METHOD IMPLEMENTED BY COMPUTER AND SYSTEM TO ISOLATE A POTENTIAL ANOMALY IN IMAGE DATA, AND, MACHINE

专利摘要:
METHOD AND SYSTEM FOR ISOLATING A POTENTIAL ANOMALY IN IMAGE FORMATION DATA, AND, MACHINE-READABLE MEANS. A method to isolate a potential anomaly in imaging data is disclosed, the method characterized by the fact that it comprises providing a set of at least a certain anomaly property representative of a particular anomaly; provide an anomaly property identifier to identify each of the at least one particular anomaly property; in image formation data, isolating a first zone having a first property and a group of at least one zone, each of at least one other zone having a corresponding different property than the first property; in image formation data, and resulting from the isolation of the first zone and a group of at least one other zone, provide a transition zone selected from a group consisting of: a closed zone separating the first zone and the hair group at least one other zone; and a closed zone extending into one of the first zone and the group of at least one other zone; apply the anomaly property identifier to identify each one (...).
公开号:BR112013021657B1
申请号:R112013021657-3
申请日:2012-02-24
公开日:2021-03-16
发明作者:Florent André Robert Chandelier；Thomas Bernard Pascal Vincent
申请人:Dog Microsystems Inc；
IPC主号:

专利说明:

CROSS REFERENCE FOR RELATED ORDERS
[0001] This application claims priority of US provisional patent application No. 61 / 446,342 with the title “Method and apparatus for isolating a potential anomaly in imaging data and its application to medical imagery”, which was filed on February 24 2011, the specification of which is incorporated for reference. FIELD OF THE INVENTION
[0002] This invention relates to data processing. More precisely, the invention pertains to a method and system for image processing, in particular computer-aided detection and diagnosis (CADe and CADx respectively) of medical data sets, more specifically for the detection of anomalies in hollow organs such as such as, but not restricted to, rectal colon injuries or abdominal aortic aneurysms. KNOWLEDGE
[0003] In medicine, CAD systems are significant in finally emitting anomalies in potentials within medical images. Previous art methods and systems have typically brought together 2D and 3D approaches with a preferred and highly successful process being a coarse to fine approach, detecting several initial stretches still refined by a classifier that only the “best” candidates can survive.
[0004] A first method for identifying colon polyp is disclosed in International PCT application No. WO 98/37517 under the title "Automatic analysis in virtual endoscopy". The proposed methods and systems require the targeting of an organ of interest, typically a colon. After successful segmentation, a mesh, that is, a set of three-dimensional surfaces oriented from normals, is used to interactively visualize the colon, in addition to support an “analysis of shape characteristics” characterized by the fact of understanding the step of determining a convexity value for each population representing a quantity and direction of curvature.
[0005] Such a method and the subsequent equivalent methods that have been proposed, such as that disclosed in US Patent No. 7,369,638, base their strategy on identifying forms of characteristic polyps when compared to the continuous appearance of the colon mucosa for example . Therefore, an accurate representation of the organ based on a segmentation process is required in order to accurately identify forms of interest. Qualified professionals will appreciate that injuries in random ways cannot be detected. For the case of CT colonography, also called virtual colonoscopy, skilled practitioners will appreciate that most prior art methods are significant for identifying anomalies of polyps shapes (spherical shapes), neither cancer nor mass (shapes with potentially no spherical protrusion).
[0006] However, and as mentioned in US Patent No. 7,236,620 (hereinafter '620) with the title "Computer-aided detection methods in volumetric imagery", detectors based on curvature calculation use derivative processes that are susceptible to produce spurious emissions due to noise in the input image. Such a limitation also affects each equivalent method involving gradient and three-dimensional surfaces considering that the zero-value three-dimensional surface of the distance map leads to the surfaces of the object and the derivative of the distance map leads to the normal surface, that is, a mesh, as mentioned in 1998 , using distance maps for precision surface representation in the sampled volumes, Gibson Sarah FF, Mitsubishi Electric Research Laboratory, IEEE. Qualified professionals will appreciate that any such distance map requires object segmentation, as stated in US Patent No. 7,113,617 under the title “Method of computing subpixel Euclidean distance maps”. As mentioned, a method of generating a distance map includes a step of identifying a curved boundary curve from a source image. Qualified professionals will appreciate that, for the case of CAD methods in CT Colonography, for example, previous art segmentation and distance map determinations are based on the precise identification of the internal mucosal wall of the colon on which normal surfaces are determined (such as mesh, gradient).
[0007] To overcome the limitations mentioned above of “derivative processes”, the '620 patent discloses a method based on simple spherical sums. The method requires a binary, that is, segmented, image to be entered, from which the shape is defined based on the proportion of segmented elements falling within the proportion of two spherical sum processes involving one 2D image at a time, but inspecting a region in 3D. Such a methodology reduces the amount of processing time required and is less susceptible to noise, but is still really dependent on image segmentation processing. As such, this only decreases the processing time and complexity required, but it does not improve the detection emitted as the difficulty moves towards the segmentation stage.
[0008] Concurrently, Gokturk introduced a standard method of three-dimensional recognition to detect shapes in medical images in the procedures of the Biomedical Computation Stanford 2000 symposium with the title "recognizing polyps from 3D CT colon data" where a random cut across a volume of candidate is used in order to extract shape features from the 2D slices, the latter being entered into a support vector machine classifier (SVM) still responsible for identifying polyp candidates. This was further detailed in “A statistical 3D pattern processing method for computer-aided detection of polyp in CT colonography, Gokturk et al., IEEE transaction on medical imaging, vol.20 (12) dec. 2001 “and entailed the US Patent of NR. 7,346,209. These developments lead to an approach similar to that of the '620 patent involving the use of 2D gradient sums in order to reduce noise artifacts, hoping that summation and smoothing operations would help to increase the difference between homogeneous and non-homogeneous structures, where gradient of local image in pixels other than edges will be more significant than it would be for homogeneous structures. As such, a limitation arises in the definition of the borders, which is the need to have a precise segmentation process of the structure of interest. Additionally, Gokturk's methods were more about building signatures of forms to be still entered in the classifiers than polyp detection itself.
[0009] Following both the preceding concepts and their combination, Cathier has disclosed a method and system for using cutting plans for colon polyps detection in US Patent No. 7,447,342 (hereinafter '342). The methods and systems disclosed involve refactoring the volumes in a whole set of data in order to detect small, rounded traces in any of these plans. However, and like the previous techniques discussed here, this method requires that the image be pre-processed to distinguish the colon from other structures in the image with high precision, necessary for successful polyp detection. In addition, qualified professionals will appreciate that such a technique is significant to be used for polyp shape recognition but does not address the needs for detecting cancers and masses (featuring random shapes).
[00010] To overcome the limitation of the '342 method with respect to its sensitivity to a binary limit, US Patent Application published under No. 2009/0016583 discloses the use of Divergence Gradient Field Response (DGFR). As mentioned, this method allows the detection of circles directly in the gradient domain, instead of edges or magnitude of the gradient as in the case of the '342 patent. However, two intrinsic limitations are expressed in such a methodology. First, a Divergence Gradient Field Response identifies circles of certain sizes, and since the size of the polyp to be found is not previously known, someone needs to compute DGFR for a multitude of subvolumes (subsampled volumes) covering the entire range of sizes of polyps. Therefore, the choice as to where to stop subsampling has to be made; and thereby, limiting the size of the smallest and largest polyp to be found. This is a matter of model matching techniques well known to qualified professionals. The second limitation is that DGFR detects circles, although polyps can depict more complex shapes than simple circles. Unfortunately, this technique does not address the needs for cancer and mass detection (ie, random forms).
[00011] In addition, in addition to a segmentation limitation that always exists, it can be for example. Likewise, the methods mentioned above are significant for detecting circular / ellipsoidal shapes for still detecting polyps. It is to be understood that looking at a sphere in a digital data set will be equivalent to either detecting rounded shapes or detecting local / global curvatures. There is therefore a lack of suitable methods to detect lesions of various sizes and shapes, as expressed by Dr C. Robinson in the 2009 European Congress of Radiology (ECR) in Vienna: “CAD algorithms were developed to detect polyps in the context of screening”, which study was significant to investigate the performance of a commercial CAD device based on “reader adjustable sphericity-settings” for the generation of cancer candidates. The author said that “The CAD algorithm was applied to each data set in four sphericity configurations (0, 50, 75, 100). Seventy-five was the standard manufacturer's configuration, 100 (plus sphericality) detected a more curved shape, and a single observer characterized all the CAD brands “. Respectively in sphericity settings of 0; 50; 75 and 100, the results in terms of “Sensitivity; false positive rate ”were {90.2%; 65}, {88.6%; 57}, {87.1%; 45} and {74.2%; 24}. Such a high false positive rate demonstrates the inability of such a morphology-based algorithm to accurately identify cancers and other lesions in various ways, considering high sensitivity can only be achieved if something else (not of clinical interest) is captured. This is well expressed by Dr. C. Robinson: "The detection of cancer increases with decreasing sphericity, at the expenses of decreasing specificity".
[00012] Finally, among other limitations is the fact that some methods involve a simple limit to distinguish the colon from another structure to differentiate the lumen from tissues. Although there is no such “simple boundary” method, a clear limitation of such methods would be the inability to treat CT colonography data sets resulting from reduced preparation with marking fluid / faeces where more than “simply separating air / tissue " is required. In fact, and considering such a reduced preparation, qualified professionals will understand that marked residues could represent small / rounded characteristics similar to polyps.
[00013] This, therefore, could be desirable to provide an improved method and apparatus that will overcome at least one of the drawbacks identified above. BRIEF SUMMARY
[00014] It is, therefore, an object of the present invention to provide a method and system for isolating anomalies in potentials and use thereof for hollow organs in medical data sets.
[00015] It is another object of the present invention to provide such a method and system to assist automated detection of anomalies without organ segmentation with previous precision and without prior morphological knowledge of the anomalies.
[00016] According to a modality, a method is provided to isolate a potential anomaly in imaging data, the method characterized by the fact that it comprises providing a set of at least a certain anomaly property representative of a given anomaly; provide an anomaly property identifier to identify each of the at least one particular anomaly property; in imaging data, isolating the first zone having a first property and a group of at least one other zone, each of at least one other zone having a corresponding different property than the first property; provide a transition zone resulting from the isolation of the first zone and a group of at least one other zone, the transition zone being selected from a group consisting of: a closed zone separating the first zone and the group of at least one other zone; and a closed zone extending into one of the first zone and the group of at least one other zone; applies the anomaly property identifier to identify each of the at least one particular anomaly property in at least one transition zone to provide a computed indication for a selected zone, the selected zone being at least the transition zone; determine if the computed indication for the selected zone is in accordance with each of the at least one particular anomaly property and if the computed indication for the selected zone is in agreement, assign a potential anomaly candidate indication for the selected zone and thereby isolate the potential anomaly mentioned.
[00017] According to one modality, the imaging data two.
[00018] According to another modality, the n-dimensional data set is one of a two-dimensional volumetric matrix of elements and a three-dimensional volumetric matrix of elements.
[00019] According to another modality, the n-dimensional data set originates from a device selected from a group consisting of a magnetic resonance imaging (MRI) device, a positron emission tomography device (PET), an X-ray device, an ultrasound device and any combination thereof.
[00020] According to another modality, the set of at least a certain anomaly property comprises at least one of, information related to composition, information related to shape, spatial location in imaging data, and a combination of them along of time.
[00021] According to another embodiment, the anomaly property identifier comprises at least one of, determination of tissue density, determination of tissue homogeneity gradient, determination of absence or presence of properties of tissue, determination of content / water distribution, presence determination and determination of a contrast agent distribution at a given time or over time.
[00022] According to one embodiment, the first property of the first zone comprises a certain region of air, the corresponding property of each of the at least one other zone comprises a certain region of the tissue.
[00023] According to another modality, the first property in the first zone comprises a certain marked region; the corresponding property of each of the at least one other zone comprises a certain region of the tissue.
[00024] According to another modality, the method also comprises applying the anomaly property identifier to the selected zone.
[00025] According to another modality, the method still comprises providing an indication of a potential anomaly.
[00026] According to another embodiment, providing an indication of a potential anomaly comprises at least one of, storing an indication of a potential anomaly and displaying an indication of a potential anomaly in a user interface.
[00027] According to another modality, the method also comprises transmitting the indication of a potential anomaly to a remote location.
[00028] In yet another embodiment, the image data comprises a large number of unitary image elements selected from the group consisting of pixels and voxels.
[00029] According to another modality, a machine-readable medium is provided with instructions written on it to carry out the method for isolating a potential anomaly in imaging data.
[00030] According to another modality, a method is provided to isolate a potential anomaly in imaging data, the method characterized by the fact that it understands to receive imaging data; isolating in imaging data the first zone having a first property and a group of at least one other zone, each of the at least one other zone having a corresponding different property than the first property; provide a transition zone resulting from the isolation of the first zone and a group of at least one other zone, the transition zone being selected from a group consisting of a closed zone separating the first zone and the group of at least one other zone ; and a closed zone extending into one of the first zone and the group of at least one other zone; applying a tissue homogeneity gradient identifier to at least the transition zone to provide a computed indication for the selected zone, the selected zone being at least the transition zone; determine if the computed indication is in accordance with the selected zone; and if the computed indication for the selected zone is in agreement, assign a potential anomaly candidate zone indication to the selected zone and thereby isolate the aforementioned potential anomaly.
[00031] According to another modality, a system is provided to isolate a potential anomaly in imaging data, the system characterized by the fact that it comprises a data collector; a central processing unit, operatively connected to the data collector; an I / O device, operatively connected to the data collector; a network interface circuit, operatively connected to the data collector; and a memory, operatively connected to the data collector, the memory characterized by the fact that it comprises at least one program to isolate a potential anomaly in imaging data characterized by the fact that at least one program is configured to run by the central processing unit, the at least one program to isolate a potential anomaly in imaging data characterized by the fact that it comprises: instructions for providing an anomaly property identifier to identify each of the at least one particular anomaly property; instructions for isolating, in imaging data, the first zone having a first property and a group of at least one other zone, each of the at least one other zone having a corresponding different property than the first property; instructions for providing a transition zone resulting from the isolation of the first zone and a group of at least one other zone, the transition zone being selected from a group consisting of a closed zone separating the first zone and the group of at least one another zone and a closed zone extending into one of the first zone and the group of at least one other zone; instructions for applying the anomaly property identifier to identify each of the at least one particular anomaly property in the at least transition zone to provide a computed indication for a selected zone, the selected zone being at least the transition zone; instructions to determine whether the computed indication for the selected zone conforms to each of the at least one particular anomaly property and whether the computed indication for the selected zone conforms to assign a potential anomaly candidate indication for the selected zone and thereby isolate the potential anomaly mentioned.
[00032] According to another modality, a system is provided to isolate a potential anomaly in imaging data, characterized by the fact that the memory still comprises the imaging data.
[00033] According to another modality, a system is provided to isolate a potential anomaly, characterized by the fact that the imaging data are received from a network interface circuit.
[00034] The method can be used to provide initial candidates or a complete detection scheme with efficient computational complexity and without requiring the organ of interest to be accurately segmented. This is of great advantage since the method can increase detection of obstructive anomalies in hollow organs as well as other lesions in tortuous regions where an accurate segmentation required by the prior art CAD methods can be difficult to achieve.
[00035] Furthermore, the method is not dependent on analysis in a restrictive way, nor is it dependent on an analysis of strict morphological characteristics, such as curvature. This is of great advantage over current state of the art methods such as lesions, which can portray varying shape and size. Although not depending on the characteristics of morphological analysis or form, the method can be combined with any of them as a subsequent classification process.
[00036] In one embodiment, the method may involve the use of uncertain regions, that is, regions whose type is not known, also called transition regions, describing partial volume artifact, for example, trying to extract some coherent information outside of it. This is of great advantage compared to the prior art methods that aim to reduce, limit, or prevent any information from the uncertain regions considering that they carry predominantly supposedly defective signals. BRIEF DESCRIPTION OF THE DRAWINGS
[00037] In order that the invention can be readily understood, modalities of the invention are illustrated by way of examples in the accompanying drawings.
[00038] Figure 1 (PREVIOUS TECHNIQUE) is a flow chart showing a prior art method for providing final candidates in an anomaly detection scheme.
[00039] Figure 2 is a flow chart showing a modality of a method for isolating a potential anomaly in imaging data.
[00040] Figure 3 is a scanned CT image showing a portion of a colon, according to one modality.
[00041] Figure 4 is an enlarged view of Figure 3.
[00042] Figure 5 is another scanned CT image showing a portion of a colon, according to another modality.
[00043] Figure 6 is another scanned CT image showing a portion of a colon, according to another modality.
[00044] Figures 7A to 7D show representative images of a portion of a colon, according to one embodiment.
[00045] Figure 8 is another scanned CT image showing a portion of a colon, according to another modality.
[00046] Figures 9A and 9B show representative images of another portion of a colon, according to another embodiment.
[00047] Figure 10 is a block diagram showing a modality of a processing device in which the method for isolating a potential anomaly in imaging data can be implemented.
[00048] Figure 11 is a schematic showing a modality in which a thick region of investigation is determined, for example, extending from a given region of air towards the given region of the tissue, for a distance Δ.
[00049] Figure 12 is an enlarged view of part of a scanned CT image showing a portion of a colon shown in Fig. 3.
[00050] Figure 13 is an enlarged view of part of a scanned CT image showing a portion of a colon shown in Fig. 3 characterized by the fact that the local maximum present in the thick region extended at a constant distance Δ from of a given air region are shown.
[00051] Figure 14 shows the schematic of Fig. 11 with singular final points obtained by projecting rays and effecting a reduction process in local maximums limiting the accumulation of Lightning intensity to a certain magnitude.
[00052] Figure 15 is an enlarged view of part of a scanned CT image showing a portion of a colon shown in Fig. 3 characterized by the fact that exemplary rays orthogonal to a given air region and passing through singular points have been designed. It can be seen that those rays intercept at singular points most likely belonging to the anomalies in potentials.
[00053] Figure 16 is an enlarged view of part of a scanned CT image showing a portion of a colon shown in Fig. 3 characterized by the fact that singular points have still been broken down by applying an “intensity” limit on each intersection of rays; the more rays culminating at a singular point, the more intensity, and the more likely those singular points belong to a potential anomaly
[00054] Figure 17 is the schematic of Fig. 11 characterized by the fact that singular points can be agglomerated in order to reconstruct portions of an approximate colon mucosa at the location of a potential anomaly, projecting rays in the outward direction from the clusters of at least one singular point, and involving the previously determined distance transformation map. Such portions of the approximate colon mucosa at potential anomaly locations can be further used to determine an approximate center of gravity (for 3D assessment using a center of rotation, for example) and approximate measurement of potential injuries.
[00055] Figure 18A is an enlarged view of part of a scanned CT image showing a distance field extending from certain regions of air towards certain regions of the tissues. Unlike traditional distance transformation approaches that converge towards the center of an object, the purpose of extending the distance is to provide information on how far certain regions of tissue are from certain regions of air. Such extension is restricted by a maximum penetration thickness depending on the “size of the anomaly property”.
[00056] Figure 18B is an enlarged view of part of a scanned CT image showing a surface flow determined from, and extending within, a distance field extension as shown in Fig. 18A. Such a surface flow will provide information on the location of the maximum site. The person skilled in the art will appreciate that the combination of both information maps (ie, Fig. 18a & Fig.18b) would discard false positives due to bubbles close to the surface, potentially resulting from air in the remaining stools, for example
[00057] Figure 19 illustrates a two-dimensional image originating from a CT Colonography exam. Figure 19 comprises a first image showing two regions under investigation, characterized by the fact that one represents a local maxim. Figure 19 comprises an enlarged image in which coarse pixels from an original image can be seen.
[00058] Details of the invention and its additional advantages will be apparent from the detailed description included below. DETAILED DESCRIPTION
[00059] In the following description of the modalities, references to the accompanying drawings are by way of illustration of examples by which the invention can be practiced. It will be understood that other modalities can be done without departing from the scope of the disclosed invention.
[00060] As previously mentioned, CAD systems can be used in medicine to detect anomalies in potentials in a given set of medical data. The present invention provides a method and apparatus for isolating a potential anomaly in imaging data that may be particularly useful for detecting anomalies in hollow organs such as rectal colon injuries or abdominal aortic aneurysm for non-limiting examples. Although modalities of the method are described in a medical imaging application, qualified professionals will nevertheless appreciate that several other applications can be envisaged, as will become apparent when reading from the present description.
[00061] Fig. 1 shows a prior art method used to provide final candidates in an anomaly detection scheme. The method comprises an initial process for detecting initial candidates, followed by a second process for providing final candidates. As described below, the second process allows you to rank the final candidates while trying to eliminate false positive candidates.
[00062] In the prior art method illustrated and according to processing step 102, an input data set is provided. It will be appreciated that the input data set can comprise 2D images as well as 3D images, as known to qualified professionals.
[00063] According to processing step 104, an organ segmentation is performed. The purpose of organ segmentation is to adequately locate the boundary of the organ of interest.
[00064] According to processing step 106, a detection based on the extraction of sphere or curvature is carried out.
[00065] According to processing step 108, initial candidates are provided. Initial candidates are provided following detection.
[00066] According to processing step 110, a characteristic extraction is carried out on the set of initial candidates.
[00067] According to processing step 112, a classification of initial candidates based on the characteristic extraction is carried out.
[00068] According to processing step 114, a reduction of false positives is carried out.
[00069] According to processing step 116, the final candidates are provided.
[00070] As mentioned earlier and as will become apparent below, contrary to known prior art methods, the method of the present invention is not based on an accurate segmentation of the organ of interest. The method can be used to provide either initial candidates or to provide a complete detection scheme with efficient computational complexity, as detailed hereinafter.
[00071] In fact, in one modality, the method described does not depend on a restrictive analysis or a strict morphological characteristic analysis, both arising from the need for an accurate object segmentation for its potential detection, which is of great importance. advantage since lesions to be found in general portray variable shape and size. Qualified professionals, however, will appreciate that the present method can be combined with any shape analysis or morphological characteristic analysis as a subsequent classification process, as they will become apparent below.
[00072] Now referring to Fig 2, a modality of the method for isolating a potential anomaly in imaging data will now be described.
[00073] It will be appreciated that in one embodiment, the imaging data comprises a set of n-dimensional data (characterized by the fact that n> 2) originating from an imaging system and supplied to a processing system.
[00074] It will be appreciated by qualified professionals that without restricting the size of such data sets, medical data sets are usually two-dimensional or three-dimensional volumetric arrays of elements, denoted as pixels and voxels respectively. Assuming an orthogonal coordinate system, an element pixel element is represented along the axes (i, j) (respectively (i, j, k) for voxels) at the location (x, y) (respectively (x, y, z ) for voxels). Consequently, a “slice” of a data set can be selected by specifying a “z” location along the “k” axis of a three-dimensional data set.
[00075] Still in the form of a medical imaging application, the data set can be acquired from a device selected from a group consisting of a magnetic resonance imaging (MRI) device, a tomography device. positron emission (PET), an X-ray device, an ultrasound device and any combination thereof. The acquired data set comprises at least a portion of the organ of interest, so that each element can be related to a specific property of the human body. For example, a set of medical data acquired by an X-ray CT scanner, characterized by the fact that it comprises at least a portion of the colon, will depict elements with density values expressed in Hounsfield units and displayed in grayscale colors. , and where the elements produced in black will typically represent air elements of Hounsfield values below -400Hu, thus allowing the visualization of hollow organs, such as the colon.
[00076] In one embodiment, as will become apparent below, the method may involve the use of certain regions in imaging data, that is, regions that have been confidentially detected as being of a particular known type such as bones, soft tissues, air regions, and marked regions, and the use of uncertain regions, that is, regions describing partial volume artifact for example. In this modality, the method uses at least the uncertain regions, also called transition zones, originating from the identification of the right adjacent regions neighboring the uncertain regions in order to extract some coherent information. This is different from other methods that aim to reduce, limit, or prevent any information from “uncertain” regions considering that they carry predominantly supposedly defective signals.
[00077] For example, and as will be more detailed hereinafter, in one embodiment, coherent information can be related to a property in a suspected region to depict a right region of the type of concentric tissue involved by a right region of the type of air. Another example of the property of a lesion is that it shows a more dense concentric and homogeneous tissue distribution.
[00078] Qualified professionals will appreciate that several other coherent information related to the properties of a potential injury can be considered. For example, in the case that the potential injury represents a coherent air density, the coherent information showing the presence of air bubbles can be used to rule out the potential injury and identify it as a false positive, as best illustrated hereinafter. According to another typical characteristic of lesions, they can depict tissues that are denser in their center than in and around them. Lesions can also portray continuous properties such as denser and denser tissues without any other fabric type jerks. Conversely, remaining non-homogeneous colon fluids or fecal matter depict highly inhomogeneous tissue characteristics and the presence of trapped air bubbles. However, typical lesions can also, in some cases, characterize surrounding high-density tissue if coated within a labeling agent, such as Barium or Iodine, for example. Finally, when considering intravenous acquisitions, lesions may depict tissue jerks due to centered upper densities whereas lesions can be highly vascularized.
[00079] Qualified professionals will appreciate that various physiological properties or other types of colon injury properties may be of interest in implementing the method described here in the given rectal colon shielding application. Qualified professionals will also appreciate that other typical properties of a given potential anomaly can be derived from the clinical knowledge of the same, according to a particular field of application and for a given type of potential anomaly.
[00080] Still referring to Fig 2 and according to processing step 202, a set of at least a certain anomaly property representative of an anomaly is provided. In the example discussed above, the set of at least a certain anomaly property can comprise a certain anomaly property that is representative of an injury, for example, a particular anomaly property can describe a region of the right concentric tissue type whose density increases in the inner direction, surrounded by a region of the right type of air.
[00081] Qualified professionals will appreciate when reading the description that several other properties of anomaly representative of a particular injury can be considered.
[00082] Qualified professionals will also appreciate that a given anomaly can be represented with a single property as well as with a large number of properties. For example, the set of at least a certain anomaly property can comprise at least one of information related to composition, information related to shape, spatial location in imaging data and a combination of them over time.
[00083] According to processing step 204, an anomaly property identifier to identify each of the at least one particular anomaly property is provided. In an embodiment described above, the anomaly property identifier may comprise the determination of a gradient, as will become apparent below.
[00084] Qualified professionals will appreciate that several anomaly property identifiers can be considered for the purpose of identifying a corresponding anomaly property. It will also be appreciated that a combination of a large number of anomaly property identifiers can be used according to a particular application, as will become apparent below. For example, in one embodiment, a first anomaly property identifier can be provided to identify a first particular anomaly while a second anomaly property identifier can be provided to identify a second particular anomaly property.
[00085] As will become apparent to qualified professionals, the anomaly property identifier can comprise at least one of tissue density determination, tissue homogeneity gradient determination, tissue distribution for a given region, absence determination or presence of certain tissue properties, determination of water content / distribution in the case of MRI images, presence and distribution of contrast agent at a given time or over the course of acquisitions featuring intravenous contrast agent.
[00086] In an additional modality, intensity profile analysis characterized by the fact that it understands any derivative processes such as Gradient and Divergence analysis [Semi-Automatic Generation of Transfer Functions for Direct Volume Rendering, G. Kindlmann & JW Durkin, IEEE Symposium on Volume Visualization, 1998] and [Fully atutomated three-dimensional detection of polyps in fecal-tagging CT Colonography, J. Nappi, H. Yoshida, Acad. Radiol., 14 (3) -287-300, March 2007], volume and surface distribution analysis characterized by the fact that it understands, but is not limited to distance transformation, intensity distributions, Euler's frontal evolution theories [Evolution, Implementation, and Application of level set and makes matching methods for advancing front, JA Journal of computational Physics, 169: 503-555, 2001] and Dynamic Frontal evolution such as advanced Pareto boundaries can be used.
[00087] Qualified professionals will appreciate that the anomaly property identifier can be derived from the image acquisition physics, the particular anomaly physiology or a combination of them, as will become apparent when reading this description. It will be appreciated that in a modality the anomaly property identifier is applied over a region, preventing further dependence on the segmentation processes.
[00088] It will be appreciated that tissue composition properties originate from the clinical understanding of any anomaly and from the physics of image acquisition, being density in Hounscampo values for CT scanners based on X-rays and water content in Resonance Imaging Magnetic as non-limiting examples.
[00089] Still referring to Fig 2 and according to processing step 206, the first zone and a group of at least one other zone are isolated in imaging data.
[00090] It will be appreciated that the first zone has a first property. Furthermore, it will also be appreciated that each zone in the group of at least one other zone has a corresponding different property than the first property, as detailed below.
[00091] For example, and in one embodiment, the first zone may comprise a certain region of air while the other zones of the group may comprise a region of certain tissue.
[00092] In another mode, the first zone can comprise a certain marked region, as will be detailed hereinafter. It will be appreciated that it certainly depends on the unambiguity of the information describing a particular physical object (thus based on the physics of acquisition and understanding of the clinical aspects related to the object). by way of non-exhaustive example, regions of Hounsfield values below - 500 most likely depict regions of air.
[00093] According to processing step 208, and resulting from the isolation of the first zone and a group of at least one other zone, the transition zone is provided in imaging data. The transition zone is selected from a group consisting of a closed zone separating the first zone and the group of at least one other zone and a closed zone extending into one of the first zone and the group of at least one other zone, as will become apparent below.
[00094] In one embodiment and as mentioned above, the transition zone may be an uncertain region whose type is not known.
[00095] As will become apparent below and in a modality, the transition zone is a closed zone in a 2D slice of imaging data. In the case where no closed zone can be provided on the imaging data slice under analysis, the process of isolating a potential anomaly can be stopped without further processing and without isolating any potential anomaly, as further detailed hereinafter.
[00096] Fig. 3 shows an example of imaging data characterized by the fact that it comprises a portion of a colon 300. The image data comprises an X-ray image of the CT scanner representing a 302 lesion. In this image, elements of darker voxels represent air, gray intensity voxel elements represent soft tissues and whitish elements represent marked bones or tissues. As shown, the imaging data comprises a first zone 304 having a first property, at least one other zone 306 having a corresponding different property than the first property and a transition zone 308. In the illustrated embodiment, the first zone 304 is a right region of air totally surrounded by an adjacent closed transition zone 314. The other zone 306 is a right region of total tissue surrounded by a transition zone 308 which is a closed zone. Transition zones 308 and 314 comprise elements that have certainly not been classified as belonging to a particular type of known elements.
[00097] Still referring to Fig 2 and according to processing step 210, once the corresponding regions have been provided, the anomaly property identifier used to identify each of the at least one particular anomaly property is applied to the hair minus transition zone to provide a computed indication for a selected zone, the selected zone being at least the transition zone. As will become apparent below for qualified professionals, in one modality, the anomaly property identifier can also be applied in at least one of the certain zones in addition to the transition zone.
[00098] According to processing step 212, a determination is made as to whether the computed indication for the selected zone is in accordance with each of the at least one particular anomaly property, as described below.
[00099] According to processing step 214 and 216, if the computed indication for the selected zone is in agreement, an indication of a potential anomaly candidate zone is assigned to the selected zone and thereby isolate the potential anomaly.
[000100] It will be appreciated that the indication of a potential anomaly can be provided according to several modalities.
[000101] In particular the provision of a potential anomaly may comprise at least one of storing the indication of a potential anomaly and displaying the indication of a potential anomaly in a user interface circuit.
[000102] It will be appreciated that the indication of a potential anomaly can still be transmitted to the remote location.
[000103] In one embodiment, the object of interest, that is, the hollow organ is defined and identified in imaging data before isolating a potential anomaly. Prior art methods and systems mainly refer to such identification as a segmentation process, with the purpose of isolating an organ and more precisely defining its boundary (contour) such as the colon mucosa. Such segmentation typically provides an object mask that can be binary with elements in the foreground representing the object and the background representing a non-object zone [Fast image segmentation, P.I. Corke & H.I. Anderson, Dpt of computer and information science - school of engineering and applied science - university of Pennsylvania, Philadelphia, July 1989], even though this binary mask may arise from previous segmentation processes involving non-binary masks [Fuzzy connectedness and image segmentation JK Udupa & PK Saha, IEEE, 91 (10): 1649-1669, October 2003].
[000104] As disclosed in the co-pending PCT application by the same applicant of No. PCT / CA2009 / 001749, with the title “Méthod for determining in the estimation of a topological support of a tubular structure in addition units of its virtual range endoscopy”, and in PCT application No. PCT / CA2009 / 001743, with the title “Method and System for filtering image data and use thereof virtual range endoscopy”, which are both incorporated here for reference in their entirety, an appropriate support can be provided without use a segmentation process to identify the object of interest. The methods described involve the definition of masks of certainties that can encompass thick regions around the border of the organ. Qualified professionals will appreciate that in such modality different masks representing different regions with different certainties in terms of their presence for the object of interest, supported by considerations of topology and connectivity, are provided. Qualified practitioners will appreciate that such modalities can enable the provision of a complete CT Colonograpy system that does not introduce a colon segmentation process, but provides information needed by the end user to perform rectal colon cancer review in the data sets of the patient, featuring a state-of-the-art tool such as electronic colon cleansing.
[000105] In order to detect anomalies in potentials, as mentioned earlier, several zones have to be provided. In one embodiment, different masks using different Window / Levels can be used in imaging data to isolate the first zone and a group of at least one other zone and provide at least one transition zone. In an additional embodiment, four masks can be used, respectively a colon mask, a bone mask, a lung mask, and an abdominal mask. The anomaly property identifiers used for each of the at least one particular anomaly property can be applied to each Window / Levels.
[000106] In fact, as it should be apparent to qualified professionals, applying the masks according to one of the corresponding methods known in the art to which the invention belongs, it is possible to approximately identify and label the different “connected components” in each slice in 2D of a set of relevant medical data [E. Davies Machine Vision: Theory, Algorithms and Practicalities, Academic Press, 1990, Chap. 6.]
[000107] Medical images can then be individually processed in order to identify disconnected regions of non-air disconnected within regions of air, that is, flotation surfaces or unconnected regions. Detection of potential anomaly can be carried out as previously described for each such non-air region.
[000108] In another mode, the masks used to identify the zones of interest can be obtained from certainty masks as provided in “Method for determining na estimation of a topological support or a tubular strucutre and use thereof virtual range endocsopy” and “Method and System for filtering image data and use thereof in virtual endoscopy” mentioned above.
[000109] For example, in one embodiment, the following masks are provided: high certainty air masks are provided: high certainty air mask, high certainty Tg / Bone mask, uncertain mask for air / soft tissue interfaces , uncertain mask for mark / bone - soft tissue interfaces, uncertain mask for mark - bone / air interfaces and the right mask to reduce colon data sets obtained using anatomical topology from within the mask chart, as detailed in the aforementioned copending orders. In yet an additional modality, improved computational efficiency can be provided by digitally scanning the elements of the given soft tissue mask belonging to the final colon identification mask subsequent to the anatomical topology processing of the mask's connectivity graph. The main reason for scanning this "layer" is the fact that any lesion will have a dense "interior" describing characteristics similar to soft tissue, muscle, fat and so on.
[000110] In a detailed form hereinafter, in each border element of the object of interest, that is, the elements in the internal direction for the colon in the case described, some cut plans can be provided in order to investigate a potential trait of a lesion close to the colon mucosa. A trace of an injury is defined as the surface of a connected labeled component not connected to another element belonging to a certain colon mucosa (US Patent 7,447,342).
[000111] In an additional detail detailed hereinafter, a refined outlet can be obtained by detecting non-air surfaces describing a topological hole due to the presence of air inside (typically air bubble inside a residual unmarked false positive layer), or the presence of marked density within the surface. Qualified professionals will appreciate that this will help to refine potential early anomaly candidates.
[000112] Additionally, in yet another modality, the initial potential anomaly can be entered on top of any other current methods and systems in order to improve its sensitivity, specifically in the case of minor injuries or obstructive injuries that are typically quite difficult to identify from typical 3D schematics.
[000113] In yet an additional modality, potential anomaly candidates can be further classified based on a set of characteristics computed for each element of the non-air region. An example of such a feature is the density distribution of each element in the non-air region as well as the non-air surface. In fact, using the appropriate masks to initiate an analysis and provide the corresponding zones in which the isolation method is applied intrinsically eliminates any false positives potentially arising from the marked residues. This is a major improvement compared to current state-of-the-art methods and systems that require false positive reduction steps during or after candidate classification. In addition, potential anomaly candidates can be detected within marked regions without the need for an appropriate electronic cleaning method. The method for isolating a potential anomaly is, as a matter of fact, not dependent on the performance of the electronic cleaning in terms of preserving the colon mucosa considering that the problem has been dealt with in a different way, as should become apparent to qualified professionals when reading this description.
[000114] In an additional modality, qualified professionals will appreciate that each trait of a potential injury can be analyzed in a different window / level in order to increase the certainty of a potential candidate by observing their presence in different “limits of characteristic of the fabric".
[000115] In yet an additional modality, the detected trace can still be analyzed. In one embodiment, the analysis refers to looking for a closed topological surface, with no holes. This can ensure that no air bubbles are present within the trace (false positive), thereby reducing the occurrence of false positives.
[000116] Again, in another mode, the step of analyzing the trace comprises the use of a gradient field in order to detect closed surfaces.
[000117] Still in an additional modality, the trace analysis can include an analysis of neighboring masks. Neighboring masks, determined from the connectivity graph tree as mentioned above, can be one of branded / soft tissue or air / soft tissue masks. In these masks, derivative characteristics describing the noise / ambient behavior can be determined.
[000118] In an additional modality, the analysis of layer masks can be used to detect a surrounding layer around the initially detected surface. Using such a mechanism, the certainty that the candidate is a relevant clinical finding is increased, despite the fact that the neighboring layer can be seen as “blurry”, describing a global behavior involving soft tissue. Such behavior is defined as a partial volume artifact around an injury and reinforces the relevance of the trait, as will be further detailed.
[000119] Based on the two copendent requests previously mentioned in the identification of hollow organs characterized by the fact that no segmentation is used, the present method can be used to only investigate the "thick regions" provided by the non-segmentation process. However, that only the present method can be implemented with a non-segmented organ, such a combination would dramatically decrease the required processing time.
[000120] Applications of the method for detecting rectal colon injuries will now be described with reference to Figs. 3 to 9B that show anomalies in potentials. The illustrated anomalies have both, polyps, cancers and obstructive cancers being optically confirmed, ranging in size from 4.5 mm to 40 mm or are false positives.
[000121] In Fig. 3 previously described, a first contour 310 outlining a certain region of soft tissue 306 and a second contour 312 outlining a certain region of air tissue 304 are shown. The region between them is a transition zone 308 which is a highly uncertain region.
[000122] It is worth mentioning that, as illustrated in Fig. 3 and as previously described, a property of a particular suspicious region is to portray a certain region of the concentric tissue type surrounded by a certain region of the air type. In fact, an expected topological property for a hollow organ is that there is no flying element inside the hollow tube, specifically during a 2D analysis. However, it can be anticipated that the intestinal junction valve may more likely appear as such a "flying object" since it is only the region of the lumen of the irregular colon of the colon, as should be apparent to qualified professionals. As such, it can be processed accordingly, as explained in US Patent No. NR. 7,440,601.
[000123] Now referring to Fig 4, sampling elements in and around the uncertain region, that is, the transition zone, were illustrated. These sample points can be used to isolate the potential anomaly, as described below.
[000124] Now referring to Fig 5, sampled gradient vectors are shown computed originating from two sampled points shown in figure 4 next to the uncertain region. This computation allows to determine if the property in the transition zone is in agreement or not in order to detect and isolate a potential anomaly. In fact, qualified professionals will appreciate that the purpose of this processing step is to make most of the information coherent in the region uncertain.
[000125] In this example, it will be appreciated that the uncertain region surrounding the colon mucosa (the transition zone) represents a gradient going inward, that is, penetrating into the soft tissue and surrounding the organs. Likewise, it can be seen that the uncertain region 308 involving the lesion 306 indicates coherent information of denser tissues within the potential findings. Using coherent information from the uncertain region, it is possible to enhance the physiological properties of injuries, e.g. property of a coherent and homogeneous gradient of denser concentric tissue originating from the uncertain region. Qualified professionals, therefore, will appreciate that it may be possible to derive relevant information using coherent information from uncertain regions.
[000126] It will be appreciated that, as mentioned above, this method is radically different from methods of the prior art that attempt, by all means, to decrease the impact of volume artifacts and to develop smoothing and summing processes to attenuate noise interference. In fact, prior art methods attempt to detect the boundary and use of those boundary points and possibly the points included in the processing. These methods do not use the uncertain points. On the contrary, in a modality of the present disclosed method, gradient analysis of uncertain regions is used in order to extract, for example, potentially coherent information of attenuation of denser concentric tissue in specific locations. Specifically, it will be appreciated by the professional qualified in the art that, in one modality, most elements of calculations for the gradient calculation originate from a thick region of uncertain elements as opposed to the use of elements included in the segmented region (certain elements) for inventions of the state of the art.
[000127] Qualified professionals will appreciate that, in an additional modality, ownership of a suspect region may be to portray a right region of the type of concentric fabric surrounded by a right region of the type of mark, for which an identifier of property could be a concentric gradient increasing the density of voxels in the region. This last modality is the previous reciprocal modality for analyzing the detection of a marked patient that does not require electronic cleaning.
[000128] Now referring to Fig 6, another exemplary method of isolating a potential anomaly will be described. This figure illustrates the application of the gradient analysis approach involved in figure 5 that can prevent the addition of false positives by requiring additional false positive reduction processing steps. A “false positive” element 600 imitating a potential bulge of the colon mucosa in the colon lumen is shown. Such a bulge would be defined as a finding of positive CAD by current methods based on analysis in an analysis way (based on curvature or based on gradient shock). On the contrary, using the principle of a present method to isolate a potential anomaly, it can be seen that a gradient analysis of the surrounding uncertain region does not suggest a coherent soft tissue concentration. In addition, the two right layers (air and soft tissue) do not end up creating a certain region of tissue surrounding a certain region of air as shown in figure 5 for example. This illustrates that the present method can make it possible to combine processing steps of false positive reduction and detection during a single analysis, which is of great advantage. Qualified professionals will also appreciate that the method can reduce processing time, which is also of great advantage.
[000129] Now referring to Fig 7A to 7D, a small polyp of 4.5 mm 700 is illustrated. Fig. 7B shows that the gross axial image does not provide enough information to still identify this small polyp since the right layer of air does not involve a certain region of tissue, as shown in Fig. 5. In this case, a processing step of cutout that cut out planes at different angles at the right air boundary can be implemented. Such a cropped image is shown in Fig. 7C and Fig. 7D as a white plane. This cropped image shown in Fig. 7C, is now possible to identify a region of surrounding air, not a region of tissue but an uncertain region.
[000130] As shown in Fig. 7C, gradient properties determined from this enclosed region show a soft tissue trend, which has been further validated by the right Gradient boundary property. Once again, information from the uncertain region provides enough properties to determine a positive finding of actual CAD revealing a 4.5 mm real polyp. Again, this is derived from the nature of a type of lesions and polyps that have soft tissue or denser tissue at its center, in an even distribution.
[000131] Now referring to Fig 8, the behavior of a modality of a present method involving the analysis of gradient property is better illustrated. In a manner similar to Figure 3, some small uncertain regions 800 are surrounded by a certain region of air. The analysis of the gradient of such regions revealed findings of false positives. This shows the robustness of the present method that leverages information from uncertain regions to discard false positive candidates in a processing step equivalent to that of identifying positive CAD findings. Such a combination of two steps in one allows to improve the processing speed, which is of great advantage.
[000132] Now referring to Fig 9A and 9B, another modality of the method that helps to avoid trap doors and false positives represented by the morphological CAD processes will be described. As shown on the left of Fig. 9A, a spherical shape 900 in the colon lumen can be seen, it would still be reported as a CAD finding for current state of the art strategies. On the contrary, the present method would inherently classify such a region as being a false positive considering the candidate region is filled with air, as illustrated in Fig. 9B. In fact, by analyzing the gradient of the uncertain region between air / tissue, it can be seen that the gradient indicates folds and colonic mucosa, but no concentration of tissue gradient right in the center of the "spherical candidate region" considering its filling with air. This rules out such a region as a reported potential CAD finding.
[000133] Now referring to Figure 10, a modality of a processing device 1000 is shown in which the method for isolating a potential anomaly in imaging data can be advantageously used.
[000134] The processing device 1000 comprises a central processing unit 1002, I / O devices 1004, a network interface circuit 1008, a data collector 1006 and a memory 1010, a central processing unit 1002, devices of I / O 1004, a network interface circuit 1008 and memory 1010 are operatively coupled using data collector 1006.
[000135] More precisely, a central processing unit 1002 is adapted to process data and instructions. A network interface circuit 1008 is adapted to operatively connect processing device 1000 to another processing device (not shown) via a data network (not shown). Qualified professionals will appreciate that various modalities of the 1008 network interface circuit can be provided. Furthermore, skilled professionals will also appreciate that a 1008 network interface circuit can operate according to various communication protocols such as TCP / IP for example.
[000136] I / O devices 1004 are used to enable a user to interact with the processing device 1000. Qualified professionals will appreciate that various modalities of I / O devices 1004 can be used. For example, I / O devices 1004 can comprise at least one of, a keyboard, a screen and a mouse.
[000137] Qualified professionals will also appreciate that various types of data collectors 1006 can be provided.
[000138] It will also be appreciated that various modalities of memory 1010 can be provided. Furthermore, it will be appreciated that memory 1010 can be used to store, in one embodiment, an operating system 1012, at least one program to isolate a potential anomaly in imaging data 1016, characterized by the fact that the at least one program it is configured to run by central processing unit 1002, and database 1014 used to operate at least one program to isolate a potential anomaly in imaging data 1016.
[000139] In one embodiment, the at least one program to isolate a potential anomaly in 1016 imaging data comprises instructions for providing a set of at least a certain anomaly property representative of a particular anomaly.
[000140] The at least one program for isolating a potential anomaly in 1016 imaging data further comprises instructions for providing an anomaly property identifier to identify each of the at least one particular anomaly property.
[000141] The at least one program to isolate a potential anomaly in imaging data 1016 still comprises instructions for isolating, in imaging data, the first zone having a first property and a group of at least one other zone, each of the at least one other zone having a corresponding different property than the first property.
[000142] It will be appreciated that the imaging data can be stored in the 1014 database. The imaging data can be obtained from at least one of the I / O devices 1004 and a network interface circuit 1008.
[000143] The at least one program to isolate a potential anomaly in 1016 imaging data further comprises instructions for providing a transition zone resulting from the isolation of the first zone and a group of at least one other zone, a transition zone being selected from a group consisting of a closed zone separating the first zone and the group from at least one other zone and a closed zone extending into one of the first zone and the group from at least one other zone.
[000144] The at least one program to isolate a potential anomaly in 1016 imaging data further comprises instructions for applying the anomaly property identifier to identify each of the at least one particular anomaly property in the at least transition zone to provide a computed indication for a selected zone, the selected zone being at least the transition zone.
[000145] The at least one program to isolate a potential anomaly in 1016 imaging data further comprises instructions for determining whether the computed indication for the selected zone conforms to each of the at least one particular anomaly property and whether the indication computed for the selected zone agrees, assign an indication of potential anomaly candidate zone for the selected zone and thereby isolate the potential anomaly anomaly.
[000146] Qualified professionals will appreciate that operating system 1012 is used to manage interactions between a central processing unit 1002, I / O devices 1004, a network interface circuit 1008, data collector 1006 and memory 1010.
[000147] In a preferred modality, a singularity analysis is performed to decrease the number of elements where analysis of cutting planes is carried out as expressed above. As illustrated in Fig. 11, a thick investigation region is determined, for example, extending from the right region of air towards the right region of the tissue, for the distance Δ. In a modality, Δ refers to the characteristics of the singularities to be identified. For the specific case of rectal colon injuries, Δ is determined such that singularities would exist for both the smallest and the largest anomaly to be reported.
[000148] In one modality, the singularity analysis is a process optimizing the detection of anomalies in potentials while making sure that all anomalies in potentials to be reported are encompassed by the thick region under examination.
[000149] Now referring to Fig 12, a 2D image of the colon is shown, where the red border determines the right region of air, and the yellow border represents the thick region Δ extending from the right region of air.
[000150] It will be appreciated that in an additional modality, the analysis of singularity is carried out through a process of transformation of distance applied in the extended thick region. Such a distance transformation could be an Euclidean Distance Transform or a Weighted Transform as detailed in “R. Kimmel et al., Sub-pixel Distance Maps and Weighted Distance Transforms, Journal of Mathematical Imaging and Vision, 1994 ”. In this process, the determination of singularities will comprise two processing steps. A first processing step is the identification of a local maximum in the distance map while a second processing step is the disposal of a local maximum with low potential to belong to an anomaly.
[000151] Now referring to Fig 13, it is illustrated the local maximum present in the thick region extended at a constant distance Δ from the right region of air. It can be seen that this modality dramatically reduces the number of elements to be further analyzed using a cutting plan methodology that is of great advantage.
[000152] In an alternative modality, the number of local maximum can be further reduced by using rays extended from the right region of air towards the right region of tissue. These rays carry a force of intensity that decreases as it moves away from the right region of air. In an additional modality, these rays can have force profile accounting for the intensities of elements along the rays. Each of these rays is perpendicular to the right air boundary.
[000153] It is, therefore, possible to obtain singular final points, as shown in Fig. 14, projecting rays and effecting a reduction process at the local maximum, limiting the accumulation of Lightning intensity to a certain magnitude.
[000154] These singular end points are still used to center the cutting plans described above. Qualified professionals will appreciate that these modalities significantly reduce the cutting plan process, since they are performed only on localized elements, and are not based on the precise segmentation of the colon mucosa, which is of great advantage.
[000155] In an additional modality, and as illustrated in Figures 18A and 18B, the analysis of uniqueness involves the processing step of extending a field away from a first right region of air in a right region of the tissue, constrained within of a certain thickness that is determined based on the property size of the anomaly. The resulting thick region carrying the distance field information would represent the transition zone in which potential anomalies could be identified. In an additional modality, the singularity analysis would involve the determination of a surface flow extending within the thick distance field extending from the right regions of air into the right regions of tissue. In yet an additional modality, the combination of both, the extension of the distance field and the surface flow would allow the determination of local maximum in locations where the surface flow is singular and at a given distance representing the size of anomaly property in potential.
[000156] The professional skilled in the art will recognize that such a method would inherently discard regions with air bubbles and does not require an accurate segmentation of the colon mucosa. In addition, the professional with qualification in the art will recognize that the support involved to determine the field extension of the distance and determination of surface flow could be filtered through a bilateral filter, eventually making the maximum strength a priori from the right regions of the world. fabric - air and fabric - brand and certain regions of fabric affecting the weights of such bilateral filters, provide what forms are kept within the thick transition region, and disregarding potential artifacts away from the thick region.
[000157] Fig. 18A shows a distance field extending from the right regions of air towards the right regions of tissue. Unlike traditional distance transformation approaches, the purpose of extending distance is to provide information on how long the right regions of tissue are of the right regions of air, and are extensions restricted by a maximum penetration thickness depending on the “property size of the anomaly”.
[000158] Fig. 18B shows a surface flow determined from, and extending within, the extent of the distance field. Such a surface flow will provide information at the local maximum location. The skilled artisan will understand that a combination of both information maps would discard false positives due to air bubbles close to the surface, potentially resulting from the remaining air in the remaining faeces, for example.
[000159] It will be appreciated that in an additional modality, these local maxims can be used in order to address the issue of 3D camera positioning. In fact, automatically positioning a 3D camera in order to support the reader examination, it is not possible since the colon mucosa is not precisely determined and since the disclosed modalities are not based on, nor do they emit, the precise segmentation of the mucosa of the colon, and thus those of the anomalies in potentials. However, it is possible to support the review of radiologists by defining this maximum final location as the center of rotation for 3D cameras based on the fact that these final local maximums are within the anomalies in potentials, specifically centered on their densest regions.
[000160] In yet an additional modality, and referring to Fig 19, an inverse problem is carried out in the rays passing through each of the maximum final location or its cluster, if close enough. This reverse problem is aimed at the reconstruction of at least a part of the colon mucosa. Qualified professionals will appreciate that multiple inverse approaches can be used such as an implicit surface problem based on distance information from singular points that are defined as "most likely accurate". In an additional embodiment, these stretches of surfaces are used to visually support an analysis of the potential anomaly describing specific colors in either 2D or 3D to mark such regions. These modalities allow to circumvent the issue that it was not possible to mark the potential anomaly during image production because no precise segmentation of the colon is performed.
[000161] Figure 19 illustrates a two-dimensional image originating from a CT Colonography exam. The main image shows two regions under investigation (the two protruding regions), characterized by the fact that one represents a local maxim. From that local maxim, an inverse problem is performed in order to determine a better implicit surface representing a reconstructed colon mucosa in that specific region. Such an implicit surface is shown in the left enlarged image on which one can see the coarse pixels of the original image, and the implicit surface of the refined reconstructed colon mucosa that can be used to determine the regions' volume and thickness of interest.
[000162] In one embodiment, the reconstructed anomaly stretches are used to determine an anomaly approach measurement.
[000163] Finally, although the qualified professionals will appreciate that the previous modalities allows to overcome the limitations of a scheme that does not involve any traditional and precise segmentation process, specifically in supporting the visual examination of potential anomalies, the qualified professionals also they will appreciate that these modalities can be used without the cutting plan modalities.
[000164] Although the above description refers to the specific preferred modalities as presently contemplated by the inventor, it will be understood that the invention in its broad aspect includes mechanical and functional equivalents of the elements described herein.

权利要求:
Claims (18)
[0001]
1. Method implemented by computer to isolate a potential anomaly in imaging data, the method characterized by the fact that it comprises: providing a set of at least a certain anomaly property representative of a certain anomaly; provide an anomaly property identifier to identify each of the at least one particular anomaly property; in imaging data, isolating the first zone having a first property and a group of at least one other zone, each of at least one other zone having a corresponding different property than the first property; provide a transition zone resulting from the isolation of a first zone and a group of at least one other zone, the transition zone being selected from a group consisting of: a closed zone separating the first zone and the group of at least one another zone; and a closed zone extending into one of the first zone and the group of at least one other zone; applying the anomaly property identifier to identify each of the at least one particular anomaly property in the at least transition zone to provide a computed indication for the selected zone, the selected zone being at least the transition zone; determine whether the computed indication for the selected zone is in accordance with each of the at least one particular anomaly property; and if the computed indication for the selected zone agrees, assign an indication of a potential anomaly candidate zone for the selected zone and thereby isolate the potential anomaly mentioned.
[0002]
2. Method according to claim 1, characterized by the fact that the imaging data comprise a set of n-dimensional data originated from an imaging system, where n is greater than or equal to two.
[0003]
3. Method according to claim 2, characterized by the fact that the n-dimensional data set is one of: a two-dimensional volumetric matrix of elements and a three-dimensional volumetric matrix of elements.
[0004]
4. Method according to claim 2, characterized by the fact that the n-dimensional data set originates from a device selected from a group consisting of a magnetic resonance imaging (MRI) device, a high resolution tomography device. positron emission (PET), an X-ray device, an ultrasound device and any combination thereof.
[0005]
5. Method according to claim 1, characterized by the fact that the set of at least one particular anomaly property comprises at least one of: information related to composition, information related to shape, spatial location in imaging data, and one combination of them over time.
[0006]
6. Method according to claim 1, characterized by the fact that the anomaly property identifier comprises at least one of: determination of tissue density, determination of tissue homogeneity gradient, determination of absence or presence of tissue properties , determination of water content / distribution, determination of presence and determination of a distribution of contrast agent at a given time or over time.
[0007]
Method according to claim 1, characterized by the fact that the first property of the first zone comprises a certain region of air, in which, in addition, the corresponding property of each of the at least one other zone comprises a certain region of tissue.
[0008]
8. Method according to claim 1, characterized by the fact that the first property of the first zone comprises a certain marked region, further characterized by the fact that the corresponding property of each of the at least one other zone comprises a certain region of tissue.
[0009]
9. Method according to claim 1, further characterized by the fact that it includes applying the anomaly property identifier to the selected zone.
[0010]
10. Method according to claim 1, further characterized by the fact that it comprises providing an indication of a potential anomaly.
[0011]
11. Method according to claim 10, characterized in that the provision of an indication of a potential anomaly comprises at least one of: storing the indication of a potential anomaly and displaying the indication of a potential anomaly in a user interface.
[0012]
12. Method according to claim 10, further characterized by the fact that it comprises transmitting the indication of a potential anomaly to a remote location.
[0013]
13. Method according to claim 1 characterized by the fact that the aforementioned image data comprises a large number of unitary image elements selected from the group consisting of pixels and voxels.
[0014]
14. Machine-readable medium characterized by the fact that it has instructions written on it to carry out the method for isolating a potential anomaly in imaging data as claimed in any of claims 1 to 13.
[0015]
15. Method to isolate a potential anomaly in imaging data, the method characterized by the fact of understanding: receiving imaging data; isolating in imaging data a first zone having a first property and a group of at least one other zone, each of the at least one other zone having a corresponding property different from the first property; provide a transition zone resulting from the isolation of a first zone and a group of at least one other zone, the transition zone being selected from a group consisting of: a closed zone separating the first zone and the group of at least one another zone; and a closed zone extending into one of the first zone and the group of at least one other zone; applying a tissue homogeneity gradient identifier to the at least transition zone to provide a computed indication for a selected zone, the selected zone being at least the transition zone; determine if the computed indication is in accordance with the selected zone; and if the computed indication for the selected zone agrees, assign an indication of a potential anomaly candidate zone for the selected zone and thereby isolate the potential anomaly mentioned.
[0016]
16. System to isolate a potential anomaly in imaging data, the system characterized by the fact that it comprises: a data collector; a central processing unit, operatively connected to the data collector; an I / O device, operatively connected to the data collector; a network interface circuit, operatively connected to the data collector; and a memory, operatively connected to the data collector, the memory comprising at least one program to isolate a potential anomaly in imaging data, in which at least one program is configured to be executed by the central processing unit, o at least one program to isolate a potential anomaly in imaging data comprising: instructions for providing an anomaly property identifier to identify each of the at least one particular anomaly property; instructions for isolating, in imaging data, the first zone having a first property and a group of at least one other zone, each of the at least one other zone having a corresponding different property than the first property; instructions for providing a transition zone resulting from the isolation of the first zone and a group of at least one other zone, the transition zone being selected from a group consisting of a closed zone separating the first zone and the group of at least one another zone and a closed zone extending into one of the first zone and the group of at least one other zone; instructions for applying the anomaly property identifier to identify each of the at least one particular anomaly property in the at least transition zone to provide a computed indication for a selected zone, the selected zone being at least the transition zone; instructions to determine if the computed indication for the selected zone conforms to each of the at least one particular anomaly property and if the computed indication for the selected zone agrees, assign a potential anomaly candidate indication for the selected zone and thereby isolate the potential anomaly mentioned.
[0017]
17. System for isolating a potential anomaly in imaging data according to claim 16, characterized by the fact that the memory additionally comprises the imaging data.
[0018]
18. System for isolating a potential anomaly according to claim 16, characterized by the fact that the imaging data is received from the network interface circuit.

类似技术:

---

公开号 | 公开日 | 专利标题

---

BR112013021657B1|2021-03-16|METHOD IMPLEMENTED BY COMPUTER AND SYSTEM TO ISOLATE A POTENTIAL ANOMALY IN IMAGE DATA, AND, MACHINE-READABLE MEDIA

---

Tan et al.2013|Segmentation of lung lesions on CT scans using watershed, active contours, and Markov random field

---

Vandemeulebroucke et al.2012|Automated segmentation of a motion mask to preserve sliding motion in deformable registration of thoracic CT

---

US7876947B2|2011-01-25|System and method for detecting tagged material using alpha matting

---

US7840051B2|2010-11-23|Medical image segmentation

---

EP1908014A1|2008-04-09|Abnormality detection in medical images

---

US8515200B2|2013-08-20|System, software arrangement and method for segmenting an image

---

Abdolali et al.2016|Automatic segmentation of maxillofacial cysts in cone beam CT images

---

Radaelli et al.2010|On the segmentation of vascular geometries from medical images

---

Sun et al.2014|Juxta-vascular nodule segmentation based on flow entropy and geodesic distance

---

Cheirsilp et al.2015|Thoracic cavity definition for 3D PET/CT analysis and visualization

---

Van Ravesteijn et al.2013|Electronic cleansing for 24-h limited bowel preparation CT colonography using principal curvature flow

---

El-Bazl et al.2003|Automatic identification of lung abnormalities in chest spiral CT scans

---

Macho et al.2018|Segmenting Teeth from Volumetric CT Data with a Hierarchical CNN-based Approach.

---

Reinhardt et al.2000|Pulmonary imaging and analysis

---

Manikandarajan et al.2013|Detection and segmentation of lymph nodes for lung cancer diagnosis

---

Zhao et al.2015|An effective brain vasculature segmentation algorithm for time-of-flight MRA data

---

EP3889896A1|2021-10-06|Model-based virtual cleansing for spectral virtual colonoscopy

---

Memon et al.2008|Deficiencies of Lung segmentation techniques using CT scan images for CAD

---

US11160528B2|2021-11-02|System and method for visualization of ultrasound volumes

---

Tran et al.2016|Liver segmentation and 3d modeling from abdominal ct images

---

Ben-Zikri2017|Development, Implementation and Pre-clinical Evaluation of Medical Image Computing Tools in Support of Computer-aided Diagnosis: Respiratory, Orthopedic and Cardiac Applications

---

Li et al.2012|PET-guided liver segmentation for low-contrast CT via regularized Chan-Vese model

---

Yigitsoy et al.2012|Dynamic graph cuts for colon segmentation in functional cine-MRI

---

Wang et al.2009|PET-enhanced liver segmentation for CT images from combined PET-CT scanners

同族专利:

---

公开号 | 公开日

---

US9717414B2|2017-08-01|

---

AU2012220301A1|2013-09-12|

---

CA2827742A1|2012-08-30|

---

EP2678827A4|2017-10-25|

---

JP6031454B2|2016-11-24|

---

KR101967357B1|2019-04-09|

---

KR20140009407A|2014-01-22|

---

CN103501699A|2014-01-08|

---

BR112013021657A2|2017-07-04|

---

WO2012113069A1|2012-08-30|

---

US20140010430A1|2014-01-09|

---

CN103501699B|2016-01-13|

---

HK1190903A1|2014-07-18|

---

EP2678827A1|2014-01-01|

---

CA2827742C|2018-07-17|

---

JP2014507231A|2014-03-27|

---

AU2012220301B2|2017-04-13|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

---

US5920319A|1994-10-27|1999-07-06|Wake Forest University|Automatic analysis in virtual endoscopy|

---

US6331116B1|1996-09-16|2001-12-18|The Research Foundation Of State University Of New York|System and method for performing a three-dimensional virtual segmentation and examination|

---

US6549646B1|2000-02-15|2003-04-15|Deus Technologies, Llc|Divide-and-conquer method and system for the detection of lung nodule in radiological images|

---

US6947784B2|2000-04-07|2005-09-20|The General Hospital Corporation|System for digital bowel subtraction and polyp detection and related techniques|

---

US7113617B2|2000-12-12|2006-09-26|Hewlett-Packard Development Company, L.P.|Method of computing sub-pixel Euclidean distance maps|

---

US7236620B1|2002-06-24|2007-06-26|Icad, Inc.|Computer-aided detection methods in volumetric imagery|

---

US7260250B2|2002-09-30|2007-08-21|The United States Of America As Represented By The Secretary Of The Department Of Health And Human Services|Computer-aided classification of anomalies in anatomical structures|

---

US7346209B2|2002-09-30|2008-03-18|The Board Of Trustees Of The Leland Stanford Junior University|Three-dimensional pattern recognition method to detect shapes in medical images|

---

JP4326563B2|2003-03-11|2009-09-09|シーメンスメディカルソリューションズユーエスエーインコーポレイテッド|Computer and computer-readable recording medium|

---

US7333644B2|2003-03-11|2008-02-19|Siemens Medical Solutions Usa, Inc.|Systems and methods for providing automatic 3D lesion segmentation and measurements|

---

US7369638B2|2003-07-11|2008-05-06|Siemens Medical Solutions Usa, Inc.|System and method for detecting a protrusion in a medical image|

---

US7447342B2|2003-09-22|2008-11-04|Siemens Medical Solutions Usa, Inc.|Method and system for using cutting planes for colon polyp detection|

---

US7440601B1|2003-10-10|2008-10-21|The United States Of America As Represented By The Department Of Health And Human Services|Automated identification of ileocecal valve|

---

JP4686279B2|2005-07-06|2011-05-25|株式会社東芝|Medical diagnostic apparatus and diagnostic support apparatus|

---

US20090016583A1|2007-07-10|2009-01-15|Siemens Medical Solutions Usa, Inc.|System and Method for Detecting Spherical and Ellipsoidal Objects Using Cutting Planes|

---

AT511683T|2007-12-28|2011-06-15|Im3D S P A|CLASSIFICATION OF MARKED MATERIAL IN A SET OF TOMOGRAPHIC PICTURES OF THE COLORECTAL REGION|

---

US8131036B2|2008-07-25|2012-03-06|Icad, Inc.|Computer-aided detection and display of colonic residue in medical imagery of the colon|

---

US20100183210A1|2009-01-22|2010-07-22|Van Uitert Robert L|Computer-assisted analysis of colonic polyps by morphology in medical images|

---

CN102763133B|2009-11-27|2016-06-15|卡丹医学成像股份有限公司|For the method and system of filtering image data and the use in virtual endoscopy thereof|

---

EP2503940B1|2009-11-27|2020-09-02|Cadens Medical Imaging Inc.|Method and system for determining an estimation of a topological support of a tubular structure and use thereof in virtual endoscopy|

---

US8724894B1|2012-12-03|2014-05-13|Rockwell Collins, Inc.|Colorization of digital imagery|JP6235886B2|2013-01-08|2017-11-22|キヤノン株式会社|Biological tissue image reconstruction method and apparatus, and image display apparatus using the biological tissue image|

---

US9832447B2|2013-04-04|2017-11-28|Amatel Inc.|Image processing system and image processing program|

---

US9940511B2|2014-05-30|2018-04-10|Kofax, Inc.|Machine print, hand print, and signature discrimination|

---

US11213220B2|2014-08-11|2022-01-04|Cubisme, Inc.|Method for determining in vivo tissue biomarker characteristics using multiparameter MRI matrix creation and big data analytics|

---

US10133614B2|2015-03-24|2018-11-20|Ca, Inc.|Anomaly classification, analytics and resolution based on annotated event logs|

---

US9922433B2|2015-05-29|2018-03-20|Moira F. Schieke|Method and system for identifying biomarkers using a probability map|

---

CN105686844B|2015-12-28|2018-11-16|南京信息工程大学|Make the human brain part water distribution volume determination method of reference area in the preceding region A%|

---

CN105748093B|2015-12-28|2018-11-16|南京信息工程大学|Cerebral gray matter makees the human brain part water distribution volume determination method of reference area|

---

US10339650B2|2016-01-07|2019-07-02|Koios Medical, Inc.|Method and means of CAD system personalization to reduce intraoperator and interoperator variation|

---

US9536054B1|2016-01-07|2017-01-03|ClearView Diagnostics Inc.|Method and means of CAD system personalization to provide a confidence level indicator for CAD system recommendations|

---

EP3479350A4|2016-07-01|2020-08-19|Cubisme, Inc.|System and method for forming a super-resolution biomarker map image|

---

US10346982B2|2016-08-22|2019-07-09|Koios Medical, Inc.|Method and system of computer-aided detection using multiple images from different views of a region of interest to improve detection accuracy|

---

US11232853B2|2017-04-21|2022-01-25|Cubisme, Inc.|System and method for creating, querying, and displaying a MIBA master file|

---

US10593033B2|2017-06-27|2020-03-17|Nec Corporation|Reconstructor and contrastor for medical anomaly detection|

---

EP3527137A1|2018-02-14|2019-08-21|Koninklijke Philips N.V.|Apparatus for iterative material decomposition of multispectral data|

---

CA3047972A1|2018-06-25|2019-12-25|The Royal Institution For The Advancement Of Learning |Method and system of performing medical treatment outcome assessment or medical condition diagnostic|

法律状态:
2018-12-18| B06F| Objections, documents and/or translations needed after an examination request according [chapter 6.6 patent gazette]|

---

2019-10-22| B06U| Preliminary requirement: requests with searches performed by other patent offices: procedure suspended [chapter 6.21 patent gazette]|

---

2021-01-05| B09A| Decision: intention to grant [chapter 9.1 patent gazette]|

---

2021-03-16| B16A| Patent or certificate of addition of invention granted|Free format text: PRAZO DE VALIDADE: 20 (VINTE) ANOS CONTADOS A PARTIR DE 24/02/2012, OBSERVADAS AS CONDICOES LEGAIS. |

优先权:

---

申请号 | 申请日 | 专利标题

---

US201161446342P| true| 2011-02-24|2011-02-24|

---

US61/446342|2011-02-24|

---

PCT/CA2012/000172|WO2012113069A1|2011-02-24|2012-02-24|Method and apparatus for isolating a potential anomaly in imaging data and its application to medical imagery|

---