System and method for automatic detection and segmentation of vertebrae in 3D images

Abstract

This article proposes a system and method for identifying and segmenting the vertebral body of a single vertebra in a 3D image (e.g., CT or micro-CT image) of a subject. In certain embodiments, the method described herein identifies multiple discrete and distinguishable regions in a graphical representation of a single vertebra in a 3D image of a subject, one of which corresponds to the vertebral body of the single vertebra. The region corresponding to the vertebral body can be classified automatically or manually (e.g., via user interaction) in this way. Identifying vertebral bodies in this way helps streamlined quantitative analysis of 3D images for orthopedic research, especially providing a basis for quickly and consistently evaluating vertebral morphological traits.

Description

Technical Field

[0001] The present invention generally relates to methods and systems for image processing and analysis. More specifically, in certain embodiments, the present invention relates to automatic detection and / or segmentation of vertebral bodies in anatomical images of small subjects (e.g., small animals; e.g., small mammals), such as captured with a computer tomography (CT) scanner. Background Art

[0002] There are a wide variety of techniques involved in in vivo imaging of mammals, such as bioluminescence, fluorescence, tomography, and multimodal imaging techniques. In vivo imaging of small mammals is performed by a large number of researchers in various fields such as oncology, infectious diseases, and drug development.

[0003] In vivo micro-computed tomography (hereafter "micro-CT") imaging is an x-ray-based technique that can image tissues, organs, and non-organic structures with high resolution, although higher throughput imaging can take advantage of lower resolution micro-CT imaging to speed up image acquisition and / or processing while maintaining acceptable accuracy and image detail. Micro-CT is rapidly evolving, and low-dose scanning and rapid imaging protocols are needed to facilitate multimodal applications and enable longitudinal experimental models. In vivo imaging typically involves the non-invasive spatiotemporal visualization of biological phenomena within living subjects using agents (such as fluorescent probes). Multimodal imaging involves the fusion of images acquired in different ways, for example, by combining FMT, PET, MRI, CT, and / or SPECT imaging data.

[0004] Image analysis applications and / or imaging systems generally allow for visualization, analysis, processing, segmentation, registration, and measurement of biomedical images. These applications and systems also provide volume rendering tools (e.g., volume synthesis, depth shading, gradient shading, maximum intensity projection, cumulative voxel projection, signal projection); manipulation functions (e.g., to define regions of interest, remove unwanted objects, edit images and object maps); and measurement functions (e.g., for calculating the number of surface voxels, the number of exposed faces, the planar area of ​​a region, and the estimated surface area or volume of a region).

[0005] Image segmentation techniques are often used to identify separate image regions corresponding to different structures, organs, and / or tissues of interest. Accurate and robust image segmentation can be challenging if the different structures of interest are similar in nature and / or are found in close proximity to each other. Specifically, while segmenting representations of individual bones (e.g., to distinguish between individual bones) is inherently difficult, further segmenting individual bones in order to distinguish between their various subregions can present an even greater challenge. For example, while individual bones are naturally physically separated from each other at joints, the physical structural divisions between various subregions of a particular bone are often not clearly discernible. However, further segmenting individual bones in this manner (e.g., to distinguish between various subregions of a particular bone) is useful for imaging methods for the study and / or diagnosis of bone formation, injury, and disease.

[0006] For example, orthopedic research often involves quantitative analysis of bone morphometric attributes. Research focused on vertebral formation, spinal injuries, and diseases such as degenerative disc disease and osteoporosis measures the morphometric attributes of vertebrae and specific sub-regions thereof to assess, for example, disease status and / or progression, severity of injury, and the like. Measurement and analysis of vertebral morphometric attributes often focus on a specific portion of each vertebra, referred to as the vertebral centrum or vertebral body. The vertebral body is the thick oval central portion of an individual vertebra, comprising cancellous bone tissue surrounded by a protective layer of compact bone, which forms a cortical compartment. Structures known as pedicles protrude from each side of the vertebral body and combine with the lamina to form the vertebral arch. The vertebral body is the main load-bearing structure in the vertebrae and is susceptible to compression fractures, particularly in patients with degenerative diseases such as osteoporosis. Therefore, measurement and analysis of the vertebral body region of the vertebrae is particularly important for orthopedic research and / or diagnosis.

[0007] Ex vivo and / or in vivo bone morphometric measurements are often obtained using micro-CT imaging, which can provide adequate contrast between bone and soft tissue. Example micro-CT images of multiple vertebrae are shown in FIG. Figure 1A and Figure 1B As shown. Figure 1B In the figure, the vertebral bodies of each of the three vertebrae were manually identified. Analyzing micro-CT images to measure the morphological properties of the vertebral bodies can provide useful insights into the diagnosis, status, and progression of disease and / or injury in a subject, as well as analyzing the efficacy of different treatments. However, to date, image-based analysis of vertebral morphological properties has relied on manual identification of vertebral bodies (e.g., via user-drawn hand-drawn boundaries, such as Figure 1B shown).

[0008] Therefore, there is a need for improved systems and methods for automatically segmenting individual bones into their component sub-regions. In particular, there is a need for systems and methods that can automatically identify the vertebral bodies of individual vertebrae. Summary of the invention

[0009] This article proposes a system and method for identifying and segmenting the vertebral bodies of individual vertebrae in a 3D image (e.g., CT or micro-CT image) of a subject. In certain embodiments, the method described herein identifies multiple discrete and distinguishable regions in a graphical representation of individual vertebrae in a 3D image of a subject, one of which corresponds to the vertebral bodies of individual vertebrae. The regions corresponding to the vertebral bodies can be classified automatically or manually (e.g., via user interaction) in this way. Identifying vertebral bodies in this way helps streamlined quantitative analysis of 3D images for orthopedic research, especially providing a basis for quickly and consistently evaluating vertebral morphological traits.

[0010] In certain embodiments, to provide accurate and robust identification and segmentation of vertebral regions in images, the methods described herein utilize a series of image processing steps that take into account and exploit insights about the specific physical structure of individual vertebrae and their vertebral bodies.

[0011] In certain embodiments, a single vertebra mask is first accessed and / or generated that identifies a portion of a 3D image corresponding to a particular individual vertebra of interest. A series of specific image processing steps are then applied to the single vertebra mask to isolate a vertebral body subregion corresponding to the vertebral body of the vertebra of interest. The specific image processing steps used take advantage of insights into the specific physical geometry of the vertebrae and the manner in which the vertebral body is physically distinguished from other regions of the individual vertebrae. The methods herein include the following steps that not only take advantage of the manner in which these physical features are represented in the image to provide segmentation, but also address a source of significant error caused by image features that correspond to certain physical structures of the vertebrae.

[0012] Specifically, Figure 2A and Figure 2B The appearance and cross-section of individual vertebrae are shown. As described herein, and Figure 2A As shown in , from the outside, the vertebral body appears to be a thick, oval-shaped solid bone structure with narrower pedicles protruding from it. The combination of a distance transform and a subsequent watershed segmentation operation can be used to subdivide it at points where the graphical representation, such as the mask, becomes narrow. Therefore, such a combination of steps provides the potential to separate the vertebral body region from other regions of a single vertebral mask based on the apparent thickness of the vertebral body compared to the narrower connecting regions connecting the vertebral body to other parts of the vertebra.

[0013] However, if Figure 2A and Figure 2BAs shown, although the vertebral body 202 appears solid from the outside, its interior (e.g., trabecular portion) 252 (not to be confused with the neural tube 204) includes fine structures and cavities occupied by bone marrow and soft tissue. Therefore, in some embodiments, a single vertebra mask generated from a 3D image to identify individual vertebrae has a hollow shell-like structure that represents an image area corresponding to solid bone, while areas corresponding to soft tissue and bone marrow are typically omitted. In addition, a single vertebra mask often includes perforations extending between the interior of the mask (e.g., a cavity corresponding to bone marrow and / or soft tissue areas) and the exterior. These perforations physically correspond to common physical structures such as blood vessels within bones, as well as other physical structures such as tumors and / or cracks. Although generally less common, features such as tumors and cracks may be present in the vertebrae of a subject imaged for orthopedic applications associated with analyzing and / or diagnosing certain diseases or injuries.

[0014] In certain embodiments, hollow cavities and / or perforations in a single vertebra mask prevent the above-described distance transform and watershed segmentation operations from accurately and robustly detecting and segmenting the vertebral regions of the single vertebra mask. Specifically, perforations and hollow regions within a single vertebra mask can produce many narrow regions within the vertebral region itself. Because the vertebral region itself has many narrow features, the narrow connections that also exist between the vertebral body and other vertebral regions cannot provide specificity for the separation between the vertebral body and other regions. Therefore, distance transform and watershed segmentation operations applied to masks such as masks can produce significant over-segmentation errors, indistinguishably subdividing the single vertebra mask at narrow features within the vertebral region in addition to at the connections between the vertebral body region and other regions.

[0015] Therefore, in certain embodiments, to address this challenge, the vertebral segmentation methods described herein utilize a filling step that artificially fills in perforations and internal (e.g., trabeculae) regions in individual vertebrae masks, such as Figure 2B 252 in . This approach transforms the individual single vertebrae masks from a shell-like structure to a solid structure - a filled single vertebrae mask. Applying the distance transform and watershed segmentation steps to the filled single vertebrae masks, as opposed to the initial single vertebrae masks, enables them to successfully exploit the narrow connections between the vertebral bodies of individual vertebrae and other regions to accurately and robustly detect and segment the vertebral body regions. Thus, including such a filling step prior to performing the distance transform and watershed segmentation steps takes into account the unique physical geometry of the vertebrae and avoids over-segmentation errors that would otherwise result from hollow and / or perforated initially obtained (e.g., generated; e.g., accessed) single vertebrae masks.

[0016] In certain embodiments, the identified vertebral body region can be used to quantitatively measure the volume, surface (e.g., surface area), connectivity, and other morphological traits of the trabecular and cortical compartments of the vertebral body of the vertebra. Such measurements are used as valuable metrics for, for example, assessing the disease state of a subject, and can be repeated over time to assess disease progression and therapeutic efficacy. For example, automatic quantification of trabecular volume in longitudinal studies can provide insight into the efficacy of different treatments for vertebral osteoporosis.

[0017] It is noteworthy that previous methods for measuring morphological traits have relied on manual identification of vertebrae (e.g., via Figure 1B The hand-drawn boundaries shown in the figure are cumbersome and prone to human error and inconsistency. In contrast, by automatically identifying only the vertebral subregions that require classification at most [e.g., via a single "affirmative" click (e.g., as a mouse-based interface) or tap (e.g., via a touch-sensitive interface) by the user], the systems and methods provided herein greatly streamline image analysis, thereby enabling more accurate and consistent analysis to be performed quickly. By improving the accuracy and rate of analysis in this way, the systems and methods described herein provide a valuable tool for orthopedic research and diagnosis.

[0018] In one aspect, the present invention relates to a method for automatically detecting and segmenting vertebral bodies of a particular vertebra in a 3D image of a subject (e.g., an anatomical image of the subject), the method comprising: (a) receiving, by a processor of a computing device, a 3D image of the subject [e.g., wherein the image is an anatomical image (e.g., a CT image, such as a micro-CT image)], wherein the 3D image includes a graphical representation of one or more vertebral portions of the subject; (b) accessing and / or generating, by the processor, a single vertebral mask that identifies a portion of the graphical representation determined to correspond to the particular vertebra [e.g., wherein the single vertebral mask is a binary mask that includes a plurality of voxels, each single vertebral mask voxel corresponding to the received 3D image of the subject (c) applying, by a processor, one or more morphological operations (e.g., morphological dilation; e.g., morphological hole filling; e.g., morphological erosion) to fill perforations and / or one or more interior regions of the single vertebra mask, thereby generating a filled single vertebra mask; (d) determining, by the processor, a distance map by applying a distance transform to the filled single vertebra mask [e.g., wherein the distance map includes a plurality of distance map voxels, each distance map voxel corresponding to a voxel of the filled single vertebrae mask and having (e.g., being assigned) a distance value representing a distance from the voxel to a nearest boundary and / or non-skeletal voxel (e.g., a voxel of the filled single vertebrae mask having a value of 0)]; (e) applying, by the processor, a watershed segmentation operation to the distance map to identify a set of catchment basins from the distance map [e.g., by dividing the distance map into a plurality of catchment basins separated from one another by watershed lines; for example, wherein the watershed mask resulting from the watershed segmentation operation includes a plurality of catchment basins (e.g., each catchment basin corresponds to a connected region of voxels assigned a first value such as a digital 1 or a Boolean 'true'), which are separated from one another by watershed lines (e.g., each watershed line corresponds to a connecting line of voxels assigned a second value, such as a digital 0 or a Boolean 'true')]; (f) determining, by the processor, a labeled intersegmental vertebral map comprising a plurality of labeled regions using the watershed set and the individual vertebral mask, one of the labeled regions corresponding to a vertebral body [e.g., the labeled intersegmental vertebral map corresponds to a labeled version of the individual vertebral mask, wherein portions of the individual vertebral mask that are within different watersheds of the watershed set are identified (e.g., by performing a logical AND operation on each watershed of the watershed set and the individual vertebral mask) and are labeled accordingly to distinguish them from one another]; and (g) rendering, by the processor, a graphical representation of the labeled intersegmental vertebral map [e.g., for display to a user;For example, where the graphical representation visually distinguishes differently labeled regions of the labeled vertebral map (e.g., using different colors, shading, etc.)].

[0019] In certain embodiments, step (b) includes segmenting, by a processor, the 3D image to generate individual vertebrae masks.

[0020] In some embodiments, step (b) includes: segmenting the 3D image by a processor to generate a labeled (segmented) skeletal map including a plurality of labeled regions that distinguish portions of the graphical representation corresponding to individual bones (e.g., including but not limited to one or more vertebral portions; for example, each labeled region of the labeled (segmented) skeletal map corresponds to a portion of the graphical representation determined to correspond to a particular individual bone); rendering by the processor a graphical representation of the labeled (segmented) skeletal map [e.g., for display to a user; for example, wherein the graphical representation visually distinguishes differently labeled regions of the labeled (segmented) skeletal map (e.g., using different colors, shading, etc.)]; receiving by the processor a user selection of at least one of the plurality of labeled regions; and generating by the processor a single vertebral mask from the user-selected labeled regions.

[0021] In certain embodiments, segmenting a 3D image includes applying one or more second-order derivative splitting filters to the 3D image [e.g., applying one or more second-order derivative splitting filters to the image to produce a split bone mask of the image with bone boundaries removed; determining a plurality of split binary components of the split bone mask by performing one or more morphological processing operations; and performing a region growing operation using the split binary components of the split bone mask as seeds to produce a labeled (segmented) bone map comprising a plurality of labeled regions that distinguish individual bones in the 3D image].

[0022] In certain embodiments, at least a portion of the single vertebrae mask is located on an edge of the 3D image, and the method includes filling an interior of the portion of the single vertebrae mask that is located on the edge of the 3D image.

[0023] In some embodiments, step (c) includes: applying, by the processor, a morphological dilation operation to grow the single vertebrae mask (e.g., to fill perforations in the single vertebrae mask) to generate an expanded single vertebrae mask; and applying, by the processor, a morphological hole filling operation to the expanded single vertebrae mask to fill one or more internal regions within the expanded single vertebrae mask to generate a filled single vertebrae mask.

[0024] In certain embodiments, the method includes refining the filled single vertebrae mask by performing a morphological erosion operation by a processor (e.g., using a morphological erosion element having a size that is the same and / or approximately equal to a size of a morphological dilation element used in a morphological dilation operation).

[0025] In some embodiments, a morphological dilation operation uses an expansion element having a size that is preset and / or automatically determined based on the resolution of the 3D image [e.g., such that the expansion element size corresponds to a particular physical size (e.g., from 100 to 240 microns along each dimension) based on (e.g., approximately equal to; e.g., slightly larger than) one or more physical features associated with a cavity extending from the exterior to the interior of a vertebra (e.g., a blood vessel within the vertebra)].

[0026] In some embodiments, the method includes receiving, by a processor, user input of an expansion element size value, and using the user-input expansion element size when applying a morphological expansion operation (e.g., allowing a user to enlarge the expansion element size to account for unusual features, such as cracks, tumors, etc., in imaged vertebrae).

[0027] In certain embodiments, the method includes: (h) after step (g), receiving, by the processor via a graphical user interface (GUI), a user selection of a marked area of ​​the intersegmental vertebral map corresponding to a vertebral body; and (i) determining, by the processor, a vertebral body area of ​​the intersegmental vertebral map that corresponds to the user selection [e.g., (A) by marking, by the processor, the marked area selected by the user as corresponding to the vertebral body (e.g., and marking, by the processor, the remaining marked areas as corresponding to other areas of the vertebra), thereby generating a marked vertebral map (e.g., a binary map) that distinguishes areas of a single vertebral mask corresponding to the vertebral body from other areas of the single vertebral mask; e.g., (B) by generating, by the processor, a vertebral body mask that identifies the marked area selected by the user].

[0028] In certain embodiments, the method includes determining, by the processor, one or more morphometric measurements using the determined vertebral body regions (eg, for diagnostic purposes; eg, for determining efficacy of a treatment).

[0029] In certain embodiments, the one or more morphometric measurements include measurements of one or more morphometric traits of the trabecular and / or cortical portions of the vertebral body (eg, the volume of the trabecular portion of the vertebral body).

[0030] In certain embodiments, the 3D image of the subject is a CT image (eg, a micro-CT image), and the method includes acquiring the CT image (eg, a micro-CT image).

[0031] In another aspect, the present invention relates to a system for automatically detecting and segmenting vertebral bodies of a particular vertebra in a 3D image of a subject (e.g., an anatomical image of the subject), the system comprising: a processor of a computing device; and a memory having instructions stored thereon, wherein the instructions, when executed by the processor, cause the processor to: (a) receive a 3D image of the subject [e.g., wherein the image is an anatomical image (e.g., a CT image, such as a micro-CT image)], wherein the 3D image includes a graphical representation of one or more vertebral portions of the subject; (b) access and / or generate a single vertebral mask, wherein the single vertebral mask identifies a portion of the graphical representation determined to correspond to the particular vertebra [e.g., wherein the single vertebral mask is a portion of the graphical representation determined to correspond to the particular vertebra [e.g., wherein the single vertebral mask is a portion of the graphical representation determined to correspond to the particular vertebra (c) applying one or more morphological operations (e.g., morphological dilation; e.g., morphological hole filling; e.g., morphological erosion) to fill perforations and / or one or more interior regions of the single vertebra mask, thereby generating a filled single vertebra mask; and (d) determining a distance map by applying a distance transform to the filled single vertebra mask. [e.g., wherein the distance map comprises a plurality of distance map voxels, each distance map voxel corresponding to a voxel of a filled single vertebra mask and having (e.g., being assigned) a distance value representing a distance from the voxel to a nearest boundary and / or non-bone voxel (e.g., a voxel of the filled single vertebra mask having a value of 0)]; (e) applying a watershed segmentation operation to the distance map to identify a set of catchment basins from the distance map [e.g., by dividing the distance map into a plurality of catchment basins separated from each other by watershed lines; for example, wherein a watershed mask resulting from the watershed segmentation operation comprises a plurality of catchment basins (e.g., each catchment basin corresponds to a connected region of voxels assigned a first value such as a number 1 or a Boolean 'true'), which are determined by the watershed segmentation operation. (e.g., each watershed line corresponds to a connecting line of voxels assigned a second value such as a digital 0 or a Boolean 'true')]; (f) determining a labeled intersegmental vertebral map including a plurality of labeled regions using the watershed set and the single vertebral mask, one of the labeled regions corresponding to a vertebral body [e.g., the labeled intersegmental vertebral map corresponds to a labeled version of the single vertebral mask, wherein portions of the single vertebral mask that are within different watersheds of the watershed set are identified (e.g., by performing a logical AND operation on each watershed of the watershed set and the single vertebral mask) and are labeled accordingly to distinguish them from each other]; and (g) rendering a graphical representation of the labeled intersegmental vertebral map [e.g., for display to a user;For example, where the graphical representation visually distinguishes differently labeled regions of the labeled vertebral map (e.g., using different colors, shading, etc.)].

[0032] In certain embodiments, in step (b), the instructions cause the processor to segment the 3D image to generate individual vertebrae masks.

[0033] In certain embodiments, in step (b), the instructions cause the processor to: segment the 3D image to generate a labeled (segmented) skeletal map comprising a plurality of labeled regions that distinguish portions of the graphical representation corresponding to individual bones (e.g., including but not limited to one or more vertebral portions; e.g., each labeled region of the labeled (segmented) skeletal map corresponds to a portion of the graphical representation determined to correspond to a particular individual bone); render a graphical representation of the labeled (segmented) skeletal map [e.g., for display to a user; e.g., wherein the graphical representation visually distinguishes differently labeled regions of the labeled (segmented) skeletal map (e.g., using different colors, shading, etc.)]; receive a user selection of at least one of the plurality of labeled regions; and generate a single vertebral mask from the user-selected labeled regions.

[0034] In certain embodiments, the instructions cause the processor to segment a 3D image by: applying one or more second-order derivative splitting filters to the 3D image [e.g., applying one or more second-order derivative splitting filters to the image to produce a split bone mask of the image with bone boundaries removed; determining a plurality of split binary components of the split bone mask by performing one or more morphological processing operations; and performing a region growing operation using the split binary components of the split bone mask as seeds, thereby producing a labeled (segmented) bone map comprising a plurality of labeled regions that distinguish individual bones in the 3D image].

[0035] In certain embodiments, at least a portion of the single vertebrae mask is located on an edge of the 3D image, and wherein the instructions cause the processor to fill in an interior of the portion of the single vertebrae mask that is located on the edge of the 3D image.

[0036] In certain embodiments, in step (c), the instructions cause the processor to: apply a morphological dilation operation to grow the single vertebra mask (e.g., to fill perforations in the single vertebra mask) to generate an expanded single vertebra mask; and apply a morphological hole filling operation to the expanded single vertebra mask to fill one or more interior regions within the expanded single vertebra mask to generate a filled single vertebra mask.

[0037] In certain embodiments, the instructions cause the processor to refine the filled individual vertebrae mask by performing a morphological erosion operation (eg, using morphological erosion elements having the same and / or approximately equal sizes to morphological dilation elements used in a morphological dilation operation).

[0038] In some embodiments, the instructions cause the processor to perform a morphological dilation operation that uses an expansion element having a size that is preset and / or automatically determined based on the resolution of the 3D image [e.g., such that the expansion element size corresponds to a specific physical size (e.g., from 100 to 240 microns along each dimension) based on (e.g., approximately equal to; e.g., slightly larger than) one or more physical features associated with a cavity extending from the exterior to the interior of a vertebra (e.g., a blood vessel within the vertebra)].

[0039] In some embodiments, the instructions cause the processor to receive user input of an expansion element size value and use the user-input expansion element size when applying a morphological expansion operation (e.g., so that a user can enlarge the expansion element size to account for unusual features such as cracks, tumors, etc. in the imaged vertebrae).

[0040] In some embodiments, the instructions cause the processor to: (h) after step (g), receive via a graphical user interface (GUI) a user selection of a marked region of the intersegmental vertebral map corresponding to a vertebral body; and (i) determine a vertebral body region of the intersegmental vertebral map that corresponds to the user selection [e.g., (A) by marking the user-selected marked region as corresponding to the vertebral body (e.g., and marking the remaining marked regions as corresponding to other regions of the vertebra), thereby generating a marked vertebral body map (e.g., a binary map) that distinguishes regions of a single vertebral mask corresponding to the vertebral body from other regions of the single vertebral mask; e.g., (B) by generating a vertebral body mask that identifies the marked region selected by the user].

[0041] In certain embodiments, the instructions cause the processor to determine one or more morphometric measurements using the determined vertebral body regions (eg, for diagnostic purposes; eg, for determining efficacy of a treatment).

[0042] In certain embodiments, the one or more morphometric measurements include measurements of one or more morphometric traits of the trabecular and / or cortical portions of the vertebral body (eg, the volume of the trabecular portion of the vertebral body).

[0043] In certain embodiments, the 3D image of the subject is a CT image (eg, a micro-CT image). BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The foregoing and other objects, aspects, features and advantages of the present disclosure will become more apparent and better understood by referring to the following description in conjunction with the accompanying drawings, wherein:

[0045] Figure 1A are grayscale micro-CT images of three lumbar vertebrae from a rat model.

[0046] Figure 1B are grayscale micro-CT images of three lumbar vertebrae from a murine model, with volume of interest (VOI) boundaries drawn (manually) around the vertebral bodies of the three lumbar vertebrae.

[0047] Figure 2A is an image corresponding to the rostral view of the mouse T2 vertebra [adapted from (IA Bab, C. Hajbi-Yonissi, Y. Gabet, and R. Müller, Micro-Tomographic Atlas of the Mouse Skeleton, New York, NY, USA. Springer, 2007; page 68)].

[0048] Figure 2B is an image corresponding to an internal view of the mouse T2 vertebra [adapted from (IA Bab, C. Hajbi-Yonissi, Y. Gabet, and R. Müller, Micro-Tomographic Atlas of the Mouse Skeleton, New York, NY, USA. Springer, 2007; p. 70)].

[0049] Figure 3 is an image showing a representation of a labeled (segmented) skeletal map distinguishing individual bones, including three identified regions that identify and distinguish three lumbar vertebrae, according to an illustrative embodiment.

[0050] Figure 4 is an image showing a representation of a labeled inter-segmental vertebral map determined using the methods described herein, according to an illustrative embodiment.

[0051] Figure 5 is a block flow diagram of a process for automatic vertebral body detection and segmentation according to an illustrative embodiment.

[0052] Figure 6 is a block flow diagram of a process for automatic vertebral body detection and segmentation according to an illustrative embodiment.

[0053] Fig. 7A is an image showing a representation of a populated single vertebral mask determined using the methods described herein, in accordance with an illustrative embodiment.

[0054] Figure 7B is an image showing a representation of a cross section of a populated single vertebral mask determined using the methods described herein, in accordance with an illustrative embodiment.

[0055] Fig. 8A are images showing representations of the results of applying a morphological dilation operation to a single vertebrae mask in accordance with an illustrative embodiment.

[0056] Figure 8B is an image showing a representation of the result of applying a morphological fill operation to a single vertebrae mask in accordance with an illustrative embodiment.

[0057] Fig. 9 is an image showing a representation of a distance map determined using the methods described herein, according to an illustrative embodiment.

[0058] Fig.10 is a block diagram of an exemplary cloud computing environment used in certain embodiments.

[0059] Fig.11 is a block diagram of an exemplary computing device and an exemplary mobile computing device used in certain embodiments.

[0060] The features and advantages of the present disclosure will become more apparent from the detailed description set forth below in conjunction with the accompanying drawings, in which like reference numerals identify corresponding elements throughout. In the accompanying drawings, the same reference numerals generally indicate identical, functionally similar, and / or structurally similar elements.

[0061] definition

[0062] In this application, unless otherwise stated, the use of "or" means "and / or". As used in this application, the term "comprising" and variations of the term, such as "comprising" and "comprising", are not intended to exclude other additives, components, integers or steps. As used in this application, the terms "about" and "approximately" are used as equivalents. Any number with or without about / approximately used in this application is intended to cover any normal fluctuations understood by a person of ordinary skill in the relevant art. In certain embodiments, the term "about" or "approximately" refers to a range of values, which is 25%, 20%, 19%, 18%, 17%, 16%, 15%, 14%, 13%, 12%, 11%, 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1% or less in either direction (greater than or less than) of the specified reference value, unless otherwise stated or apparent from the context, and unless this number would exceed 100% of the possible value.

[0063] Image: As used herein above, the term "image" (e.g., a three-dimensional image of a mammal) includes any visual representation, such as a photograph, a video frame, a streaming video, and any electronic, digital, or mathematical analog of a photograph, a video frame, or a streaming video. In certain embodiments, any of the devices described herein include a display for displaying the image or any other result produced by the processor. In certain embodiments, any of the methods described herein include a step of displaying the image or any other result produced by the method.

[0064] 3D, three-dimensional: As used herein, "3D" or "three-dimensional" when referring to an "image" refers to conveying information about three spatial dimensions. A 3D image can be presented as a data set in three dimensions and / or can be displayed as a set of two-dimensional representations or a three-dimensional display. In certain embodiments, a 3D image is represented as voxel (e.g., volume pixel) data.

[0065] Various medical imaging devices and other 3D imaging devices (e.g., computer tomography scanners (CT scanners), micro-CT scanners, etc.) output 3D images containing voxels, or convert their output into 3D images containing voxels for analysis. In some embodiments, a voxel corresponds to a unique coordinate in a 3D image (e.g., a 3D array). In some embodiments, each voxel exists in a filled or unfilled state (e.g., binary ON or OFF).

[0066] Mask: As used herein, a "mask" is a graphical pattern that identifies a 2D or 3D region and is used to control the elimination or retention of portions of an image or other graphical pattern. In certain embodiments, the mask is represented as a binary 2D or 3D image, where each pixel of the 2D image or each voxel of the 3D image is assigned to one of two values ​​of a binary value set (e.g., each pixel or voxel can be assigned 1 or 0, e.g., each pixel or voxel can be assigned a Boolean "true" or "false" value).

[0067] Second order derivative splitting filter: As used herein, applying a "second order derivative splitting filter" is an image processing operation of a 3D image (e.g., a grayscale 3D image) based on the second order derivative of intensity (or its approximation) at each voxel in a plurality of voxels. In some embodiments, the splitting filter is derived from a Gaussian second order derivative filter selected from Laplacian of Gaussian (LoG), the highest Hessian eigenvalue (HEH) with preliminary Gaussian filtering, and the lowest Hessian eigenvalue (LEH) with preliminary Gaussian filtering.

[0068] Split-line voxel: As used herein, the term "split-line voxel" refers to a voxel of a given image and / or mask that is identified and used to remove the voxel from a particular mask, thereby splitting the particular mask.

[0069] Seed: As used herein, the term "seed" refers to a group of voxels (e.g., a group of connected voxels) that is used as the initial starting region for a growing operation that expands the size of the seed until a specific stopping criterion is met. In certain embodiments, the growing operation expands the size of the seed by repeatedly adding neighboring voxels to the seed.

[0070] Label: As used herein, the term "label" refers to an identifier (eg, a computer representation of an identifier such as a text value, a numeric value, a Boolean value, etc.) that is linked to a particular region of an image.

[0071] Subject: As used herein, the term "subject" refers to an individual being imaged. In certain embodiments, the subject is a human. In certain embodiments, the subject is a small animal.

[0072] Small animals: As used herein, "small animals" refers to small mammals that can be imaged with micro-CT and / or micro-MR imagers. In some embodiments, "small animals" refers to mice, rats, voles, rabbits, hamsters, and animals of similar size.

[0073] Bone, skeletal tissue: As used herein, the terms "bone" and "skeletal tissue" refer to any bone tissue, and include, for example, normal bone and heterotopic ossification (HO).

[0074] Vertebral portion: As used herein, the term "vertebral portion" refers to a portion of an individual vertebra, including up to an entire individual vertebra (eg, a vertebral portion can be an entire individual vertebra).

[0075] Link: As used herein, the terms "link" and "linked," as in the case of a first data structure or data element being linked to a second data structure or data element, refer to a computer representation of an electronically stored (e.g., in a computer memory) association between two data structures or data elements.

[0076] Provide: As used herein, the term "provide" as in "provide data" refers to the process of transferring data between different software applications, modules, systems and / or databases. In certain embodiments, providing data includes executing instructions by a process to transfer data between software applications or between different modules of the same software application. In certain embodiments, a software application can provide data to another application in the form of a file. In certain embodiments, an application can provide data to another application on the same processor. In certain embodiments, standard protocols can be used to provide data to applications on different resources. In certain embodiments, a module in a software application can provide data to another module by passing parameters to the other module. DETAILED DESCRIPTION

[0077] It is contemplated that the systems, architectures, devices, methods, and processes of the present invention include changes and adaptations developed using information from the embodiments described herein. As contemplated by this specification, adaptations and / or modifications of the systems, architectures, devices, methods, and processes described herein may be performed.

[0078] Throughout this specification, when articles, devices, systems, and architectures are described as having, containing, or including particular components, or when processes and methods are described as having, containing, or including particular steps, it is contemplated that, in addition, there are articles, devices, systems, and architectures of the present invention that consist essentially of or consist of the described components, and that there are processes and methods according to the present invention that consist essentially of or consist of the described processing steps.

[0079] It should be understood that the order of steps or the order in which certain actions are performed is not important as long as the present invention remains operable. In addition, two or more steps or actions may be performed simultaneously.

[0080] Any reference to a publication herein (eg, in the Background section) is not an admission that the publication is prior art to any claim presented herein. The Background section is provided for clarity purposes and is not meant as a description of prior art to any claim.

[0081] As noted, the documents are incorporated herein by reference. If there is any disagreement in the meaning of a particular term, the meaning provided in the "Definitions" section above shall control.

[0082] Headings are provided for the convenience of the reader, and the presence and / or arrangement of headings are not intended to limit the scope of the subject matter described.

[0083] Described herein are systems and methods for detecting and segmenting graphical representations of vertebral bodies within 3D images. In certain embodiments, the systems and methods described herein provide a tool that receives a 3D image of a subject and uses a combination of image processing operations to identify regions of the 3D image that correspond to graphical representations of vertebral bodies of individual vertebrae.

[0084] In certain embodiments, the method described herein operates on a single vertebra mask identifying an individual vertebra, so that each single vertebra mask is further segmented into a plurality of discrete and distinguishable sub-regions, including a vertebral body sub-region (e.g., a single vertebral body sub-region). In this way, the vertebral body segmentation method described herein generates a labeled inter-segmental vertebral mask in an automated manner from a single vertebra mask. The labeled inter-segmental lamina mask comprises a plurality of labeled regions, one of which corresponds to a vertebral body region. The vertebral body region can be automatically classified (e.g., classified as corresponding to a vertebral body) or can be manually classified, e.g., via user interaction.

[0085] For example, Figure 3 Shows from Figure 1A and Figure 1B A labeled (segmented) bone map 300 generated from a 3D micro-CT image is shown. The labeled (segmented) bone map 300 includes a plurality of labeled regions that distinguish individual bones, including three labeled regions 302, 304, and 306 that identify and distinguish three lumbar vertebrae. As described herein, specific labeled regions corresponding to individual vertebrae of interest can be selected and used to generate individual vertebrae masks corresponding to the individual vertebrae of interest.

[0086] The individual vertebral masks are then analyzed to generate a labeled inter-segmental map comprising a plurality of distinguishable and labeled regions, one of which corresponds to the vertebral body of a particular vertebra of interest. Figure 4 Shows the Figure 3 304 of the present invention. The inter-segmental single vertebra map 400 is generated by labeling the middle (L4) vertebra identified by the region 304 of the segmental vertebra. The inter-segmental single vertebra map 400 includes a plurality of labeled sub-regions 402, 404, 406, 408, and 410. The sub-regions 404 correspond to vertebral bodies. As described herein, once the labeled inter-segmental vertebra map 400 is generated, the sub-regions corresponding to the vertebral bodies 404 can be classified by further automatic processing or manually through a simple streamlined user interaction (such as a single "yes" click on the vertebral body region in the representation presented on the graphical user interface (GUI).

[0087] As described herein, it is non-trivial to ensure that a labeled inter-segmental vertebral map is generated that includes sub-regions that accurately and consistently identify vertebral body sub-regions. Specifically, the vertebral body segmentation method described herein utilizes a series of image processing steps that take into account and utilize information about, for example, 2A and Figure 2B Shown are insights into the specific physical structure of individual vertebrae and their vertebral bodies.

[0088] Specifically, the vertebral segmentation method described herein utilizes a filling step that artificially fills in the perforations and internal (e.g., trabeculae) regions of individual vertebrae masks (such as Figure 2B 252) (not to be confused with the neural tube 204). The method allows for the identification of vertebral body regions 202 of the representation of individual vertebrae via a distance transform and watershed segmentation step that exploits narrow connections (e.g., the 'neck') between the vertebral body and other regions of the individual vertebrae. As described herein, the distance transform and watershed segmentation steps provide separation of the vertebral body region from other portions of the vertebral representation by identifying these narrow connections (e.g., the 'neck'), while the filling step avoids over-segmentation errors that would otherwise occur due to, for example, internal trabecular regions (such as Figure 2B This occurs because the lower density and fine substructure of the vertebrae (e.g., shown) causes the initially obtained (e.g., accessed; e.g., generated) single vertebrae mask to have a hollow shell-like structure.

[0089] Specifically, the vertebral segmentation method described herein utilizes a filling step that artificially fills in the perforations and internal (e.g., trabeculae) regions of individual vertebrae masks (such as Figure 2B 252) (not to be confused with the neural tube 204). The method allows for the identification of vertebral body regions 202 of the representation of individual vertebrae via a distance transform and watershed segmentation step that exploits narrow connections (e.g., the 'neck') between the vertebral body and other regions of the individual vertebrae. As described herein, the distance transform and watershed segmentation steps provide separation of the vertebral body region from other portions of the vertebral representation by identifying these narrow connections (e.g., the 'neck'), while the filling step avoids over-segmentation errors that would otherwise occur due to, for example, internal trabecular regions (such as Figure 2B This occurs because the lower density and fine substructure of the vertebrae (e.g., shown) causes the initially obtained (e.g., accessed; e.g., generated) single vertebrae mask to have a hollow shell-like structure.

[0090] Figure 5 An example process 500 for detecting and segmenting the vertebral bodies of individual vertebrae in an image is shown. The process 500 begins with receiving a 3D image 502 of a subject, such as a 3D micro-CT image. In certain embodiments, the 3D micro-CT image includes a plurality of voxels, each voxel representing a specific 3D volume within a region of the subject being imaged. Each voxel of the 3D image has an intensity value that provides a measure of contrast, as detected by a specific imaging modality used to obtain the 3D image. For example, the voxel intensity of the 3D micro-CT image can be represented using a Hounsfield unit value that provides a measure of the attenuation that an X-ray experiences when passing through various regions of the subject before being detected by an X-ray detector of the micro-CT detector.

[0091] In certain embodiments, the region of the subject being imaged includes various bones, including individual vertebral portions. Thus, the received 3D image includes (e.g., among other things) a graphical representation of one or more individual vertebral portions. As described herein, Figure 1A and Figure 1B Images of micro-CT images of three lumbar vertebrae of a murine model are shown. In these figures, the dark grey areas correspond to the graphical representation of the bones. Figure 1B Specific regions of the image shown in , which correspond to vertebral bodies of individual vertebrae (which have been manually identified (eg, via a user manually drawing on the image)), are outlined in the figure.

[0092] A. Segmentation of individual vertebrae and single vertebrae masks

[0093] return Figure 5 In another step 504, a single vertebra mask is accessed and / or generated 506. For example, the systems and methods described herein may access and operate on a generated single vertebra mask, which is then further segmented as described herein, or may include the step of generating a single vertebra mask, which is further segmented to generate an inter-segmental vertebra mask.

[0094] The individual vertebra mask 506 is a mask that identifies a portion of the 3D image that is determined to correspond to a particular vertebra of interest whose vertebral body is to be identified and segmented. For example, the individual vertebra mask 506 can be a binary mask that includes a plurality of voxels, each voxel corresponding to a voxel of the 3D image. Voxels identified as corresponding to a particular vertebra are assigned a first value, such as a digital 1 or a Boolean 'true', while other voxels are assigned a second value, such as a digital 0 or a Boolean 'false'.

[0095] In some embodiments, the single vertebra mask 506 is a generated single vertebra mask, and step 504 includes accessing the generated single vertebra mask. Such a single vertebra mask can be stored in a memory and accessed, for example.

[0096] In certain embodiments, step 504 includes generating individual vertebrae masks 506. A variety of methods may be used to generate individual vertebrae masks, including manual identification of specific individual vertebrae of interest, such as via user interaction where a user manually draws the boundaries of the specific vertebrae of interest.

[0097] In certain embodiments, a more streamlined and robust approach is utilized in which individual bones (including, but not limited to, individual vertebral portions) are identified within an image using automated segmentation methods.

[0098] For example, Figure 6A specific embodiment of a process 600 for vertebral body detection and segmentation is shown, which includes an additional step for generating individual vertebrae masks 506. In some embodiments, the additional step is used to automatically segment individual bones represented in a 3D image. Figure 6 In the illustrated embodiment, a thresholding operation 602 is applied to the 3D image to generate a binary bone mask 604 that identifies the region of the 3D image corresponding to the bone. For example, a first value or a second value may be assigned to the voxels of the binary bone mask based on whether the intensity of the corresponding voxel of the 3D image 502 is above or below a specific threshold. The thresholding operation 602 may use the same single threshold as a specific threshold compared to the intensity of each voxel of the 3D image, or may select a specific threshold from a plurality of thresholds, such as in a hysteresis thresholding method. In certain embodiments, when the intensity of a voxel of the 3D image is above a specific threshold, it is identified as a bone, and a first value (e.g., digital 1; e.g., Boolean 'false') is assigned to the corresponding voxel of the binary bone mask 602, and when the intensity of the voxel of the 3D image is below a specific threshold, it is identified as not corresponding to the bone, and a second value (e.g., digital 0; e.g., Boolean 'false') is assigned to the corresponding voxel of the binary bone mask 602. The thresholding method for generating a mask that distinguishes bone voxels from non-bone voxels in a 3D image is described in more detail in the following documents: U.S. patent application 14 / 812,483 filed on July 29, 2015; PCT application PCT / US15 / 42631 filed on July 29, 2015; and U.S. patent application 15 / 604,350 filed on May 24, 2017, the contents of each of which are incorporated herein by reference in their entirety.

[0099] The binary bone mask 604 may then be split into a plurality of regions, each corresponding to a different individual bone, via a bone separation step 606. The different regions may be distinguishably labeled to generate a labeled (segmented) bone map 608 that distinguishes the regions of the graphical representation corresponding to different individual bones.

[0100] In certain embodiments, the bone separation step 606 includes applying one or more second-order derivative filters to the 3D image, for example, as described in U.S. Patent Application 14 / 812,483 filed on July 29, 2015 and PCT Application PCT / US15 / 42631 filed on July 29, 2015. Specifically, in this method, one or more second-order derivatives can be applied to the 3D image to generate a bone mask of the image with the bone boundary removed through splitting. Morphological processing operations can be performed to determine the split binary components of the bone mask through splitting, which are then used as seeds for region growing operations to generate a labeled (segmented) bone map ("segmentation map" in patent application 14 / 812,483 filed on July 29, 2015 and PCT Application PCT / US15 / 42631 filed on July 29, 2015), which includes a plurality of labeled regions that distinguish individual bones in the 3D image.

[0101] After generating the marked (segmented) skeletal map 608, the specific marked area corresponding to the specific individual vertebra of interest can be selected 610 and used to generate a single vertebra mask 506. The area corresponding to the specific individual vertebra can be selected automatically or semi-automatically based on the input from the user. For example, a graphical representation of the marked (segmented) skeletal map can be rendered to display to the user. The areas marked differently in the rendered graphical representation can be visually distinguished, for example, by different colors, grayscale shades, etc. Then, the user can simply identify the specific area corresponding to the individual vertebra of interest, for example, by 'clicking' (for example, with a mouse) or 'tapping' (for example, using a touch-sensitive interface). Then a mask identifying this area can be generated and used as a single vertebra mask 506. In this way, the user can select a specific individual vertebra for segmentation and / or analysis by a single quick 'click' or 'tapping' in a graphical user interface (GUI).

[0102] B. Detection and segmentation of vertebral regions

[0103] In certain embodiments, once a single vertebra mask 506 is obtained (e.g., accessed or generated by the systems and methods described herein), the vertebral segmentation methods described herein operate on the single vertebra mask 506 to generate a labeled inter-segmental vertebral map 522, such as Figure 4 The example shown in .

[0104] It is non-trivial to separate the vertebral body region from other regions of a single vertebra mask corresponding to other parts of the individual vertebrae. The method described herein includes specific processing steps that exploit the physical characteristics of the individual vertebrae and address the image processing challenges that certain characteristics present.

[0105] Generation of Bi-filled individual vertebrae masks

[0106] In certain embodiments, the methods described herein include performing one or more morphological operations to fill in perforations and / or interior regions of the individual vertebrae mask 508, thereby generating a filled individual vertebrae mask 510. This filling step 508 addresses the image processing challenges presented by the structural features of individual vertebrae, particularly their internal trabecular regions, as well as blood vessels and other delicate structures that pass through the cortical mantle of the vertebral body and form openings.

[0107] like Figure 3 A and Figure 3 The vertebral body corresponds to the cylindrical region of an individual vertebra that is connected to other regions of the individual vertebrae by relatively narrow structures, as shown in B. However, the interior of the vertebrae is not solid, dense bone, but includes bone marrow, soft tissue, and various other delicate structures, such as Figure 2B As shown. Due to the different densities of the outer and inner parts of the vertebrae, the outer and inner parts of the vertebrae appear as different grayscale intensities in the micro-CT images (e.g., inner soft tissue regions with lower intensity values, which represent less dense tissue). Subsequently, the individual vertebrae masks generated from such images are not solid, but rather shell-like and include hollow inner regions (e.g., inner voxels marked as digital '0' or Boolean false values).

[0108] The method described herein transforms the hollow, shell-like individual vertebrae mask 506 into a solid structure by filling these interior regions to generate a filled individual vertebrae mask 510. Performing subsequent distance transform 512 and watershed segmentation operations 516 can separate the vertebral body regions of individual vertebrae from other sub-regions based on the relative thickness of the vertebral body regions in the filled individual vertebrae mask compared to the portions of the mask connecting them to the other sub-regions.

[0109] Notably, the distance transform is used to identify thin structures (e.g., the 'neck') in the mask by determining the distance from each voxel of the mask to the nearest boundary (e.g., to the nearest voxel of a digital '0' or Boolean 'false' value). Thus, generating a padded single vertebra mask and performing a subsequent distance transform 512 using the padded single vertebra mask avoids severe over-segmentation errors that would otherwise result from applying the distance transform to the hollow, shell-like single vertebra mask that was initially accessed and / or generated. Such over-segmentation errors would, for example, replace Figure 4 The single, easily identifiable pyramidal region 404 of the example in FIG. 4 results in multiple smaller, possibly ambiguous sub-regions.

[0110] In some embodiments, the filling step 508 for generating the filled individual vertebrae mask 510 is performed using a morphological dilation operation 632 and a morphological hole filling operation 634, as shown in FIG. Figure 6 Detailed example process 600 is shown in FIG. A morphological expansion 632 is performed to grow the individual vertebrae mask 506 and fill the perforations therein, and to generate an expanded individual vertebrae mask. Such perforations generally correspond to small holes extending from the interior of the shell-shaped individual vertebrae mask 506 to the exterior. These perforations are caused by physical structures such as blood vessels extending from the interior (e.g., the bone marrow portion) of the vertebrae to the exterior. Therefore, the morphological expansion operation 632 may use expansion elements whose sizes are based on the sizes of physical structures such as blood vessels that generally cause perforations in individual vertebrae. For example, based on the resolution of the 3D image 502, the size of the expansion element in the voxel may be determined (e.g., automatically) to correspond to (e.g., approximately greater than or equal to) a specific physical size associated with the blood vessel. For example, for a 3D image with a resolution of approximately 20 to 30 μm along each dimension per voxel, an expansion element having a size of 5 to 8 voxels along one or more dimensions (e.g., corresponding to a physical size of approximately 100 to 240 μm along one or more dimensions) will be used.

[0111] In some embodiments, the size of the expansion element can be a user exposure parameter that the user can adjust himself. This is useful to consider certain situations where there are unusually large perforations, such as due to tumors and / or cracks in the vertebrae. The user can increase or enter a specific value for the size of the expansion element to take into account such characteristics.

[0112] In some embodiments, a morphological hole filling operation 634 is performed to fill one or more interior regions in the expanded single vertebra mask generated after the morphological dilation operation 632. Thus, the morphological hole filling operation 634 fills the interior regions of the single vertebra mask that physically correspond to the marrow and soft tissue interior regions of the particular individual vertebra represented by the single vertebra mask. In some embodiments, it is necessary to first eliminate perforations in the single vertebra mask via the morphological dilation operation 632 before performing the morphological hole filling operation 634. Specifically, certain morphological hole filling operations may fail when applied to a mask having perforations that prevent the interior region from being clearly delineated. Fig. 8A shows an exemplary dilated single vertebra mask after a morphological dilation operation, and Figure 8B An exemplary filled single vertebrae mask is shown after a morphological hole filling operation.

[0113] In certain embodiments, the filled individual vertebrae mask generated by applying the morphological dilation and hole filling operations is refined using a morphological erosion operation 636. Because the morphological dilation operation causes the individual vertebrae mask to grow outward in addition to filling the perforations, the morphological erosion operation 636 performed using an erosion element having a size that is approximately the same as the dilation element can be used to counteract this growing effect.

[0114] Thus, by filling the perforations and / or interior regions of the shell-like individual vertebrae mask 506 as described herein, a filled individual vertebrae mask 510 corresponding to the filled solid object may be generated. Fig. 7A and Figure 7B It is shown that after the morphological corrosion step 636, Figure 6 A representation of a populated single vertebrae mask generated by the example process 600 is shown. Figure 7B Shown through Fig. 7A A cross-sectional view of a filled single vertebra mask is shown. As shown, the interior of the single vertebra mask is solid and filled - the single 'hole' physically corresponds to the neural tube 204 of the physical vertebra represented by the single vertebra mask. The fine structure and interior areas of the physical individual vertebra represented by the filled single vertebra mask, such as Figure 2B 252 shown in does not exist and has been filled by the above filling method.

[0115] In certain embodiments, additional steps are performed, for example, prior to the filling step 508. For example, Figure 6 As shown in process 600, an optional automatic cropping step 620 and / or a fill image border step 622 may be performed. The automatic cropping step 620 crops the 3D image to a localized area surrounding a single vertebra mask. Reducing the image size in this manner may, for example, increase the speed of downstream processing steps. A fill image border step 622 may be included when a single vertebra mask identifies a particular individual vertebra that is partially invisible. In this case, a portion of the single vertebra mask is located on the border of the 3D image. Similar to the manner in which, in certain embodiments, it is necessary to fill (e.g., by a morphological dilation operation) perforations in a single vertebra mask prior to performing a morphological hole filling step, the open areas in a cross-section of a single vertebra mask that are located on the image border are filled via the fill image border step 622, thereby 'capping' the open ends of the single vertebra mask on the image border.

[0116] B.ii Distance Transformation and Distance Map Determination

[0117] In certain embodiments, process 500 includes the step of applying a distance transform 512 to the filled single vertebra mask 510 to determine a distance map 514. The distance transform determines [e.g., assigns a first value (e.g., a numerical 1; e.g., a Boolean 'false')] for each voxel in the filled single vertebra mask that corresponds to bone, the distance from the voxel to the nearest boundary or soft tissue region of the 3D image [e.g., the distance to the nearest voxel of the filled single vertebra mask having a second value (e.g., a numerical 0; e.g., a Boolean 'false')]. The distance transform thus produces a distance map 514 that includes a plurality of voxels, each voxel corresponding to a voxel of the filled single vertebra mask 510 and having (e.g., being assigned) a distance value representing the distance from the voxel to the nearest boundary and / or non-bone voxel (e.g., a voxel of the filled single vertebra mask having a value of 0).

[0118] By applying the distance transform to Fig. 7A Example distance map determined by filling a single vertebra mask in Fig. 9 . The values ​​of the distance map voxels are represented in grayscale and using contour lines. The areas inside the contour lines correspond to thicker areas than the areas outside the contour lines. Outside the contour lines, shading from dark to light gray indicates decreasing thickness (e.g., decreasing distance to the border), with white representing 0 distance. Thus, the thickest areas of the bone (primarily corresponding to the vertebral body areas) are displayed within the larger central contour, while the thinnest areas in the image (such as pedicles) are displayed outside the contour lines and fade to white as the distance approaches 0.

[0119] B.iii Watershed segmentation

[0120] In certain embodiments, once the distance map is determined, a watershed segmentation step 516 is applied to the distance map 514. The watershed segmentation step 516 includes a watershed segmentation operation, such as H-extrema watershed segmentation, which identifies a set of catchment basins 518 and / or watershed lines in the distance map. The catchment basins 518 of the distance map correspond to thicker bone regions, represented by larger distance values ​​within the distance map. The catchment basins 518 are separated from each other by watershed lines, which correspond to connecting lines of voxels corresponding to narrow connectors. Thus, the thick central region of the vertebra of the filled single vertebra mask 510 is represented by one catchment basin, while the regions corresponding to the pedicles and other structures of the particular individual vertebra attached to the vertebral body by narrow connections are represented by other catchment basins.

[0121] In some embodiments, a watershed segmentation operation divides the distance map into a plurality of catchment basins separated from one another by watershed lines. In some embodiments, a watershed segmentation operation produces a watershed mask comprising a plurality of catchment basins (e.g., each catchment basin corresponds to a connected region of voxels assigned a first value, such as a number 1 or a Boolean 'true'), the plurality of catchment basins being separated from one another by watershed lines (e.g., each watershed line corresponds to a connected line of voxels assigned a second value, such as a number 0 or a Boolean 'true').

[0122] B.iv Generation of masked and labeled intersegmental vertebral maps

[0123] In some embodiments, the masking step 520 uses the catchment basin group 518 generated via the watershed segmentation step 516 and the individual vertebrae masks 506 to generate an inter-segmental vertebrae map 522. The masking step 520 includes identifying portions of the individual vertebrae mask that are within different catchment basins in the group 518 and labeling them accordingly to distinguish them from each other. For example, a particular portion of the individual vertebrae mask 506 that is within a particular catchment basin can be identified as follows: a voxel-by-voxel logical AND operation is performed between the individual vertebrae mask 506 and the particular catchment basin, and the result is labeled [e.g., a label value (e.g., a particular integer value) is assigned to each voxel having a first value (e.g., a number 1; e.g., a Boolean 'true'). This process can be repeated for each catchment basin in the group 518, labeling each result differently to distinguish different areas of the individual vertebrae mask 506. In this way, a labeled inter-segmental vertebrae map is generated, such as Figure 4 Example 400 is shown. As described herein, the labeled inter-segmental vertebral map 522 corresponds to a labeled version of the individual vertebral mask 506, wherein portions of the individual vertebral mask 506 that are within different catch basins in the catch basin set are identified and labeled accordingly. With the combination of processing operations described herein, a vertebral body can be represented via a single, easily identifiable labeled region in the inter-segmental vertebral map.

[0124] C. Additional Processing

[0125] Ci is used for user interaction in the classification of vertebral sub-regions

[0126] In certain embodiments, once the labeled intersegmental vertebral map 522 is generated, a graphical representation of the labeled intersegmental vertebral map 522 is rendered 524 for presentation to a user, such as within a graphical user interface (GUI). The user can then select, via the GUI, a region corresponding to a vertebral body. Once the user has selected, for example, which labeled region of the labeled intersegmental vertebral map corresponds to a vertebral body, the region can be labeled as such (e.g., labeled as corresponding to a vertebral body). This method can be used, for example, to generate a binary labeled map that distinguishes regions in a 3D image that correspond to vertebral bodies of a particular individual vertebra of interest from other portions of the particular individual vertebra of interest. Additionally or alternatively, a vertebral mask can be generated that identifies the vertebral bodies of the particular individual vertebra of interest. Typically, as Figure 4 As shown in the exemplary intersegmental vertebral map 400 of FIG. 4 , the vertebral body is represented by a single, easily identifiable region ( Figure 4 404 in .

[0127] In this way, the systems and methods described herein allow users to identify the vertebral body region of a specific individual vertebra of interest represented in a 3D image by simply selecting a specific area of ​​the displayed intersegmental vertebral map as corresponding to a vertebral body. In certain embodiments, this can be achieved by a single positive 'click' (e.g., using a mouse) or 'tap' (e.g., using a touch-sensitive interface) in a graphical user interface (GUI). Therefore, the vertebral body detection and segmentation tool described herein eliminates the cumbersome and laborious process of users manually drawing boundaries to identify the region corresponding to the vertebral body of an individual vertebra in the image. In addition, since the marked area of ​​the marked intersegmental vertebral map is automatically generated, errors and inconsistencies between different users are greatly reduced (e.g., it is almost guaranteed that two or more users select the same area as corresponding to a vertebral body, but it is unlikely that two or more users will draw exactly the same boundaries on the image).

[0128] C.ii Metric Determination

[0129] Thus, by providing a tool for automatically detecting and segmenting the vertebral bodies of individual vertebrae in images of a subject, the systems and methods described herein facilitate the detection and segmentation methods described herein, thereby facilitating streamlined quantitative analysis of vertebral images for applications such as orthopedic research and disease / injury diagnosis. Specifically, the methods described herein provide a basis for analyzing the morphological properties, density, and structural parameters of the vertebral body regions of individual vertebrae. As described herein, such analysis can provide useful insights into the diagnosis, status, and progression of a subject's disease and / or injury, as well as analyzing the efficacy of different treatments.

[0130] For example, once a vertebral region of a marked intersegmental vertebral map is identified, it can be used (e.g., as a mask) to calculate one or more morphometric values, such as the volume or surface (e.g., surface area) of a vertebral body. Other measurements, such as connectivity, can also be determined. In certain embodiments, the identified vertebral region is used to determine the region of the image corresponding to the trabecular and / or cortical portion of the vertebral body. Therefore, morphometric values ​​such as volume, surface (e.g., surface area) can also be obtained for these specific components. For example, the identified vertebral region can be used to determine the trabecular subregion corresponding to the trabecular portion of the vertebral body. The volume of the trabecular subregion can be determined (e.g., automatically) to measure the trabecular volume of the vertebra. Automatically quantifying trabecular volume in this way can provide insight into the efficacy of different treatments for vertebral osteoporosis.

[0131] As described herein, since vertebral subregions are automatically generated and user interaction is limited at most to identifying (e.g., by selecting) only vertebral subregions, inter-user and intra-user errors and variations in measuring vertebral morphological traits are greatly reduced. Thus, the vertebral detection and segmentation methods described herein provide a valuable tool for assessing orthopedic disease status and / or progression in a subject and evaluating treatment efficacy.

[0132] C.iii Imaging modality

[0133] Although the images presented and analyzed by the methods described herein are micro-CT images, other imaging modalities may also be used. For example, the methods described herein may also be used to detect and segment the vertebral bodies of individual vertebrae in MRI images, optical images, and other types of images. Specifically, the vertebral body segmentation and detection tools described herein may be used to analyze any imaging modality that allows for imaging of vertebrae and bone tissue (e.g., any modality that provides sufficient contrast between bone tissue and soft tissue).

[0134] D. Computer system and network environment

[0135] like Fig.10 , an implementation of a network environment 1000 for providing the systems and methods described herein for automatic vertebral body detection and segmentation is shown and described. In a brief overview, reference is now made to Fig.10, a block diagram of an exemplary cloud computing environment 1000 is shown and described. The cloud computing environment 1000 may include one or more resource providers 1002a, 1002b, 1002c (collectively referred to as 1002). Each resource provider 1002 may include computing resources. In some implementations, computing resources may include any hardware and / or software for processing data. For example, computing resources may include hardware and / or software capable of executing algorithms, computer programs, and / or computer applications. In some implementations, exemplary computing resources may include application servers and / or databases with storage and retrieval capabilities. Each resource provider 1002 may be connected to any other resource provider 1002 in the cloud computing environment 1000. In some implementations, the resource providers 1002 may be connected via a computer network 1008. Each resource provider 1002 may be connected to one or more computing devices 1004a, 1004b, 1004c (collectively referred to as 1004) via a computer network 1008.

[0136] The cloud computing environment 1000 may include a resource manager 1006. The resource manager 1006 may be connected to resource providers 1002 and computing devices 1004 through a computer network 1008. In some implementations, the resource manager 1006 may facilitate one or more resource providers 1002 to provide computing resources to one or more computing devices 1004. The resource manager 1006 may receive a request for computing resources from a particular computing device 1004. The resource manager 1006 may identify one or more resource providers 1002 that can provide the computing resources requested by the computing device 1004. The resource manager 1006 may select a resource provider 1002 to provide the computing resources. The resource manager 1006 may facilitate a connection between a resource provider 1002 and a particular computing device 1004. In some implementations, the resource manager 1006 may establish a connection between a particular resource provider 1002 and a particular computing device 1004. In some implementations, the resource manager 1006 may redirect a particular computing device 1004 to a particular resource provider 1002 that has the requested computing resources.

[0137] Fig.11 Examples of computing device 1100 and mobile computing device 1150 that can be used to implement the techniques described in this disclosure are shown. Computing device 1100 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. Mobile computing device 1150 is intended to represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are exemplary only and not limiting.

[0138] The computing device 1100 includes a processor 1102, a memory 1104, a storage device 1106, a high-speed interface 1108 connected to the memory 1104 and a plurality of high-speed expansion ports 1110, and a low-speed interface 1112 connected to a low-speed expansion port 1114 and the storage device 1106. Each of the processor 1102, the memory 1104, the storage device 1106, the high-speed interface 1108, the high-speed expansion port 1110, and the low-speed interface 1112 is interconnected using various buses and can be installed on a common motherboard or in other suitable ways. The processor 1102 can process instructions for execution within the computing device 1100, including instructions stored in the memory 1104 or on the storage device 1106, to display graphical information of a graphical user interface (GUI) on an external input / output device (such as a display 1116 coupled to the high-speed interface 1108). In other implementations, multiple processors and / or multiple buses, as well as multiple memories and memory types, can be used appropriately. Furthermore, multiple computing devices may be connected, with each device providing a portion of the necessary operations (e.g., as a server bank, a group of blade servers, or a multi-processor system). Thus, as the terminology is used herein, where multiple functions are described as being performed by a "processor," this encompasses implementations in which the multiple functions are performed by any number of processor(s) of any number of computing devices(s). Furthermore, where a function is described as being performed by a "processor," this encompasses implementations in which the function is performed by any number of processor(s) of any number of computing devices(s) (e.g., in a distributed computing system).

[0139] The memory 1104 stores information within the computing device 1100. In some implementations, the memory 1104 is one or more volatile storage units. In some implementations, the memory 1104 is one or more non-volatile storage units. The memory 1104 may also be another form of computer-readable medium, such as a magnetic disk or optical disk.

[0140] The storage device 1106 can provide mass storage for the computing device 1100. In some implementations, the storage device 1106 can be or include a computer-readable medium, such as a floppy disk device, a hard disk device, an optical disk device, or a tape device, a flash memory or other similar solid-state storage device, or a device array (including devices in a storage area network or other configuration). Instructions can be stored in an information carrier. When the instructions are executed by one or more processing devices (e.g., processor 1102), one or more methods, such as the above methods, are performed. Instructions can also be stored by one or more storage devices (e.g., memory 1104, storage device 1106, or memory on processor 1102) such as computer-readable media or machine-readable media.

[0141] The high-speed interface 1108 manages bandwidth-intensive operations of the computing device 1100, while the low-speed interface 1112 manages less bandwidth-intensive operations. This allocation of functions is exemplary only. In some implementations, the high-speed interface 1108 can be coupled to the memory 1104, the display 1116 (e.g., through a graphics processor or accelerator), and can be coupled to a high-speed expansion port 1110 that can accept various expansion cards (not shown). In an implementation, the low-speed interface 1112 can be coupled to the storage device 1106 and the low-speed expansion port 1114. Various communication ports (e.g., USB, A low-speed expansion port 1114 (such as Ethernet, wireless Ethernet) can be coupled to one or more input / output devices, such as a keyboard, a pointing device, a scanner, or a networking device (such as a switch or a router), for example, through a network adapter.

[0142] The computing device 1100 can be implemented in a variety of different forms, as shown. For example, the computing device can be implemented as a standard server 1120, or multiple times in a group of such servers. In addition, the computing device can be implemented in a personal computer such as a laptop computer 1122. The computing device can also be implemented as part of a rack server system 1124. Alternatively, the components in the computing device 1100 can be combined with other components in a mobile device (not shown) such as a mobile computing device 1150. Each of such devices can contain one or more of the computing device 1100 and the mobile computing device 1150, and the entire system can be composed of multiple computing devices that communicate with each other.

[0143] The mobile computing device 1150 includes a processor 1152, a memory 1164, an input / output device such as a display 1154, a communication interface 1166, and a transceiver 1168, among other components. The mobile computing device 1150 may also be provided with a storage device such as a micro drive or other device to provide additional storage. Each of the processor 1152, the memory 1164, the display 1154, the communication interface 1166, and the transceiver 1168 may be interconnected using various buses, and several of these components may be mounted on a common motherboard, or may be mounted in other suitable ways.

[0144] The processor 1152 may execute instructions within the mobile computing device 1150, including instructions stored in the memory 1164. The processor 1152 may be implemented as a chipset including separate and multiple analog and digital processors. The processor 1152 may provide, for example, coordination of other components of the mobile computing device 1150, such as control of a user interface, applications executed by the mobile computing device 1150, and wireless communications performed by the mobile computing device 1150.

[0145] The processor 1152 can communicate with the user through the control interface 1158 and the display interface 1156 coupled to the display 1154. The display 1154 can be, for example, a thin film transistor liquid crystal display (TFT) or an organic light emitting diode (OLED) display or other suitable display technology. The display interface 1156 may include appropriate circuits for driving the display 1154 to present graphics and other information to the user. The control interface 1158 can receive commands from the user and convert them for submission to the processor 1152. In addition, the external interface 1162 can provide communication with the processor 1152 to enable near-area communication of the mobile computing device 1150 with other devices. The external interface 1162 can provide, for example, wired communication in some implementations, or wireless communication in other implementations, and multiple interfaces can also be used.

[0146] The memory 1164 stores information within the mobile computing device 1150. The memory 1164 may be implemented as one or more of a computer-readable medium, a volatile storage unit, or a non-volatile storage unit. An expansion memory 1174 may also be provided and connected to the mobile computing device 1150 via an expansion interface 1172, which may include, for example, a single in-line memory module (SIMM) card interface. The expansion memory 1174 may provide additional storage space for the mobile computing device 1150, or may also store applications or other information for the mobile computing device 1150. Specifically, the expansion memory 1174 may include instructions for executing or supplementing the above-mentioned processes, and may also include security information. Thus, for example, the expansion memory 1174 may be provided as a security module for the mobile computing device 1150, and may be programmed with instructions that allow the mobile computing device 1150 to be used securely. In addition, secure applications and other information may be provided via a SIMM card, such as placing identification information on a SIMM card in an unbreakable manner.

[0147] The memory may include, for example, flash memory and / or NVRAM memory (non-volatile random access memory), as discussed below. In some implementations, instructions are stored in an information carrier. The instructions, when executed by one or more processing devices (e.g., processor 1152), perform one or more methods, such as the above methods. The instructions may also be stored by one or more storage devices (e.g., memory 1164, expansion memory 1174, or memory on processor 1152), such as one or more computer-readable media or machine-readable media. In some implementations, instructions may be received, for example, by a transceiver 1168 or an external interface 1162 as a propagated signal.

[0148] The mobile computing device 1150 may communicate wirelessly via the communication interface 1166 and, if necessary, may include digital signal processing circuitry. The communication interface 1166 may provide for communication in various modes or protocols, such as global system for mobile communications (GSM) voice calls, short message service (SMS), enhanced message service (EMS) or multimedia message service messages (MMS), code division multiple access (CDMA), time division multiple access (TDMA), personal digital cellular (PDC), wideband code division multiple access (WCDMA), CDMA2000, or general packet radio service (GPRS). For example, such communication may be performed using radio frequencies via the transceiver 1168. In addition, wireless communication such as the wireless cellular network may be used. Wi-Fi TMOr other such transceiver (not shown) for short-range communication. In addition, the global positioning system (GPS) receiver module 1170 can provide additional navigation and location-related wireless data to the mobile computing device 1150, which can be used as appropriate by applications running on the mobile computing device 1150.

[0149] The mobile computing device 1150 may also communicate auditorily using an audio codec 1160 that may receive spoken information from a user and convert it into usable digital information. The audio codec 1160 may also generate audible sounds for the user, such as through a speaker, for example, in an earpiece of the mobile computing device 1150. Such sounds may include sounds from voice phone calls, may include recorded sounds (e.g., voice messages, music files, etc.), and may also include sounds generated by applications running on the mobile computing device 1150.

[0150] The mobile computing device 1150 may be implemented in a variety of different forms, as shown. For example, the mobile computing device may be implemented as a cellular phone 1180. The mobile computing device may also be implemented as part of a smart phone 1182, a personal digital assistant, or other similar mobile device.

[0151] Various implementations of the systems and techniques described herein can be implemented by digital electronic circuitry, integrated circuits, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementations in one or more computer programs that are executable and / or interpretable on a programmable system that includes at least one programmable processor, which can be special purpose or general purpose, coupled to receive data and instructions from and send data and instructions to a storage system, at least one input device, and at least one output device.

[0152] These computer programs (also referred to as programs, software, software applications or code) include machine instructions for a programmable processor and may be implemented using high-level procedural and / or object-oriented programming languages ​​and / or using assembly / machine languages. The terms machine-readable medium and computer-readable medium as used herein refer to any computer program product, apparatus and / or device (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term machine-readable signal refers to any signal for providing machine instructions and / or data to a programmable processor.

[0153] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having a display device (e.g., a cathode ray tube (CRT) or liquid crystal display (LCD) monitor) for displaying information to the user and a keyboard and pointing device (e.g., a mouse or trackball) by which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with a user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input from the user can be received in any form, including auditory, voice, or tactile input.

[0154] The systems and techniques described herein can be implemented in a computing system that includes a back-end component (e.g., as a data server), or includes a middleware component (e.g., an application server), or includes a front-end component (e.g., a client computer with a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any digital data communication form or medium (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0155] A computing system may include clients and servers. Clients and servers are generally remote from each other and typically interact through a communication network. The relationship of client and server arises from computer programs running on the respective computers and having a client-server relationship to each other.

[0156] In some implementations, any modules described herein may be separated, combined, or incorporated into a single or combined module.The modules depicted in the figures are not intended to limit the systems described herein to the software architecture shown therein.

[0157] The elements of different implementations described herein can be combined to form other implementations not specifically described above. Processes, computer programs, databases, etc. described herein can exclude several elements without adversely affecting their operation. In addition, the logic flow depicted in the accompanying drawings does not require the particular order shown or the order in sequence to achieve the desired result. Various separated elements can be combined into one or more single elements to perform the functions described herein.

[0158] Throughout the specification, when devices and systems are described as having, containing, or including specific components, or when processes and methods are described as having, containing, or including specific steps, it is expected that, in addition, there are devices and systems of the present invention that are primarily composed of or consist of the described components, and there are processes and methods of the present invention that are primarily composed of or consist of the described processing steps.

[0159] It should be understood that the order of steps or the order in which certain actions are performed is not important as long as the present invention remains operable. In addition, two or more steps or actions may be performed simultaneously.

[0160] While the present invention has been particularly shown and described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined in the following claims.

Claims

1. A method for automatically detecting and segmenting a vertebral body of a specific vertebra in a 3D image of a subject, the method comprising: (a) receiving, by a processor of a computing device, a 3D image of a subject, wherein the 3D image includes a graphical representation of one or more vertebral portions of the subject; (b) applying a thresholding operation to the 3D image to generate a binary bone mask identifying areas of the 3D image corresponding to bones; (c) accessing and / or generating, by the processor, a single vertebrae mask that identifies a portion of the graphical representation determined to correspond to the particular vertebra; (d) applying, by the processor, a cropping operation to reduce the 3D image to a localized region surrounding the single vertebrae mask; (e) applying, by the processor, a morphological dilation operation to grow the single vertebrae mask, thereby generating a dilated single vertebrae mask; (f) applying a morphological hole filling operation to the dilated single vertebra mask to fill one or more interior regions within the dilated single vertebra mask, thereby generating a filled single vertebra mask; (g) determining, by the processor, a distance map by applying a distance transform to the filled single vertebrae mask, the distance transform being used to identify thin structures in the mask by determining the distance from each voxel of the filled single vertebrae mask to the nearest boundary; (h) applying, by the processor, a watershed segmentation operation to the distance map to identify a set of catchment basins from the distance map; (i) determining, by the processor, a labeled intersegmental vertebral map using the set of catch basins and the single vertebral mask, the labeled intersegmental vertebral map comprising a plurality of labeled regions, one of the labeled regions corresponding to the vertebral body; as well as (j) rendering, by the processor, a graphical representation of the labeled inter-segmental vertebral map.

2. The method of claim 1, wherein step (c) comprises segmenting, by the processor, the 3D image to generate the individual vertebrae masks.

3. The method of claim 1 or 2, wherein step (c) comprises: segmenting, by the processor, the 3D image to generate a labeled skeletal map, the labeled skeletal map comprising a plurality of labeled regions that distinguish portions of the graphical representation corresponding to respective bones; rendering, by the processor, a graphical representation of the labeled skeletal map; receiving, by the processor, a user selection of at least one of the plurality of marking areas; as well as The single vertebrae mask is generated by the processor from the user-selected marked region. 4 . The method of claim 2 , wherein said segmenting said 3D image comprises applying one or more second order derivative splitting filters to said 3D image.

5. A method as claimed in any one of the preceding claims, wherein at least a portion of the single vertebrae mask is located on an edge of the 3D image, and wherein the method comprises filling the interior of the portion of the single vertebrae mask located on the edge of the 3D image.

6. The method of claim 1, comprising refining the filled individual vertebrae mask by performing a morphological erosion operation by the processor.

7. The method of claim 6, wherein the morphological dilation operation uses a dilation element having a size that is preset and / or automatically determined based on a resolution of the 3D image.

8. The method of claim 6, wherein the method comprises: A user input of a dilation element size value is received by the processor and the user input dilation element size is used in applying the morphological dilation operation.

9. A method as claimed in any one of the preceding claims, comprising: (k) after step (j), receiving, by the processor via a graphical user interface (GUI), a user selection of the marked region of the intersegmental vertebral map corresponding to the vertebral body; as well as (l) Determining, by the processor, a vertebral region of the intersegmental vertebral map, the vertebral region corresponding to the user selection.

10. The method of claim 9, comprising determining, by the processor, one or more morphometric values ​​using the determined vertebral body regions.

11. The method of claim 10, wherein the one or more morphometric measurements include measurements of one or more morphometric traits of the trabecular and / or cortical portions of the vertebral body.

12. The method of any one of the preceding claims, wherein the 3D image of the subject is a CT image, and wherein the method comprises acquiring the CT image.

13. A system for automatically detecting and segmenting vertebral bodies of specific vertebrae in a 3D image of a subject, the system comprising: a processor of a computing device; and a memory having stored thereon instructions, wherein the instructions, when executed by the processor, cause the processor to: (a) receiving a 3D image of a subject, wherein the 3D image includes a graphical representation of one or more vertebral portions of the subject; (b) applying a thresholding operation to the 3D image to generate a binary bone mask identifying an area of ​​the 3D image corresponding to bone; (c) accessing and / or generating a single vertebra mask identifying a portion of the graphical representation determined to correspond to the particular vertebra; (d) applying, by the processor, a cropping operation to reduce the 3D image to a localized region surrounding the single vertebrae mask; (e) applying a morphological dilation operation to grow the single vertebra mask, thereby generating a dilated single vertebra mask; (f) applying a morphological hole filling operation to the dilated single vertebra mask to fill one or more interior regions within the dilated single vertebra mask, thereby generating a filled single vertebra mask; (g) determining a distance map by applying a distance transform to the filled single vertebrae mask, the distance transform being used to identify thin structures in the mask by determining the distance from each voxel of the filled single vertebrae mask to the nearest boundary; (h) applying a watershed segmentation operation to the distance map to identify a set of catchment basins from the distance map; (i) determining a labeled intersegmental vertebral map using the set of catch basins and the single vertebral mask, the map comprising a plurality of labeled regions, one of the labeled regions corresponding to the vertebral body; and (j) Rendering of a graphical representation of the labeled intersegmental vertebral map.

14. The system of claim 13, wherein: In step (c), the instructions cause the processor to segment the 3D image to generate the individual vertebrae masks.

15. The system of claim 13 or 14, wherein: In step (c), the instructions cause the processor to: segmenting the 3D image to generate a labeled skeletal map, the labeled skeletal map comprising a plurality of labeled regions that distinguish portions of the graphical representation corresponding to respective bones; rendering a graphical representation of the labeled skeletal map; receiving a user selection of at least one of the plurality of marking areas; and The single vertebra mask is generated from the user-selected marked region.

16. The system of claim 14 or 15, wherein the instructions cause the processor to segment the 3D image by applying one or more second order derivative splitting filters to the 3D image.

17. The system of any one of claims 13 to 16, wherein at least a portion of the single vertebrae mask is located on an edge of the 3D image, and wherein the instructions cause the processor to fill in an interior of the portion of the single vertebrae mask that is located on an edge of the 3D image.

18. The system of claim 13, wherein the instructions cause the processor to refine the filled single vertebrae mask by performing a morphological erosion operation.

19. The system of claim 13 or 18, wherein the instructions cause the processor to perform the morphological dilation operation using a dilation element having a size that is preset and / or automatically determined based on a resolution of the 3D image.

20. The system of claim 13 or 18, wherein the instructions cause the processor to receive user input of a dilation element size value and use the user input dilation element size in said applying the morphological dilation operation.

21. The system of any one of claims 13 to 20, wherein the instructions cause the processor to: (k) after step (j), receiving, via a graphical user interface (GUI), a user selection of the marked region of the intersegmental vertebral map corresponding to the vertebral body; and (l) Determining a vertebral region of the intersegmental vertebral map, the vertebral region corresponding to the user selection.

22. The system of claim 21, wherein the instructions cause the processor to determine one or more morphometric values ​​using the determined vertebral body regions.

23. The system of claim 22, wherein the one or more morphometric measurements include measurements of one or more morphometric traits of the trabecular and / or cortical portions of the vertebral body.

24. The system of any one of claims 13 to 23, wherein the 3D image of the subject is a CT image.

Images