then determined from the signal intensities in the image that are defined by the mask.

The mask is created by removing undesired portions of the image. These undesired portions comprise other tissues within the body that are not of interest. These other tissues are removed, thus isolating the bowel wall within the images. The undesired tissues are identified by identifying intensity values associated with the respective tissues. This process is described in more detail with reference to figure 4.

In a first step 28, the images are input into the image processing system (i.e. the computer system). As shown in figure 5a, the image comprises an MRI scan of a bowel. In this example, the brightness of the image corresponds to measured intensity during the MRI scan (i.e. black equals low intensity, white equals high intensity).

High intensity images may be saturated at a given threshold. For example, the images may be saturated at the 95th percentile, such that any signal intensity above the 95th percentile of the maximum signal intensity is set to the intensity at the 95th percentile. This allows for the histogram of signal intensities to be redistributed with less weight on the higher signal intensities.

The image intensities are quantised and/or scaled to a nominal scale, for example, such that the intensities in each respective image have a value of between 0 and 1024. This scaling allows the same threshold values to be used across all data sets despite varying signal intensities between captured images and/or subjects (i.e. the intensity data is normalised).

In a second step 30, the undesired portion of the image is determined. Therefore, any voxels (i.e. a pixel or group of pixels) outside of a certain intensity range and/or above/below a certain threshold are identified accordingly. These voxels can then be disregarded. In a first example, noise is removed from the image. This is achieved by disregarding any intensity values below a certain threshold. Any disregarded intensity values are recorded as having a zero/null value. As shown in figure 5b, any noise values are set to zero. The remaining portions of the image contain only tissues, including the bowel wall.

In a third step 32, edge detection is performed on the resulting image. The edge detection identifies all boundaries in an image. Additionally, this provides the length of each detected boundary.

In a fourth step 34, the largest boundary/bounded area is identified therefore defining a mask 40 (in black) with the remaining area (in white) having a zero/null value and, as shown in figure 5c. The mask 40 therefore captures the portion of the image comprising body tissue (black area) and excludes the portion of the image containing white noise (white area).

In fifth step 36, the mask 40 is combined with the original image 5a (e.g. by multiplying the intensity values thereof), with the area of the image within the mask 40 remaining in the new image 5d, and the area outside of the mask having a zero/null value.

Therefore, in a sixth step 38, the noisy regions of figures 5a are removed to provide figure 5d.

As shown in figure 6, the process is repeated to remove fat. In figure 7a, the noiseless image 5d from the previous imaging process is input into the image processing system.

In figure 7b, any intensity values lying outside a given range/are set to zero. This removes any portion of the image containing fat.

In figure 7c, edge detection is performed to create a mask containing the remaining tissues excluding the fat. The largest bounded area may be identified. In figure 7d, the mask is combined with the original image in figure 7a, thus creating an image excluding the fat. With reference back to figure 6, the same process is repeated for other undesired portions of the image. For example, fat 46, muscle 48 and visceral fat 50 are removed to only leave the portion of the image comprising the bowel. Further undesired tissues/structure may be removed as desired. This is performed iteratively so that each undesired portion is removed sequentially. Flowever, it can be appreciated that one or more undesired tissue, or all the undesired tissues may be removed in a single step.

The intensity ranges/thresholds for the respective undesired portions (e.g. fat, muscle etc.) may be predetermined by experiment, or may be determined on an ad-hoc basis, for example, by the operator manually inspecting portions of the images and determining the intensity values associated with each tissue.

Figures 8a-d show the removal of muscle from the image. Figures 9a-d show the removal of visceral fat. It can be appreciated that the process is substantially the same as previously described.

Figure 9d shows an image where the fat and muscle is removed, thus leaving only the bowel and the bowel wall. This image may then be used to create a mask of the bowel 52.

This process is applied to each image in the time series simultaneously.

In order to determine the permeability of the bowel wall, the bowel wall must then be isolated from its contents. As previously discussed, this may be performed manually, but this is time-consuming, so the processing system is further configured to isolate the bowel wall in steps 54 and 56. This process is summarised in figure 10. In a first step 58, the image containing only the bowel and bowel wall (shown in figure 11a) is input into the imaging processing system.

In a second step 60, edge detection is performed on the image. As shown in figure 9b, the edge detection selects the portion of the image corresponding to the edges of the bowel wall. This produces a binary mask where the edges are given a value of 1 (white) and the rest of the image a value of 0 (black). The required edge detection input thresholds are optimised to reduce the number of false positives in the mask (e.g. vasculature in the surrounding mesentery). This optimisation is performed by visually inspecting the edges produced in a training data set using different thresholds in the edge detection function.

In a third step 62, the edges are dilated (e.g. expanded). As shown in figure 11c, the edge portions are expanded compared with the edge portions in figure 11 b. This effectively fills in the area between the edge portion to include the full bowel wall. The edges are dilated by a predetermined number of voxels.

In a fourth step 64, a mask of the bowel content is created. In a similar process to previously discussed, the threshold intensities of the bowel image (figure 12a) corresponding to the content is selected to produce a mask containing only the bowel content (figure 12b).

In a fifth step 66, the mask is inverted. The content is now a zero/null value (in black), as shown in figure 12c.

In a sixth step 68, the content mask is combined/overlaid with bowel wall mask (figure 11c). This produces a mask of the bowel wall with the bowel content removed, as shown in figure 12d. The bowel wall mask no longer contains portions of any undesired tissues (i.e. fat, muscle, content etc. are removed), and represents only the portion of the image containing the bowel wall.

Again, this process is repeated for each image in the series. In a seventh step 70, the bowel wall masks for each image in the series are combined (i.e. by multiplying the intensity values) to create a single mask, as shown in figure 13a. Due to the motion correction, this creates a mask of the bowel wall voxels that have a consistent spatial location across all images.

In an eighth step 72, any missing voxels are filled in to ensure the bowel walls are continuous. This is performed using a “nearest neighbour” algorithm to determine whether an end portion of the bowel wall mask is adjacent to another end portion of the bowel wall mask. Firstly, all continuous segments of the wall mask are identified. A continuous segment is defined as a voxel connected by its eight nearest neighbours. The end points of these segments are then located. If a first end point is found to be within a predetermined voxel radius of a second end point, the empty voxels between the end points were filled, thus joining the segments. This creates a more connected and continuous mask, as shown in figure 13b. Figure 13c shows the bowel wall mask overlaid with the original MRI image.

In a ninth step 74, a correction is performed to offset unwanted dilation of the bowel wall mask performed in the third step 62. As shown in figure 14a, the overlaid mask includes some portions of the bowel content. This is typically due to unintentional joining of two or more sections of wall close or adjacent to one another during the dilation process. At the junction of two or more walls, the solidity and the eccentricity of the junction was calculated. The solidity measures the density of filled area (i.e. how much of the area bound by the junction is filled) and the eccentricity measures the shape of the junction (i.e. how much it curves). The inventor has found that the bowel wall typically comprises high eccentricity area with low solidity (i.e. arched shaped). Therefore, an area with high eccentricity and high solidity typically do not indicate bowel wall. As such, any areas above a predetermined threshold of eccentricity and solidity respectively are removed from the mask.

In some examples, such a solution is not effective, for example, when a high- density junction area has a plurality of arms extending in several directions. This produces an asterisk like shape, as shown in figure 14b. As shown in figure 14c, the dilation process therefore produces a dense area in the centre of the junction. Such areas have low solidity and eccentricity and therefore are not removed. To counter this, an area of a predetermined size is removed from the centre of the junction, as shown in figure 14d. For example, a 7x7 voxel area is removed. Where applicable, this process is then combined with the removal of areas with high eccentricity and high solidity.

The mask therefore more accurately represents the bowel wall of the image, as shown in figure 14e. A bowel wall mask is generated for each image of the series.

Referring back to figure 1 , a misidentification step 14 is performed. This step removes any misidentified portions of the bowel wall mask. For example, as shown in figure 15, a portion of the bowel wall mask 76 overlies the bladder 78. Therefore, this portion of the mask is removed from the overall mask 76. This step is typically performed manually, due to the easy identification and removal of the misidentifications from the mask.

In other embodiments, the misidentifications may be removed using an image processing system, for example, by using artificial intelligence (Al). The steps of creating the bowel wall mask may be used to train a data set for an Al system. This may allow automated detection of the bowel wall.

Now that an accurate mask of the bowel wall has been generated, in a seventh step 16, the mask is overlaid onto the captured images. The mask therefore indicates which portion of the images corresponds to the bowel wall. This allows the system to sample the characteristics of the bowel wall. The mask may be overlaid on any or all of the images generated by the series (e.g. the T1 and T2 images).

In an eighth step 16, the characteristics of the bowel wall are extracted from the portion of the image corresponding thereto. The T1 and T2 values are therefore determined within the bowel wall. The portion of the image to be analysed is divided into a number of discrete sampling areas (or regions of interest, ROI). The size of each ROI was determined by balancing the processing requirements of using a large number of ROIs versus the effective resolution of each ROI (i.e. a large ROI may be unable to detect any variation in the measured parameter). The T2 values are calculated within each ROI.

The inventor has found that the measured T2 value of the bowel wall corresponds to the permeability of the bowel wall. As the permeability of the bowel wall is linked to inflammatory bowel disease and other bowel diseases, then measurement of the T2 values may provide an effective indicator/diagnosis thereof.

Figure 16 shows a graph of the respective measured T2 values for a set of healthy volunteers first given a placebo and then administered Indomethacin, a drug well- documented to increase the permeability of the bowel wall. As can be seen from the graph, the T2 value is responsive to (i.e. correlates with) the administration of Indomethacin and therefore the permeability of the bowel wall. Indomethacin provocation induced a statistically significant increase in bowel wall T2 compared to placebo, from 0.070 s (standard deviation 0.036 s) to 0.115 s (standard deviation 0.063 s) with a p value of 0.017.

Figure 17a shows a graph of the median of the measured T2 values for each participant in the study versus the Lactulose:Mannitol Ratio (LMR). The LMR is typically indictive of the bowel wall permeability. Similarly, it can be seen that T2 values are correlated with the LMR and therefore with the permeability of the bowel wall. There was a significant positive correlation between LMR and bowel wall T2 (Pearson correlation coefficient 0.68, p <0.01). There was also a significant positive correlation between change in T2 and change in LMR (Pearson correlation coefficient 0.63, p <0.01), as shown in Figure 17b.

All subjects were scanned on a Philips 3T Achieva (N=46) or Ingenia (N=2). Participants lay in the prone position with their arms by their head. After the anatomical locator and motility scans subjects were given two 20 mg doses of intravenous Buscopan separated by 10 minutes. Buscopan is an antispasmodic drug that temporarily reduces peristalsis of the gut wall. The purpose of the Buscopan was to enable T2 acquisitions to be carried out without peristalsis.

Each FLASH acquisition and bTFE acquisition for the T2 measurements was performed in a single breath hold with a 15 second wait time between each acquisition to allow for full recovery of the longitudinal magnetization.

The acquisitions were single slice with the slice placed coronally in the plane where the terminal ilium enters the cecum.

T2 values of a given patient can be compared with reference or other control values, thus giving an indicator as to the health of the bowel walls. Additionally or alternatively, the locations of abnormal T2/permeability values may be determined. This may allow mapping of the variation in the permeability of the bowel, for example, to allow a practitioner to identify areas for a targeted treatment.

The present method/apparatus provides a quantitative method of determining the permeability of the bowel wall. Therefore, the method may be used to investigate a wide variety of bowel diseases or disorders. The method can be used to analyse large areas of the bowel simultaneously, thereby allowing mapping of the bowel wall functionality.

The method is non-invasive and does not require the use of sedation of the patient. The method does not require the use of potentially dangerous biomarkers and/or contrast agents.

The method may be used in MRI apparatus, without significant modification thereto. For example, the method may be provided by merely installing additional software on a conventional system. The method provides a high degree of automation, thereby providing quick, reliable and efficient processing. The method further corrects for movement of the patient, thus allowing a more comfortable experience for the patient.