Method for generating quantitative data from medical imaging data

Abstract

A computer implemented method of generating quantitative parameters for pancreatobiliary ducts and lesions in a visceral organ from volumetric medical imaging scan data wherein the scan shows all, or part of a visceral organ is described. The method comprising: locating the position of at least one lesion in the visceral organ on the medical imaging scan data; segmenting the at least one detected lesion and determining one or more quantitative structural parameters for the at least one segmented lesion; performing segmentation of a volume of the medical imaging scan data to identify tubular ducts in the visceral organ within the volume of the medical imaging scan; for at least one segmented tubular ducts within the volume of the medical imaging scan data, computing at least one set of quantitative structural parameters for at least one location along the length of the tubular duct; and generating at least one set of quantitative structural parameters for the at least one segmented tubular duct. An apparatus for performing the method is also described.

Description

[0001] Method for Generating Quantitative Data from Medical Imaging Data

[0002] Field of the Invention

[0003] This invention relates to the analysis of medical scan imaging data. In particular to the assessment of cystic lesions and pancreatobiliary ducts within a visceral organ, by analysing medical scan imaging data.

[0004] Background

[0005] Cystic lesions, particularly pancreatic cystic lesions, are a common incidental finding during abdominal imaging examinations. The vast majority of cystic lesions will remain asymptomatic and benign for the patient’s lifetime. However, a small proportion of the lesions will become malignant and develop into pancreatic cancer. Given that pancreatic cancer is usually only detected in late stages when prognosis is very poor, the ability to monitor incidentally detected cystic lesions offers an opportunity to diagnose pancreatic cancer either at early stages or ideally even before the lesion has become malignant. This early diagnosis could unlock access to curative, surgical treatments that are not possible at later stages of the disease. The remaining challenge is differentiating cystic lesions that reguire this intervention from their benign counterparts.

[0006] Correctly identifying benign and malignant pancreatic lesions is especially important due to the abundance of lesions detected. More than 2% of the population, and over 20% of the population of over 70-years-old, have asymptomatic pancreatic cystic lesions which leads to over 800,000 cystic lesions being incidentally detected every year in the US. As mentioned previously, these lesions can serve as precursors to pancreatic cancer and hence warrant monitoring. However, the vast majority will remain completely benign and so it is critical to identify the rare high-risk lesions. The confusion surrounding pancreatic cystic lesions leads to significant over-treatment and monitoring which costs an estimated $5.3 billion in the US annually.

[0007] Pancreatic cystic lesions have good image contrast on T2 weighted magnetic resonance imaging (MRI) and so current monitoring programmes rely on non-invasive MRI with magnetic resonance cholangiopancreatography (MRCP) to assess the lesions. Figure 1 shows a schematic of part of a human subject, showing the location of various different organs including parts of the digestive system, the liver, gall bladder and pancreas. MRI scans with MRCP sequences are recommended in guidelines and are commonly used for monitoring a patient’s health. Figure 2 shows an MRI image of this region from a human subject. MRCP is an MRI technique used for imaging of bile and pancreatic ducts. Pancreatic cystic lesions also have high contrast in MRCP images. When monitoring pancreatic cystic lesions, MRCP images are used to: manually assess the sizes of the pancreatic duct and pancreatic lesion and qualitatively assess interaction between lesions and the pancreatic duct.

[0008] A known method exists for quantitatively measuring the bile and pancreatic ducts. This method is described in EP3443533B and also shown in figure 3. As shown in figure 3, the known method includes a standardised protocol for image acquisition, tubular enhancement and segmentation, a quantitative biliary tree model, and then the determination of metrics for a whole tree and individual ducts. The method forms the basis of an existing product called MRCP+. Figure 4(a) is an example of a report showing the results of MRCP analysis for a healthy subject. Figure 4(b) is an example of a report showing the results of MRCP analysis for a subject with chronic pancreatitis. However, this existing method does not measure pancreatic lesions, or their interaction with any ducts.

[0009] Figures 4(a) and 4(b) show MRCP+ models and corresponding reports obtained for a healthy volunteer, and a subject with chronic pancreatitits for quantification of the pancreatic duct. As shown, no strictures or dilatations were detected in the healthy subject, but multiple strictures and dilatations were detected in the subject with chronic pancreatitis.

[0010] Clinical guidelines recommend that radiologists manually measure features on a medical scan image, such as the maximum diameter of the cystic lesion or pancreatic duct and qualitatively assess worrisome features such as mural nodules, connection of the cystic lesion to the pancreatic duct, and changes in the lesion over time. These assessments are then used in a binary way to risk-stratify the lesions.

[0011] As the current clinical assessments are performed manually and rely heavily on subjective features, there is considerable variation between different readers and even the same reader at different timepoints.

[0012] Additionally, these subjective assessments struggle to differentiate benign, pre-malignant, and malignant lesions. The variation in the assessments and difficulties in identifying high-risk cases mean that many patients either get diagnosed at a late stage or are over treated - a majority of patients with pancreatic cystic lesions who are referred for surgery do not have any high-grade dysplasia or malignancy on histology after resection.

[0013] There have been some advances in the risk stratification of lesions by academic consortiums, such as the COMP Cyst consortium, who have developed a method based upon a combination of genetic markers obtained through invasive assessments of the cystic lesions and standard clinical assessments. However, these methods still only include manual, qualitative assessments from imaging data.

[0014] This invention utilises novel measurements of quantitative metrics from medical scan imaging data of a visceral organ. The method may characterise biophysical features of cystic lesions, the interaction of the lesions with the pancreatic duct, common bile duct and the pancreas parenchyma, or other visceral ducts to enhance the stratification of patients current and future risk of malignancy. Using these methods, we aim to improve the accuracy of cystic lesion assessments, decrease the time-taken to perform assessments, and reduce inter-operator variability compared with traditional methods.

[0015] Summary

[0016] There is a method of generating quantitative parameters for pancreatobiliary ducts and lesions in a visceral organ from volumetric medical imaging scan data obtained from a medical imaging scan of all or part of the visceral organ. The method may comprise locating a position of at least one lesion and at least one pancreatobiliary duct in the visceral organ on the volumetric medical imaging scan data.

[0017] The method may comprise performing segmentation of a volume within the medical imaging scan using the volumetric medical imaging scan data to segment the at least one lesion. The method may comprise using the volumetric medical imaging scan data to segment the at least one pancreatobiliary duct. Within the volume may be viscera scanned by the medical imaging scan. Within the volume the viscera may include all or part of the visceral organ.

[0018] The method may include determining at least one structural parameter of the at least one lesion that is segmented. The method may include determining at least one structural parameter for the at least one pancreatobiliary duct that is segmented.

[0019] The pancreatobiliary ducts are typically tubular, and therefore are tubular pancreatobiliary ducts. The pancreatobiliary ducts are involved in the transport of bile and pancreatic juices, specifically bile ducts, pancreatic ducts, and their associated ductules and canialiculi. Therefore, the procedure to segment and identify the pancreatobiliary ducts is a procedure that segments and identifies tubular ducts involved in the transport of bile and pancreatic juices, specifically bile ducts, pancreatic ducts, and their associated ductules and canialiculi, in the visceral organ within the volume of the medical imaging scan.

[0020] The structural parameters of the at least one pancreatobiliary duct that is segmented may comprise a quantitative parameter.

[0021] In a preferred embodiment of the invention, the set of structural parameters or quantitative structural parameters for the at least one segmented pancreatobiliary or tubular duct comprises at least one parameter representative of at least one of: a centre-line position of the at least one segmented pancreatobiliary duct at the at least one location; a centre-line orientation of the at least one segmented pancreatobiliary duct at the at least one location; a radius of the at least one segmented pancreatobiliary duct at the at least one location; a diameter of the at least one segmented pancreatobiliary duct at the at least one location; a cross-sectional orientation of the at least one segmented pancreatobiliary duct at the at least one location; a cross-section profile of the at least one segmented pancreatobiliary duct at the at least one location; an indication of the presence of a branching point for the at least one segmented pancreatobiliary duct at the at least one location; and an orientation of a branching point for the at least one segmented pancreatobiliary duct at the at least one location. Further preferably, the set of quantitative structural parameters for the at least one segmented visceral organ lesion comprises at least one parameter representative of at least one of: a maximum diameter of the segmented lesion; a volume of the segmented lesion; determination of the presence of septa within the segmented lesion; a determination of the shape of the segmented lesion; determination of the content of the segmented lesion.

[0022] The structural parameters of the at least one lesion that is segmented may comprise a quantitative parameter or qualitative parameter or both. In a preferred embodiment of the invention the method further comprising the step of using one of more of the structural parameters or qualitative structural parameters for the at least one segmented lesion and one or more of the structural parameters or quantitative structural parameters for the segmented pancreatobiliary duct to determine a distance between a segmented duct and a segmented lesion.

[0023] Preferably, the method further comprises: partitioning the identified segmented pancreatobiliary duct into a region upstream of the lesion, a region downstream of the lesion, and a region interacting with the lesion, and determining the diameter and variation in diameter of the pancreatobiliary or tubular duct for each of these different regions.

[0024] In an embodiment of the invention, a width profile of at least one of the upstream, downstream and interacting regions of the tubular pancreatobiliary or tubular duct is used to determine additional metrics for the duct.

[0025] Further preferably, the method further comprises detecting patterns of change within the at least one segmented pancreatobiliary duct based on the median diameter estimations along the length of the at least one segmented pancreatobiliary duct. Preferably, the method comprises detecting patterns of change within the at least one segmented pancreatobiliary duct by applying at least one pattern detection window to the median diameter estimation data.

[0026] In a preferred embodiment of the invention the method comprises detecting at least one of: strictures; dilatations; and beading. Further preferably, the at least one lesion is identified based on the maximum intensity protection of voxels that make up the volumetric medical imaging scan data.

[0027] Preferably, the at least one lesion is identified by applying a threshold to all of the medical imaging scan data, and reviewing the volume rendering of every voxel that exceeds the threshold.

[0028] Preferably, wherein the visceral organ is a pancreas.

[0029] In a preferred embodiment of the invention the method further comprising the step of performing tubular enhancement on the volumetric medical imaging scan data to derive at least one tubeness measure for each voxel within the volumetric medical imaging scan data; wherein the tubular enhancement comprises, for each voxel within the volume of the medical imaging scan data, computing at least one metric distinguishing tubular structures from other structures and computing a tubeness measure for the respective voxel based at least partly on the at least one metric.

[0030] Preferably, the method comprises using multi-scale Hessian based tubular pancreatobiliary structure enhancement.

[0031] In a preferred embodiment of the invention, the volumetric medical imaging scan data comprises imaging scan data obtained via a magnetic resonance imaging modality.

[0032] Preferably, the method further comprises determining the total number of pancreatic lesions shown in the volumetric imaging scan data.

[0033] The method may be implemented by a system. The method may include locating the position of at least one lesion in the visceral organ on the medical imaging scan data. The structural parameters of the at least one pancreatobiliary duct that is segmented may comprise a quantitative parameter. The method may include segmenting the at least one detected lesion and determining one or more quantitative structural parameters for the at least one segmented lesion.

[0034] The method may include performing segmentation of a volume of the medical imaging scan data to identify tubular ducts involved in the transport of bile and pancreatic juices, specifically bile ducts, pancreatic ducts, and their associated ductules and canialiculi, in the visceral organ within the volume of the medical imaging scan. The method may include a procedure for at least one of the segmented pancreatobiliary ducts within the volume of the medical imaging scan data, computing at least one set of quantitative structural parameters for at least one location along a length of the segmented pancreatobiliary duct.

[0035] The method may include generating at least one set of quantitative structural parameters for the at least one segmented tubular duct.

[0036] An image processing system may be arranged to generate quantitative parameters for pancreatobiliary ducts and lesions in a visceral organ from volumetric medical imaging scan data wherein the scan shows all or part of a visceral organ; the image processing system comprising at least one processing device arranged to: locate the position of at least one lesion in the visceral organ on the medical imaging scan data; segment the at least one detected lesion and determining one or more quantitative structural parameters for the at least one segmented lesion; perform segmentation of a volume of the medical imaging scan data to identify tubular pancreatobiliary ducts in the visceral organ within the volume of the medical imaging scan; for at least one segmented tubular pancreatobiliary ducts within the volume of the medical imaging scan data, compute at least one set of quantitative structural parameters for at least one location along the length of the tubular pancreatobiliary duct; and generate at least one set of quantitative structural parameters for the at least one segmented pancreatobiliary tubular duct.

[0037] In an example embodiment of the invention the set of quantitative structural parameters for the at least one segmented pancreatobiliary duct comprises at least one parameter representative of at least one of: a centre-line position of the at least one segmented pancreatobiliary duct at the at least one location; a centre-line orientation of the at least one segmented pancreatobiliary duct at the at least one location; a radius of the at least one segmented pancreatobiliary duct at the at least one location; a diameter of the at least one segmented pancreatobiliary duct at the at least one location; a cross-sectional orientation of the at least one segmented pancreatobiliary duct at the at least one location; a cross-section profile of the at least one segmented pancreatobiliary duct at the at least one location; an indication of the presence of a branching point for the at least one segmented pancreatobiliary duct at the at least one location; and an orientation of a branching point for the at least one segmented pancreatobiliary duct at the at least one location.

[0038] Further preferably, the set of quantitative structural parameters for the at least one segmented visceral organ lesion comprises at least one parameter representative of at least one of: a maximum diameter of the segmented lesion; a volume of the segmented lesion; determination of the presence of septa within the segmented lesion; a determination of the shape of the segmented lesion; determination of the content of the segmented lesion.

[0039] In a preferred embodiment of the invention the processing device is further configured to perform the step of using one of more of the qualitative structural parameters for the at least one segmented lesion and one or more of the quantitative structural parameters for the segmented pancreatobiliary duct to determine a distance between a segmented pancreatobiliary duct and a segmented lesion.

[0040] Preferably, the processing device is further configured to perform the steps of: partition the identified segmented pancreatobiliary duct into a region upstream of the lesion, a region downstream of the lesion, and a region interacting with the lesion, and determine the diameter and variation in diameter of the pancreatobiliary duct for each of these different regions.

[0041] Further preferably, a width profile of the different regions of the tubular pancreatobiliary duct is used to determine additional metrics for the duct.

[0042] Preferably, the processing device is further configured to perform the steps of detecting patterns of change within the at least one segmented pancreatobiliary duct based on the median diameter estimations along the length of the at least one segmented duct.

[0043] In a preferred embodiment of the invention, the processing device is further configured to perform the steps of detecting patterns of change within the at least one segmented pancreatobiliary duct by applying at least one pattern detection window to the median diameter estimation data.

[0044] Further preferably, the processing device is further configured to perform the steps of detecting at least one of: strictures; dilatations; and beading.

[0045] Preferably, the processing device is further configured to identify at least one lesion by applying a threshold to all of the medical imaging scan data , and reviewing the volume rendering of every voxel that exceeds the threshold.

[0046] Further preferably, the processing device is further configured to perform the steps: of performing tubular enhancement on the volumetric medical imaging scan data to derive at least one tubeness measure for each voxel within the volumetric medical imaging scan data; wherein the tubular enhancement comprises, for each voxel within the volume of the medical imaging scan data, computing at least one metric distinguishing tubular structures from other structures and computing a tubeness measure for the respective voxel based at least partly on the at least one metric.

[0047] In the preferred embodiment of this invention, any reference to the terms ‘duct’ or ‘pancreatobiliary duct’ refer to ducts involved in the transportation of bile and pancreatic juices, specifically encompassing all bile ducts, pancreatic ducts and their associated ductules and canialiculi.

[0048] Brief Description of the Figures

[0049] Further details, aspects and embodiments of the invention will be described, by way of example only, with reference to the drawings. In the drawings, like reference numbers are used to identify like or functionally similar elements. Elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale.

[0050] Figure 1 is a schematic diagram showing various different human organs (prior art).

[0051] Figure 2 illustrates an example MRCP image of the region containing the bile duct, the pancreas and the gall bladder (prior art). Figure 3 shows the process of performing quantitative imaging of the biliary tree using MRCP+ (prior art); Figure 4(a) is an example of a report showing the results of MRCP analysis for a healthy subject.

[0052] Figure 4(b) is an example of a report showing the results of MRCP analysis for a subject with chronic pancreatitis.

[0053] Figures 5(a)-(c) are examples of medical scan images used in an embodiment of the invention.

[0054] Figures 6(a)-(d) show medical scan images and resulting data obtained at vexatious stage of an embodiment of the method of this invention.

[0055] Figure 7 illustrates a schematic of the measurements between a duct and a cystic lesion according to an embodiment of the invention.

[0056] Figure 8 is a flowchart showing some of the steps in the method of this invention; Figure 9 is a flowchart showing the steps involved in loading the data for processing;. Figure 10 is a flowchart showing the steps involved in segmenting a cystic lesion.

[0057] Figure 11 is a flowchart showing the steps for segmenting a duct in accordance with an embodiment of the invention.

[0058] Figure 12 is a flowchart showing the steps for computing one or more metrics according to an embodiment of the invention.

[0059] Figure 13 is a simplified flowchart of an example of a method of detecting and quantifying strictures and dilatations within a duct structure.

[0060] Figure 14 illustrates a simplified block diagram of an example of an image processing system.

[0061] Detailed Description

[0062] The present invention will now be described with reference to the accompanying drawings in which there is illustrated an example of a method and apparatus for generating quantitative data for biliary tree and pancreatic ductal structures and / or information on cystic or malignant lesions from volumetric medical imaging scan data. However, it will be appreciated that the present invention is not limited to the specific examples herein described and as illustrated in the accompanying drawings.

[0063] Furthermore, because the illustrated embodiments of the present invention may for the most part, be implemented using electronic components and circuits known to those skilled in the art, details will not be explained in any greater detail than that considered necessary as illustrated below, for the understanding and appreciation of the underlying concepts of the present invention and in order not to obfuscate or distract from the teachings of the present invention.

[0064] Preferably, an embodiment of this this invention allows the quantitative measurements of cystic or malignant lesions that are present in visceral organs, and their interaction with the ducts in the visceral organ, from medical scan images. In a preferred embodiment of the invention the medical scan images are MRCP images. Preferably, the visceral organ is the pancreas, but the method may be used for other organs. An embodiment of the invention also allows for combining a range of automated, quantitative measurements to create a risk-score for visceral organ malignancy, which can aid clinical decision making. In an embodiment of the invention, the method enhances the characterization of a duct within a visceral organ, such as a pancreatic duct.

[0065] This invention can be used to predict the current and future malignancy risk of a lesion in a visceral organ based upon one or more quantitative measures of the proximity of a lesion to a duct, the size of a lesion, and a lesion’s impact on one or more ducts in the visceral organ. The invention includes measurements of the visceral organ duct morphology downstream and upstream of the lesion, as well as the morphology of the duct in the region of closest approach to the lesion. Because cystic lesions in the head of the visceral organ may also interact with the common bile duct, in some embodiments of the invention, measurements of the common bile duct morphology are also performed.

[0066] Furthermore, an embodiment of the invention may also incorporate information from additional MRI sequences to characterise both the contents of the cyst and the parenchyma of the visceral organ. These may include T1 mapping (standardised for field strength), proton density fat fraction (PDFF) and diffusion weighted imaging (DWI).

[0067] Figures 5(a)-(c) are maximum intensity projections of MRCP coronal images with pancreatic cystic lesions highlighted on the image at 502, 504 and 506.

[0068] Figure 6 shows some of the steps used to assess a medical scan image of a visceral organ of a patient. Preferably, the organ is the pancreas, and the assessment is in respect of a medical scan image of the pancreatic system. In a preferred embodiment of the invention, the medical scan image is an MRI scan, and MRI data of the patient showing the pancreas is loaded into the software. At a minimum, this data must contain an MRCP scan. Scan data from additional 2D or 3D MRI sequences (e.g. T1 , PDFF or DWI data) may also be loaded in.

[0069] Firstly, in a preferred embodiment of the invention the MRCP scan is a scan which includes the pancreas, and the pancreatic lesion, common bile duct and pancreatic duct. These are identified in the MRCP data, as shown in figure 6(a), where the lesion may be identified on a maximum intensity projection (MIP) of the 3D MRI array from the MRCP data, and figure 6(b) which locates the 3D coordinates. In a preferred embodiment of the invention, these selected regions of the scan are segmented. Segmentation of medical scan images, and of digital images in general, is a well-known process of identifying objects and boundaries within an image and assigning a label to each pixel / voxel in the image such that pixels that share certain characteristics are assigned the same label.

[0070] The simplest method of image segmentation is called ‘thresholding’, where greyscale digital image data is converted into binary image data based on a comparison of the greyscale data for each pixel / voxel to a threshold value. Preferably, the segmentation is performed using a local threshold, computed automatically at locations in the lesions and pancreatobiliary tree, as shown in figure 6(c).

[0071] In a preferred embodiment of the invention, a tubular enhancement algorithm, which labels pixels according to how likely they are to contain a tubular structure, is also incorporated into the segmentation of pancreatic and common bile ducts. In a preferred embodiment of the invention, the method also comprises the step of performing tubular enhancement on the volumetric medical imaging scan data to derive at least one tubeness measure for each voxel within the volumetric medical imaging scan data; wherein the tubular enhancement comprises, for each voxel within the volume of the medical imaging scan data, computing at least one metric distinguishing tubular structures from other structures and computing a tubeness measure for the respective voxel based at least partly on the at least one metric. Further preferably, the method may comprise using multi-scale Hessian based tubular biliary structure enhancement.

[0072] One or more metrics of the lesion morphology (e.g. maximum diameter, volume, sphericity, interaction with a pancreatic duct) are measured directly from the lesion segmentation masks, as shown in figure 6(d). Duct morphology is assessed by first measuring the centrelines of the duct segmentation masks and from these computing the diameter of the ducts at every point along these centrelines. Interpolation-based methods are used to provide sub-pixel scale diameter estimates. From these point-wise diameter estimates, duct morphological metrics (e.g. median diameter, localised regions of change in diameter) are computed.

[0073] If additional MRI parametric maps (e.g. T1 , PDFF) were loaded into the software, the values of these parametric maps are computed within the parenchyma of the visceral organ and, if requested by the user, the lesion. These values can be computed in either user-defined regions of interest, or in automated segmentation masks of the visceral organs within the parametric maps (e.g. via a convolutional neural network).

[0074] Finally, one or more of the calculated metrics and any visualisations of the medical images and segmentations are outputted from the invention in a digital, machine-readable file format (e.g. a json file, PDF, png or other standard formats for the images).

[0075] Figure 7 shows the different measurements that may be carried out according to the method of this invention. Increased diameter of the duct upstream of the lesion (upstream), due to interaction with a lesion may be an indicator of malignancy. In an embodiment of the invention, the interaction between a pancreatic cystic lesion and the main pancreatic duct (MPD) may be indicative of malignancy. The output of this function will enable automatic detection and quantitation of phenomena qualitatively assessed by radiologists such as a cystic lesion causing upstream dilation of the pancreatic duct or common bile duct - a known high-risk feature which could require the lesion to be surgically removed.

[0076] The various metrics that can be obtained by the method of this invention include, but are not limited to: measurements of lesion morphology in the visceral organ (e.g., maximum diameter of the lesion); detection of one or more interactions between lesion and MPD / common bile duct (CBD); measurements of MPD / CBD morphology (median diameter, strictures, dilations); shape of lesions; detection of septa within a lesion; number of septa present within a lesion; number of lesions present (monocystic, oligocystic, polycystic); content of the lesion; changes in the MRI signal detected in pancreas parenchyma between the lesion and the duct ; and changes in the MRI signal detected in remote areas of pancreas parenchyma away from the lesion.

[0077] Figure 8 shows a summary view of the steps of an example embodiment of the method 900 for analysing medical scan images. In a preferred embodiment of the invention the medical scan image is an image obtained via magnetic resonance imaging modality. At step 902 data for analysis is loaded. This is followed by step 904, the segmentation of a lesion in the visceral organ shown in the scan data. The steps for the segmentation of the lesion are explained in more detail in figure 11 below. At step 906, ductal structures in the medical scan image are segmented. This is followed by step 908, potentially mapping the segmentations of the lesion and / or the duct onto other imaging data. At step 910, quantitative metrics for the lesion and / or the duct are computed. At step 912, a risk score is calculated, and at step 914 the results are all saved. In an embodiment of the invention, a method of generating quantitative parameters for ducts and lesions in a visceral organ from volumetric medical imaging scan data wherein the scan shows all, or part of a visceral organ is described. Preferably, the method comprises: locating the position of at least one lesion in the visceral organ on the medical imaging scan data; segmenting the at least one detected lesion and determining one or more quantitative structural parameters for the at least one segmented lesion; performing segmentation of a volume of the medical imaging scan data to identify tubular ducts both within and in the vicinity of the visceral organ within the volume of the medical imaging scan; for at least one segmented tubular ducts within the volume of the medical imaging scan data, computing at least one set of quantitative structural parameters for at least one location along the length of the tubular duct; and generating at least one set of quantitative structural parameters for the at least one segmented tubular duct.

[0078] Contemplated examples of the types of quantitative structural parameters that may be generated by way of the above method include: one or more centre-line positions of the segmented tubular pancreatobiliary structure; centre-line orientation of the segmented tubular pancreatobiliary structure at one or more locations along the length of the segmented tubular pancreatobiliary structure; radius values of the segmented tubular pancreatobiliary structure at one or more locations along the length of the pancreatobiliary structure; diameter values of the segmented tubular pancreatobiliary structure at one or more locations along the length of the pancreatobiliary structure; cross-sectional orientations of the segmented pancreatotubular biliary structure at one or more locations along the length of the pancreatobiliary structure; cross-section profiles (shapes) of the segmented tubular pancreatobiliary structure at one or more locations along the length of the pancreatobiliary structure; indications of the presence of branching points for the segmented tubular pancreatobiliary structure at one or more locations along the length of the pancreatobiliary structure; and orientations of branching points for the segmented tubular pancreatobiliary structure at one or more locations along the length of the pancreatobiliary structure.

[0079] In particular for some example embodiments of the present invention, and as described in greater detail below, the method may comprise determining such quantitative structural parameters at a plurality of locations along the length of each identified tubular pancreatobiliary structure.

[0080] Advantageously, by generating such quantitative duct data and lesion data comprising the generated quantitative structural parameters from volumetric scan data obtained via, say, a magnetic resonance imaging modality, a clinician is provided with enhanced information relating to the visceral organ of a patient that enables the clinician to more accurately assess the pathology of the patient’s visceral organ , and in particular to more accurately assess topological and structural features of one or more ducts and or lesions within the visceral organ, such as the presence of strictures and dilatations in the ducts. In this manner, embodiments of the present invention provide an improved non-invasive technique that enables a clinician to detect changes in a visceral organ , and more generally an improved non-invasive diagnostic technique for providing a clinician with an accurate assessment of interactions between a duct and a lesion in a visceral organ.

[0081] Figure 9 shows the steps 1000 needed to load the medical scan data into a system for execution of the method, this corresponds to step 902 in figure 9. At step 1002 the directory is loaded with cases, at step 1004, the validity of each case in the directory is checked, and at step 1006 valid cases are loaded into the app.

[0082] Figure 10 describes the steps required in the method for a lesion segmentation workflow 1100. At step 1102 a lesion location is identified on the medical scan image. The inputs at this step are preferably a 3D MRI image, stored as a three-dimensional array of voxels (a voxel is a 3D pixel), and the outputs are (row, column, slice) coordinates of any voxel within the lesion within the 3D array. Of course, in an alternative embodiment of the invention, 2D images may also be used.

[0083] In a preferred embodiment of the invention a maximum intensity projection (MIP) of the 3D MRI array is utilised. A MIP is a 2D array of pixels that is calculated by taking the maximum intensity of the 3D image along one direction. When calculating the MIP, a lookup table recording the location of each pixel in the MIP within the 3D MRI image is also calculated. Preferably, at least one lesion is identified based on the maximum intensity protection of voxels that make up the volumetric medical imaging scan data.

[0084] A pixel within the MIP that contains the lesion of interest to measure is identified. From the selected pixel, the 3D location of the lesion voxel is obtained from the lookup table.

[0085] In an alternative embodiment of the invention, 3D MRI data could be utilised alone to locate the lesion or on the medical scan image, and directly identify a voxel within the lesion to measure. Alternatively, a threshold could be applied to the entire image, and a volume rendering of every voxel above this threshold presented to the user - the user could then directly click on the parts of the volume rendering they believe to be lesions. In an embodiment of the invention, at least one lesion is identified by applying a threshold to all of the medical imaging scan data , and reviewing the volume rendering of every voxel that exceeds the threshold.

[0086] The next step is Lesion segmentation with local threshold at step 1104. This step has an input of 3D MRI image, coordinates of any voxel within the lesion (output from step 1102), and the output is a 3D lesion mask: a voxel array of the same size as the input image, with every voxel containing the lesion set to 1 , and all other voxels set to zero.

[0087] In a preferred embodiment of the invention, every pixel in the 3D MRI image within a defined distance of the selected point within the lesion is selected. Using these selected pixels, an intensity threshold separating these pixels into foreground (MRI voxels above the threshold) and background (MRI voxels below the threshold) classes is computed. This may be done, for example, using a method (commonly known as Otsu’s method) which, for a series of thresholds varying from the lowest to the highest image intensity value, calculates the inter-class variance at each of the thresholds and returns the threshold maximizing this variance.

[0088] A 3D mask, in which every voxel in the 3D MRI image above the threshold is set to 1 , and all others are set to zero, is computed. The connected components of this mask are then identified. A connected component is any set of foreground voxels that form a continuous object - i.e. for every pair of voxels in a connected component there must exist a continuous path between them through the voxel array that contains only other foreground voxels. The connected component containing the selected voxel within the lesion is identified, and all other components are removed from the mask. This final mask is returned as the 3D lesion mask.

[0089] However in an alternative embodiment of the invention, step 1104 could be omitted, and the method passes directly to step 1106 to find a threshold or skip to step 1110 to manually outline the lesion with a paintbrush tool. Another valid alternative method would be to use the selected point as a seed point in the 3D image, and from that seed point iteratively grow a lesion mask where at every iteration the neighbouring MRI voxels to the current mask are identified, and any voxels whose intensity is within a certain threshold of those within the current mask is added to the mask. The method would be repeated until either the lesion mask stopped growing, or fixed number of iterations was reached.

[0090] Step 1106 is an optional manual fine tuning of the threshold. The input of this step is preferably, a 3D MRI image, coordinates of any voxel within the lesion (output from step 1102) and, if step 1104 has been run, the 3D cyst mask from this step. In a preferred embodiment of the invention, the user is presented with a visualisation of the 3D MRI image with the current lesion mask overlayed on top of it. The user can manually enter an alternative image threshold value than the one computed in step 1104, in which the same method is applied to re-compute the lesion mask with the manually selected threshold value.

[0091] Step 1110 is an optional step to split the mask from the surrounding structures. In a preferred embodiment of the invention, in this step a user is presented with a visualisation of the 3D MRI image with the current lesion mask (outputted from either step 1104 or 1106) overlayed on top of it. If the lesion is connected to another anatomical structure which has high intensity in the MRI image, for example the pancreatic duct, then the mask may contain both the mask and that connected structure. If this situation arises, the user can split the lesion from the surrounding structure using a watershed segmentation algorithm.

[0092] In this method, a distance transform of the lesion mask is calculated. This returns a 3D voxel array of the same size as the 3D lesion mask, recording an appropriate distance of each voxel to the closest background voxel in the 3D lesion mask. The locations of the local maxima of the distance transform image are then recorded. Using the local maxima as seed points, a series of masks are iteratively grown out from these seed points until neighbouring masks meet each other at ‘ridges’ in the distance map and all voxels in the original foreground region have been labelled. The mask containing the selected point within the lesion is taken as the new lesion mask. If this splitting causes the lesion to also be split into sub-components, in step 1112 any sub-masks contained within the lesion can be recombined.

[0093] In an alternative embodiment of the invention, the separation can also be carried out by iteratively eroding the combined mask containing both the duct and the lesion, until the duct is no longer visible, and then iteratively dilating this eroded mask, which now contains only the lesion, with the same number of iterations that were needed to remove the duct. This alternative approach exploits the fact that the duct is generally thinner than most lesions, and therefore will disappear with fewer erosion iterations than the lesion mask.

[0094] The final step 1114 of method 1100 is possible editing of lesion mask with paintbrush tool (optional). In a preferred embodiment of the invention, in this step of the method, a user is presented with a visualisation of the 3D MRI image with the current lesion mask overlayed on top of it. If the mask does not correctly cover the lesion, even after applying the methods in steps 1106, 1110 and 1112, the user can manually edit the lesion mask using a paintbrush tool. Here the user can scroll through the slices of the 3D MRI image and manually mark any voxels within the lesion, and also manually remove any voxels that are outside the lesion.

[0095] Figure 11 is a flowchart showing the steps required 1200 for segmentation of the ducts, as shown step 906 in figure 9. This is a more detailed explanation of the processes performed in step 906. At step 1202 a duct will be selected, either automatically or by a user, and thresholding performed using local contrast. This is followed by step 1204, where the start and end of a pancreatic duct and / or the start and end of a common bile duct will be selected. Next is step 1206, detecting tubular structures using a measure of tubeness. At step 1208, path searching is performed to move from start to end of a duet to generate a centre line model. Finally at step 1210, the diameter, and change(s) in the diameter of the duct are quantified. In a preferred embodiment of the invention, the set of quantitative structural parameters for the at least one segmented duct comprises at least one parameter representative of at least one of: a centre-line position of the at least one segmented duct at the at least one location; a centre-line orientation of the at least one segmented duct at the at least one location; a radius of the at least one segmented duct at the at least one location; a diameter of the at least one segmented duct at the at least one location; a cross-sectional orientation of the at least one segmented duct at the at least one location; a cross-section profile of the at least one segmented duct at the at least one location; an indication of the presence of a branching point for the at least one segmented duct at the at least one location; and an orientation of a branching point for the at least one segmented duct at the at least one location.

[0096] Figure 13 describes a Workflow 1300 for measuring quantitative metrics concerning the lesions and their interaction with the pancreatic and common bile ducts. The steps of this method are an expansion of step 910, determining the quantitative metrics, in figure 9 above. In a preferred embodiment of the invention, the set of quantitative structural parameters for the at least one segmented visceral organ lesion comprises at least one parameter representative of at least one of: a maximum diameter of the segmented lesion; a volume of the segmented lesion; determination of the presence of septa within the segmented lesion; a determination of the shape of the segmented lesion; determination of the content of the segmented lesion.

[0097] The initial step 1302 is to Calculate lesion dimensions. The input for this step is a 3D lesion segmentation mask as produced from the method shown in figure 11 and described above, as well as 3D MRI image metadata (includes information on voxel dimensions in mm). The output of this step is the maximum diameter of lesion in three viewing planes, as well as the 3D volume of the lesion.

[0098] In a preferred embodiment of the invention, in this step the lesion mask is stored as a 3D voxel array of dimensions Nx x Ny x Nz where every voxel containing a lesion is set to 1 , and all others are set to zero. To calculate the maximum lesion diameter along the first viewing plane, we loop over every slice along the first dimension (i.e. from slice 1 to slice Nx), and for each slice containing at least one voxel set to 1 , we identify the edge voxels of the lesion (i.e. those voxels that are set to 1 , but with at least one neighbouring voxel that is set to zero) and measure the maximum distance between any pair of edge voxels. We repeat this method for the other two viewing planes, looping over the second and third dimensions to get the maximum diameter in the second and third viewing planes, respectively. The volume of the lesion is also recorded.

[0099] In step 1304 we measure distance between lesion and pancreatic and common bile ducts. The input for this step is a 3D lesion segmentation mask, or a 3D segmentation mask, or both containing the pancreatic and common bile ducts (obtained using the same methods outlined in the MRCP+ patent). The output from this step is the closest distance between the lesion and the pancreatic and common bile ducts, as well as the (row, column, slice) coordinates of the closest point on the both these ducts to the lesion. In a preferred embodiment of the invention, one of more of the qualitative structural parameters for the at least one segmented lesion and one or more of the quantitative structural parameters for the segmented duct are used to determine a distance between a segmented duct and a segmented lesion. In a preferred embodiment of the invention, a distance transform of the lesion mask is calculated. This returns a 3D voxel array of the same size as the 3D lesion mask, recording the distance of each background voxel to the closest foreground pixel in the lesion mask. For every foreground voxel within the pancreatic and common bile duct segmentation masks, we look up the corresponding value in the distance transform image and record the foreground voxel with the minimum distance transform value. The location of this voxel, and its corresponding distance transform value, are the closest point on the duct to the lesion, and the minimum distance between the duct and lesion, respectively.

[0100] In cases where there are several duct voxels that are located at the same minimum distance from the lesion, the average location (computed by taking the mean row, column and slice coordinates of each voxel) are computed.

[0101] In cases where there is very severe interaction between the lesion and the duct, the duct may not be modelled in the vicinity of the lesion, causing a gap to be present in the duct mask close to the lesion. In such cases, the proximity from the lesion to the duct is computed using the centreline representation of the duct produced by MRCP+. This is an ordered list of coordinates (x,y,z,d) of points along the centreline of the duct, starting from the most downstream point, with the first three numbers recording the coordinates of the point in the image and the fourth the diameter at that point. Firstly, an interpolation, for example spline, is used to estimate the (x,y,z,d) coordinates of any gaps in the centreline model. Then, the value of the distance transform of the lesion mask is calculated at every point of the interpolated centreline. The centreline point with the minimum distance transform value is the closest centreline point to the lesion. The closest distance from the duct to the lesion is the distance transform value at the closest centreline point to the lesion minus the radius of the duct at this point. In cases where several centreline points are found to have equal minimum distances from the lesion, the median of the centreline points (ordered from the start to the end of the duct) is used as the closest point.

[0102] Alternatively, for every foreground voxel in the duct mask, one could calculate the distance to each foreground voxel in the lesion mask. The closest distance between the lesion and duct would be the minimum over all of these distances, and the voxel in the duct mask with this minimum distance would be the closest point on the duct to the lesion. In Step 1306, we determine connectivity between the lesion and the pancreatic and common bile ducts. In a preferred embodiment of the invention , if the minimum distance (from step 1304) is below a specific threshold, then the two are recorded as connected. Otherwise, they are recorded as not connected.

[0103] If the above case also fails, but the user visually sees that the lesion and the duct appear connected, then they also have the option to manually label them as connected, and the distance from step 1304 is also set to zero.

[0104] In an alternative embodiment of the invention, we detect interaction between the lesion and the duct by first taking the lower of the image thresholds used for either the lesion or the duct, creating a mask where any voxel in the original MRI image greater than this threshold is set to 1 , and all others set to zero. If any connected components in this mask overlap with both the lesion and the duct mask, then the lesion and duct are marked as connected, and the calculated distance in step 1304 is overwritten as zero.

[0105] Another method would be to iteratively apply dilatations to either the lesion mask or the duct segmentation mask, and record how many iterations are needed to make one mask connect to the other. If the number of iterations is below some threshold, the two are considered connected.

[0106] We could further supplement the implemented and alternative methods outlined here by also measuring the difference in pancreatic duct diameter upstream and downstream of the lesion- if the diameter increases from downstream to upstream or if the change in diameter deviates from a precharacterised normal range, this implies there may be interaction (in the absence of any interaction we would expect the diameter of the pancreatic duct to reduce as we move from downstream to upstream). If either the pancreatic duct diameter increases from downstream to upstream, or the output from any of the other methods outlined here indicates the duct is connected to the lesion, we could flag the lesion as interacting with the duct. In a preferred embodiment of the invention, the method also comprises partitioning the identified segmented duct into a region upstream of the lesion, a region downstream of the lesion, and a region interacting with the lesion, and determining the diameter and variation in diameter of the duct for each of these different regions. Further preferably, a width profile of the different regions of the tubular duct is used to determine additional metrics for the duct. In step 1308 we measure the duct dimensions. In a preferred embodiment of the invention, (x,y,z,d) points along the centreline of the duct, starting from the head of the pancreas and ending at the tail are used. The first three coordinates record the position of that point within the image and the fourth is the measured diameter of the duct at that point.

[0107] The coordinate of the closest point on the pancreatic duct to the lesion will most likely lie on the surface of the pancreatic duct. Firstly, find which of the points on the pancreatic duct centreline is closest to this point. This is recorded as the closest centreline point to the cyst. If there was a gap in the duct mask, and the centreline-based method was used to find the closest point, then the closest centreline point from this method is used instead. Every point on the centreline that lies within r mm of the closest centreline point to the lesion is labelled as being in an ‘interacting region’, where r is half the maximum diameter of the lesion calculated in step 1302. Because the lesion centreline points are ordered from head to tail (i.e. downstream to upstream), all centreline points before the interacting region are labelled as downstream, and all those occurring after the interacting region are labelled as upstream. The median duct diameter in the upstream, downstream and interacting regions are returned, as well as the median diameter of the entire duct.

[0108] Figure 13 illustrates a simplified flowchart 1400 of an example of a method of detecting and quantifying patterns of change indicative of at least one of strictures, dilatations and beading within a biliary tree or pancreatic ductal structure. The method of Figure 13 starts at 1410 and moves on to 1420 where the quantitative biliary tree and pancreatic ductal data is loaded. In step 1430 the centreline points and duct with are measured, as described in Figure 11 (steps 1208 and 1210). Each duct is represented as an ordered list of centreline coordinates containing both the spatial location and the duct width at each centreline point. In the illustrated example the extracted tube-width estimations can optionally be smoothed out at step 1440 to remove noise that may give rise to false detections - for example, a Gaussian kernel or local averaging may be applied.

[0109] In step 1450 local minima along the width profile of the ordered centreline points are detected. The prominence, or ‘severity’, of each local minimum is characterised by measuring both the absolute and percentage difference in tube width at the minimum point and that of the neighbouring local maxima. If either or both of these differences exceed a given threshold for a given local minimum, that local minimum is labelled as a stricture.

[0110] In step 1460, local maxima along the width profile of the ordered centreline points are detected. The prominence, or ‘severity’, of each local maximum is characterised by measuring both the absolute and percentage difference in tube width at the maximum point and that of the neighbouring local minima. If either or both of these differences exceed a given threshold for a given local maximum, that local maximum is labelled as a dilatation.

[0111] Beading may also be detected, for example based on detected sequences of strictures and dilatations. In step 1470, quantitative measures of each detected stricture and dilatation, for example the severity defined in steps 1450 and 1460, are calculated. Finally, in 1480 the stricture and dilatation data comprising the location and each quantitative measure are recorded, as well as summary metrics for example the total number of strictures and dilatations and the percentages of the length of each duct comprising strictures and dilatations.

[0112] Figure 14 illustrates a simplified block diagram of an example of an image processing system 1500 that may be adapted in accordance with examples of the present invention. The system 1500 comprises one or more processing devices 1510 arranged to perform various processing functions. In some example embodiments, one or more of the processing devices 1510 may comprise one or more processor cores arranged to execute computer program code. Additionally, / alternatively, at least one of the processing devices may comprise a hardware processing device, such as an application specific integrated circuit (ASIC) device or hardware accelerator module comprising hardware circuity arranged to perform predefined processing of data provided thereto.

[0113] The system 1500 further comprises one or more memory elements 1520. The memory element(s) 1520 may consist of one or more non-transitory computer program products such as, for example, a hard disk, an optical storage device such as a CD-ROM device, a magnetic storage device, a Read Only Memory, ROM, a Programmable Read Only Memory, PROM, an Erasable Programmable Read Only Memory, EPROM, an Electrically Erasable Programmable Read Only Memory, EEPROM, and a Flash memory, etc. The memory element 1520 may additionally / alternatively comprise one or more volatile memory elements such as, for example, Random Access Memory (RAM), cache memory, etc.

[0114] For simplicity and ease of understanding, a single processing device 1510 and a single memory element 1520 will hereinafter be referred to. However, it will be appreciated that such references to a single processing device 1510 or a single memory element 1520 are intended to encompass multiple processing devices 1510 and multiple memory elements 1520 respectively.

[0115] The memory element 1520 may have stored therein executable computer program code to be executed by the processing device 1510. The memory element 1520 may further have stored therein data to be accessed and / or processed by the processing device 1510 when executing computer program code.

[0116] The system 1500 illustrated in Figure 14 further comprises one or more output devices, indicated generally at 1530. Such output devices may comprise, by way of example, a display device, a printer device, a network interface device, etc. The system 1500 illustrated in Figure 14 further comprises one or more user input devices, indicated generally at 1540. Such input devices may include, by way of example, a keyboard, a keypad, a mouse, a touchscreen, etc.

[0117] In accordance with some examples of the present invention, the processing device 1510 is arranged to generate quantitative data for biliary tree and pancreatic ductal structures from volumetric medical imaging scan data, for example in accordance with the various methods hereinbefore described with reference to the accompanying drawings. Accordingly, the processing device 1510 is arranged to: perform segmentation of a volume of the medical imaging scan data to identify tubular duct structures within the volume of the medical imaging scan data; compute a set of quantitative structural parameters for each location along the lengths of identified tubular biliary and pancreatic structures; and output quantitative biliary tree and pancreatic data comprising the sets of quantitative structural parameters for the identified tubular biliary and pancreatic structures.

[0118] For example, the processing device 1510 may be arranged to execute computer program code stored within the memory element 1520 for performing at least some of the steps of the various methods hereinbefore described with reference to the accompanying drawings.

[0119] As illustrated in Figure 14, the image processing system 1500 may be operably coupled to at least one data storage device 1550 from which the processing device 1510 is able to retrieve imaging scan data, and to which the processing device 1510 may be able to output the quantitative biliary tree and pancreatic data comprising the set of quantitative structural parameters for the segmented tubular biliary and pancreaticstructures.

[0120] It is further contemplated that the various values, metrics, measures and other data generated by the image processing system 1500 in generating the quantitative biliary tree and pancreatic data may also be output to and stored within the data storage device 1550, either collectively or separately, such data including, by way of example only:

[0121] Hessian volumes;

[0122] Eigenvectors;

[0123] Ra ratios;

[0124] Rb ratios;

[0125] Intensity metrics S;

[0126] Tubular measures (To / To(o));

[0127] Scales (o);

[0128] Segmented (binarized) volumes

[0129] Centre-line points;

[0130] Tube-width estimations;

[0131] Branch nodes;

[0132] Medialness values; and

[0133] Stricture and / or dilatation data.

[0134] As discussed with reference to the preferred embodiments of the invention. This invention allows for the following features:

[0135] Lesion segmentation, detection of interaction of the lesion with the main pancreatic duct and the common bile duct;

[0136] Measurements of pancreatic lesion morphology (e.g., maximum diameter, volume, sphericity); Measurements of pancreatic duct and common bile duct morphology (e.g., median diameter, localised regions of change in diameter); Quantification of interaction between pancreatic cystic lesion and ducts; Interaction of lesions with pancreatic duct / CBD;

[0137] Detecting closest point of lesion and duct;

[0138] Measurement of distance between lesion and pancreatic duct / CBD;

[0139] Split pancreatic duct / CBD into upstream, downstream and interacting; Detect changes in pancreatic duct morphology;

[0140] Detect branch duct connecting to lesions; and

[0141] Quantify mural nodules / solid components of lesions.

[0142] Characterisation of pancreatic parenchyma is also possible using the method of this invention. This will provide the following measurements or organ analysis, including:

[0143] Characterisation of region between lesion and pancreatic duct through multiparametric MRI;

[0144] - Assessment of pancreatic fat and inflammation through PDFF and T1 in the parenchyma; and

[0145] Regional change in pancreatic volume and regional changes in T1 and PDFF.

[0146] Characterisation of lesions , is also possible with the method of this invention. This allows for:

[0147] Characterisation of shape;

[0148] Maximum diameter;

[0149] Volume;

[0150] Detection of septa within a lesion, determination of the number of septa present within a lesion; and

[0151] The number of lesions present (including categorisation of whether monocystic, oligocystic, or polycystic depending on cystic lesion presence).

[0152] Determination of cyst content through changes in MRI contrast including T 1 and proton density fat fraction [PDFF]), can also result from the method of this invention. These include one or more of: changes in the MRI signal detected in pancreas parenchyma between the lesion and the duct (T 1 , PDFF, diffusion weighted imaging [DWI] signals. Longitudinal changes predictive of malignancy, can also result from the method of this invention. These may include one or more of:

[0153] Changes in any of the above metrics indicative of increasing risk;

[0154] Coregistration of images from the same patient at multiple timepoints to match lesions over time.

[0155] As discussed above, this invention provides novel automated measurements of quantitative metrics that characterise cystic lesions and their interaction with the pancreatic duct, CBD and pancreas parenchyma. These provide for more robust measurements that reduce variability between clinicians; improved risk prediction for pancreatic cancer; and time saving due to automation.

[0156] Changes may be assessed in the MRI signal detected in remote areas of pancreas parenchyma away from the cystic lesion including T1 , PDFF, and DWI signals.

[0157] There is potential for this method to be applied in broader assessments, including other modalities (e.g. computed tomography) or other organs (e.g. characterisation of kidney cysts).

[0158] As described above, the invention may be implemented in a computer program for running on an image processing system, at least including code portions for performing steps of a method according to the invention when run on a programmable apparatus, such as an image processing system or enabling a programmable apparatus to perform functions of a device or system according to the invention.

[0159] A computer program is a list of instructions such as a particular application program and / or an operating system. The computer program may for instance include one or more of: a subroutine, a function, a procedure, an object method, an object implementation, an executable application, an applet, a servlet, a source code, an object code, a shared library / dynamic load library and / or other sequence of instructions designed for execution on a computer system.

[0160] The computer program may be stored internally on a tangible and non-transitory computer readable storage medium or transmitted to the computer system via a computer readable transmission medium. All or some of the computer program may be provided on computer readable media permanently, removably or remotely coupled to an information processing system. The tangible and non-transitory computer readable media may include, for example and without limitation, any number of the following: magnetic storage media including disk and tape storage media; optical storage media such as compact disk media (e.g., CD ROM, CD R, etc.) and digital video disk storage media; non-volatile memory storage media including semiconductor-based memory units such as FLASH memory, EEPROM, EPROM, ROM; ferromagnetic digital memories; MRAM; volatile storage media including registers, buffers or caches, main memory, RAM, etc.

[0161] A computer process typically includes an executing (running) program or portion of a program, current program values and state information, and the resources used by the operating system to manage the execution of the process. An operating system (OS) is the software that manages the sharing of the resources of a computer and provides programmers with an interface used to access those resources. An operating system processes system data and user input responds by allocating and managing tasks and internal system resources as a service to users and programs of the system.

[0162] In the foregoing specification, the invention has been described with reference to specific examples of embodiments of the invention. It will, however, be evident that various modifications and changes may be made therein without departing from the scope of the invention as set forth in the appended claims and that the claims are not limited to the specific examples described above.

[0163] Those skilled in the art will recognize that the boundaries between logic blocks are merely illustrative and that alternative embodiments may merge logic blocks or circuit elements or impose an alternate decomposition of functionality upon various logic blocks or circuit elements. Thus, it is to be understood that the architectures depicted herein are merely exemplary, and that in fact many other architectures can be implemented which achieve the same functionality.

[0164] Any arrangement of components to achieve the same functionality is effectively ‘associated’ such that the desired functionality is achieved. Hence, any two components herein combined to achieve a particular functionality can be seen as ‘associated with’ each other such that the desired functionality is achieved, irrespective of architectures or intermediary components. Likewise, any two components so associated can also be viewed as being ‘operably connected,’ or ‘operably coupled,’ to each other to achieve the desired functionality.

[0165] Furthermore, those skilled in the art will recognize that boundaries between the above described operations are merely illustrative. The multiple operations may be combined into a single operation, a single operation may be distributed in additional operations and operations may be executed at least partially overlapping in time. Moreover, alternative embodiments may include multiple instances of a particular operation, and the order of operations may be altered in various other embodiments.

[0166] However, other modifications, variations and alternatives are also possible. The specifications and drawings are, accordingly, to be regarded in an illustrative rather than in a restrictive sense.

[0167] In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word ‘comprising’ does not exclude the presence of other elements or steps than those listed in a claim. Furthermore, the terms ‘a’ or ‘an,’ as used herein, are defined as one or more than one. Also, the use of introductory phrases such as ‘at least one’ and ‘one or more’ in the claims should not be construed to imply that the introduction of another claim element by the indefinite articles ‘a’ or ‘an’ limits any particular claim containing such introduced claim element to inventions containing only one such element, even when the same claim includes the introductory phrases ‘one or more’ or ‘at least one’ and indefinite articles such as ‘a’ or ‘an.’ The same holds true for the use of definite articles. Unless stated otherwise, terms such as ‘first’ and ‘second’ are used to arbitrarily distinguish between the elements such terms describe. Thus, these terms are not necessarily intended to indicate temporal or other prioritization of such elements. The mere fact that certain measures are recited in mutually different claims does not indicate that a combination of these measures cannot be used to advantage.

[0168] The invention has been described by way of examples only. Therefore, the foregoing is considered as illustrative only of the principles of the invention. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation shown and described, and accordingly, all suitable modifications and equivalents may be resorted to, falling within the scope of the claims.

Claims

Claims:

1. A method of generating quantitative parameters for pancreatobiliary ducts and lesions in a visceral organ from volumetric medical imaging scan data obtained from a medical imaging scan of all or part of the visceral organ; the method comprising: locating a position of at least one lesion and at least one pancreatobiliary duct in the visceral organ on the volumetric medical imaging scan data; performing segmentation of a volume within the medical imaging scan using the volumetric medical imaging scan data to identify and segment the at least one lesion and the one pancreatobiliary duct; determining at least one structural parameter of the at least one lesion that is segmented; determining at least one structural parameter for the at least one pancreatobiliary duct that is segmented.

2. A method of claim 1 wherein the structural parameters of the at least one pancreatobiliary duct that is segmented comprise a quantitative parameter.

3. A method of claim 1 wherein at least one of the structural parameters of the at least one pancreatobiliary duct that is segmented is determined for at least one location along a length of the at least one pancreatobiliary duct.

4. A method of claim 3 wherein the structural parameters for the at least one pancreatobiliary duct that is segmented comprises at least one parameter representative of at least one of: a centre-line position of the at least one pancreatobiliary duct at the at least one location; a centre-line orientation of the at least one pancreatobiliary duct at the at least one location; a radius of the at least one pancreatobiliary duct at the at least one location; a diameter of the at least one pancreatobiliary duct at the at least one location; a cross-sectional orientation of the at least one pancreatobiliary duct at the at least onelocation; a cross-section profile of the at least one pancreatobiliary duct at the at least one location; an indication of the presence of a branching point for the at least one pancreatobiliary duct at the at least one location; and an orientation of a branching point for the at least one pancreatobiliary duct at the at least one location.

5. A method of any preceding claim wherein the structural parameters of the at least one lesion that is segmented comprise a quantitative parameter.

6. A method of any preceding claim wherein the set of structural parameters of the at least one lesion that is segmented comprises at least one parameter representative of at least one of: a maximum diameter of the lesion; a volume of the lesion; a determination of the presence of septa within the lesion; a determination of the shape of the lesion; determination of the content of the lesion.

7. A method of any preceding claim wherein the structural parameters of the at least one lesion that is segmented comprise a qualitative parameter.

8. A method of any preceding claim further comprises: using at least one of the structural parameters of the at least one lesion that is segmented and the at least one of the structural parameters of the pancreatobiliary duct that is segmented to determine a distance between a lesion that is segmented and a pancreatobiliary duct that is segmented.

9. A method of any preceding claim, wherein the method further comprises: partitioning the at least one pancreatobiliary duct that is segmented and identified into a region upstream of the lesion, a region downstream of the lesion, and a region interacting with the lesion, and determining the diameter and variation in diameter ofthe pancreatobiliary duct for each of these different regions.

10. A method as claimed in claim 9 wherein a width profile of at least one of the upstream, downstream and interacting regions of the pancreatobiliary duct is used to determine additional metrics for the pancreatobiliary duct.

11. A method of claim 9 or claim 10 further comprises detecting patterns of change within the at least one pancreatobiliary duct based on median diameter estimations along the length of the at least one pancreatobiliary duct.

12. A method of claim 11 further comprises detecting patterns of change within the at least one pancreatobiliary duct by applying at least one pattern detection window to the median diameter estimation data.

13. A method of claim 11 or claim 12 further comprises detecting at least one of: strictures; dilatations; and beading.

14. A method as claimed in any preceding claim wherein the at least one lesion is identified based on a maximum intensity protection of voxels that make up the volumetric medical imaging scan data.

15. A method as claimed in any preceding claim wherein the at least one lesion that is segmented is identified by applying a threshold to all of the medical imaging scan data, and reviewing the volume rendering of every voxel that exceeds the threshold.

16. A method as claimed in any preceding claim wherein the visceral organ is a pancreas.

17. A method of any preceding claim further comprising a step of performing tubular enhancement on the volumetric medical imaging scan data to derive at least one tubeness measure for each voxel within the volumetric medical imaging scan data; wherein the tubular enhancement comprises, for each voxel within the volume of the medical imaging scan data, computing at least one metric distinguishing tubular structures from other structures and computing a tubeness measure for the respective voxel based at least partly on the at least one metric.

18. A method of any preceding claim further comprising using multi-scale Hessian based tubular pancreatobiliary structure enhancement.

19. A method as claimed in any preceding claim wherein the volumetric medical imaging scan data comprises imaging scan data obtained via a magnetic resonance imaging modality.

20. A method as claimed in any preceding claim further comprising determining the total number of pancreatic lesions shown in the volumetric imaging scan data.

21. An image processing system arranged to generate quantitative parameters for pancreatobiliary ducts and lesions in a visceral organ from volumetric medical imaging scan data from a scan that shows all or part of a visceral organ; the image processing system comprising at least one processing device arranged to: locate a position of at least one lesion in the visceral organ on the medical imaging scan data; segment the at least one lesion that is located and determine one or more structural parameters for the at least one lesion that is segmented; perform segmentation of a volume of the medical imaging scan data to identify pancreatobiliary ducts in the visceral organ within the volume of the medical imaging scan; generate at least one set of quantitative structural parameters for the at least one pancreatobiliary duct that is segmented.

22. An image processing system of claim 1 arranged for at least one of the pancreatobiliary ducts identified within the volume of the medical imaging scan data, to compute at least one set of quantitative structural parameters for at least one location along a length of the pancreatobiliary duct.

23. The image processing system of claim 22 wherein the set of quantitative structural parameters for the at least one pancreatobiliary duct that is segmented comprises at least one parameter representative of at least one of: a centre-line position of the at least one pancreatobiliary duct at the at least one location; a centre-line orientation of the at least one pancreatobiliary duct at the at least one location; a radius of the at least one pancreatobiliary duct at the at least one location;a diameter of the at least one pancreatobiliary duct at the at least one location; a cross-sectional orientation of the at least one pancreatobiliary duct at the at least one location; a cross-section profile of the at least one pancreatobiliary duct at the at least one location; an indication of the presence of a branching point for the at least one pancreatobiliary duct at the at least one location; and an orientation of a branching point for the at least one pancreatobiliary duct at the at least one location.

24. The image processing system of claim 21 , 22, or 23, wherein the set of structural parameters for the at least one lesion that is segmented comprise quantitative structural parameters.

25. The image processing system of claim 21 , 22, 23, or 24 wherein the set of structural parameters for the at least one lesion that is segmented comprises at least one parameter representative of at least one of: a maximum diameter of the lesion; a volume of the lesion; determination of the presence of septa within the lesion; a determination of the shape of the lesion; determination of the content of the lesion.

26. The image processing system according to any of claims 21 to 25 wherein the structural parameters for the at least one lesion that is segmented comprise qualitative structural parameters.

27. The image processing system of any of claims 21 to 26 wherein the processing device is further configured to perform the steps of: using one of more of the structural parameters for the at least one lesion that is segmented and one or more of the quantitative structural parameters for the pancreatobiliary duct that is segmented to determine a distance between the pancreatobiliary duct and the lesion.

28. The image processing system of claims 21 to 27 wherein the processing device is further configured to perform the steps of:partition the pancreatobiliary duct into a region upstream of the lesion that is segmented, a region downstream, and a region interacting with the lesion that is segmented and determine the diameter and variation in diameter of the pancreatobiliary duct for each of these different regions.

29. The image processing system of claim 28 wherein a width profile of the different regions of the pancreatobiliary duct is used to determine additional metrics for the duct.

30. The image processing system of any of claims 21 to 29 wherein the processing device is further configured to perform the steps of detecting patterns of change within the at least one pancreatobiliary duct based on median diameter estimations along a length of the at least one pancreatobiliary duct.31 . The image processing system of any of claims 21 to 30 wherein the processing device is further configured to perform the steps of detecting patterns of change within the at least one pancreatobiliary duct by applying at least one pattern detection window to the median diameter estimation data.

32. The image processing system of any of claims 21 to 31 wherein the processing device is further configured to perform the steps of detecting at least one of: strictures; dilatations; and beading.

33. The image processing system of any of claims 21 to 32 wherein the processing device is further configured to identify at least one lesion that is segmented by applying a threshold to all of the medical imaging scan data , and reviewing a volume rendering of every voxel that exceeds the threshold.

34. The image processing system of any of claims 21 to 33 wherein the processing device is further configured to perform the steps: of performing tubular enhancement on the volumetric medical imaging scan data to derive at least one tubeness measure for each voxel within the volumetric medical imaging scan data; wherein the tubular enhancement comprises, for each voxel within the volume of the medical imaging scan data, computing at least one metric distinguishing tubular structures from other structures and computing a tubeness measure for the respective voxel based at least partly on the at least one metric.

Images