Image processing of combined ribs and spine for rapid scan evaluation

Abstract

A visualization scheme is generated based on automated segmentation to display the patient's anatomy. Data representing a three-dimensional diagnostic image is received. The ribs and vertebrae are segmented, and rib centerlines and vertebral center landmarks are detected and labeled. Three-dimensional position coordinates corresponding to the received data are mapped to a defined two-dimensional manifold plane, and then an interpolation technique is performed to complete the remaining three-dimensional position coordinates and deform the two-dimensional manifold plane. Image intensities at each coordinate of the deformed two-dimensional manifold plane are sampled from the three-dimensional diagnostic image space. From the sampled image intensities, a reformatted image is generated as manifold slices that display a continuous and linearized visual representation of the thorax. A three-dimensional visual representation of the thorax is generated from the multiple manifold slices.

Description

Technical Field

[0001] The present invention relates to the field of medical images, and more particularly, to a method and apparatus for automatically generating visual representations of ribs and vertebrae.

Background Art

[0002] Medical professionals generally use medical images to visualize a patient's anatomical structure, in which case the visual representation data is used to diagnose diseases or injuries. In trauma situations, medical staff can rely on medical image data in the form of computed tomography (CT) scans for diagnostic purposes, such as detecting rib fractures.

Summary of the Invention

Problems to be Solved by the Invention

[0003] Looking at a three-dimensional (3D) CT scan of the chest or the whole body slice by slice is often a time-consuming process, especially when the target anatomical structure spans multiple slices (e.g., tracing 24 individual ribs).

[0004] Reading image scans, more specifically scans in the trauma or emergency department, is a time-critical task that needs to be done with great care to avoid missing important findings. Often, in these settings, the imaging is based on whole-body scans, resulting in a large amount of image data that needs to be thoroughly examined. Ribs and vertebrae are particularly important structures to be evaluated. These are repetitive structures that require a significant amount of time for examination, and in the case of vertebrae, these structures are important from a neurological perspective.

[0005] Visualizing a patient's anatomical structure and certain important structures in an intuitive way can simplify the time-consuming examinations currently required. Several approaches have been proposed to simplify the visualization and evaluation of a patient's anatomical structure and important structures, with a particular focus on approaches aimed at visualizing the thoracic cage. While offering several advantages, current approaches are accompanied by clear fundamental limitations.

[0006] One well-known visualization scheme is the "filet view" or "fishbone view." This view involves segmenting the ribs (for example, using a deep convolutional neural network) and then using a centerline extractor to label pairs of ribs within the field of view. Each rib is sampled along its trace, thereby allowing visualization of each rib and all ribs in a normalized and linearized form (curved planar reformat). This view allows medical professionals to accurately examine the rib centerlines in a normalized view (reformatted view) where all ribs are linearized and positioned uniquely on the examination canvas.

[0007] However, this type of view has several drawbacks. One drawback of this type of view is that its nature of treating each rib independently can lead to discontinuities between ribs, which can result in image artifacts from adjacent ribs appearing in the rib shape. In addition, this view is limited to ribs only, and adding contextual information such as the spine requires additional steps and image processing that must be done separately.

[0008] Another visualization scheme is the visceral cavity view. In this view, a segmentation algorithm (e.g., using a model-based approach) is applied to segment the inside of the thoracic cavity with respect to the deformation of a cylindrical manifold. Once segmentation is performed, the manifold can be unwrapped, and the maximum intensity projection near the surface (MIP) can be calculated. This view allows the user to inspect the entire thoracic cavity with respect to a continuous visual representation on the inspection canvas.

[0009] One drawback of this type of view is that relative rib lengths are not maintained. Due to the nature of unwrapping the cylindrical manifold, it is not possible to visualize the correct rib lengths. For example, the first rib may appear longer than the other ribs. In addition, due to the nature of the MIP visual representation, microfractures may not be detected. For example, in the MIP visual representation, small rib fractures may become invisible and undetectable in the generated view. Another drawback is that this view adds unrealistically large distortions (wavy ribs) that limit the clinical reliability of the generated visceral cavity view.

[0010] Therefore, there is a need for an innovative visualization scheme that combines the advantages of fillet views and visceral cavities without the drawbacks associated with these views.

[0011] The objective of the present invention is to provide a technique for generating a novel manifold view that overcomes the shortcomings of existing visualization schemes. [Means for solving the problem]

[0012] This technology can be applied to several imaging systems, including CT, CT arms, single-photon emission computed tomography (SPECT-CT), magnetic resonance imaging (MR-CT), positron emission tomography (PET-CT), and magnetic resonance imaging (MRI) systems.

[0013] According to a first aspect of the present invention, an image processing apparatus is provided. The image processing apparatus includes a memory configured to store computer executable instructions, and at least one processor that executes computer executable instructions to cause the image processing apparatus to perform the step of receiving data representing a three-dimensional diagnostic image including the ribs and spine of a target. The computer executable instructions cause the image processing apparatus to perform the steps of segmenting the ribs and spine according to the received data representing the three-dimensional diagnostic image, detecting and labeling the rib centerlines from the rib segmentation, detecting and labeling vertebral center landmarks from the spine segmentation, and defining a two-dimensional manifold plane representing a visualization canvas for displaying the reformatted image. A computer-executable instruction causes an image processing device to perform the following steps: mapping the three-dimensional positions of the rib centerline and vertebral center landmarks corresponding to received data representing a three-dimensional diagnostic image to two-dimensional positions on a defined two-dimensional plane; interpolating missing three-dimensional position coordinates on a defined two-dimensional manifold plane, which includes deforming the two-dimensional manifold plane so that it aligns with the detected rib centerline and vertebral center landmarks; and sampling image intensity at each coordinate of the deformed two-dimensional manifold plane from the three-dimensional diagnostic image space. From the sampled image intensity, a reformatted image is generated as a manifold slice, which displays a continuous and linearized visual representation of the rib cage and spine. The deformed two-dimensional manifold plane is shifted along its normal direction, and sampling is repeated to generate a stack of manifold slices that cover a complete three-dimensional visual representation of the rib cage. The three-dimensional visual representation of the rib cage and vertebral centers is generated as a stack of manifolds that cover the entire rib cage.

[0014] A second aspect of the present invention provides a method for processing a three-dimensional image. This method includes the steps of: receiving data representing a three-dimensional diagnostic image including the ribs and spine of a subject; segmenting the ribs and spine according to the received data representing the three-dimensional diagnostic image; detecting and labeling the rib centerline from the rib segmentation; detecting and labeling the vertebral body center landmark from the spine segmentation; and defining a two-dimensional manifold plane representing a visualization canvas for displaying the reformatted image. This method comprises the steps of: mapping the three-dimensional positions of the rib centerline and vertebral center landmarks corresponding to received data representing a three-dimensional diagnostic image to two-dimensional positions on a defined two-dimensional plane; interpolating missing three-dimensional position coordinates on a defined two-dimensional manifold plane, which includes deforming the two-dimensional manifold so that the two-dimensional manifold plane aligns with the detected rib centerline and vertebral center landmarks; and sampling image intensity at each coordinate of the deformed two-dimensional manifold plane from the three-dimensional diagnostic image space. The reformatted image is generated as a manifold slice from the sampled image intensity, and the manifold slice displays a continuous and linearized visual representation of the rib cage and spine. The deformed two-dimensional manifold plane is shifted along its normal direction, and sampling is repeated to generate a stack of manifold slices that cover a complete three-dimensional visual representation of the rib cage. The three-dimensional visual representation of the rib cage and vertebral centers is generated as a stack of manifolds that cover the entire rib cage.

[0015] A third aspect of the present invention provides a non-temporary computer-readable medium storing instructions for a processing circuit to execute a process. The process includes: receiving a three-dimensional diagnostic image including the ribs and spine of a subject; segmenting the ribs and spine according to the received data representing the three-dimensional diagnostic image; detecting and labeling the rib centerlines from the rib segmentation; detecting and labeling vertebral center landmarks from the spine segmentation; and defining a two-dimensional manifold plane representing a visualization canvas for displaying a reformatted image, wherein the two-dimensional manifold plane is deformed to align with the detected rib centerlines and vertebral center landmarks. The process further includes the steps of: mapping the three-dimensional positions of the rib centerline and vertebral center landmarks corresponding to the received data representing a three-dimensional diagnostic image to two-dimensional positions on a defined two-dimensional plane; and interpolating missing three-dimensional position coordinates on a defined two-dimensional manifold plane, which includes deforming the two-dimensional manifold plane so that it aligns with the detected rib centerline and vertebral center landmarks. Image intensity at each coordinate of the deformed two-dimensional manifold plane is sampled from the three-dimensional diagnostic image space. From the sampled image intensity, a reformatted image is generated as a manifold slice, which displays a continuous and linearized visual representation of the rib cage and spine. The deformed two-dimensional manifold plane is shifted along its normal direction, and sampling is repeated to generate a stack of manifold slices that cover a complete three-dimensional visual representation of the rib cage. The three-dimensional visual representation of the rib cage and vertebral centers is generated as a stack of manifolds that cover the entire rib cage.

[0016] In a preferred embodiment, rib and vertebral segmentation is performed using machine learning or deep learning techniques. These machine learning or deep learning techniques include at least one of neural networks, logistic regression, random forests, nearest neighbor analysis, and cluster or multivariate analysis. The deep learning-based segmentation approach is applied to segment the ribs and vertebrae, extract rib centerlines, and detect vertebral body center landmarks. As a result, all rib pairs and vertebral body centers are accurately detected and labeled in an automated manner.

[0017] A two-dimensional manifold plane is defined, used to display a visual representation canvas for displaying the reformatted image. The three-dimensional positions of the detected rib centerlines and vertebral center landmarks are mapped to two-dimensional positions on the visualization canvas that visualize the unfolded ribs. In other words, a coordinate correspondence from two-dimensional space to three-dimensional space is defined.

[0018] Once the correspondence from 2D coordinates to 3D coordinates is established, the correspondence from the remaining pixels in 2D space still lags behind. In one aspect of the present invention, an interpolation scheme is realized to interpolate the x / y / z coordinate correspondence from 3D space to 2D space in such areas where this information is not yet available.

[0019] In a preferred embodiment, interpolation is performed using a thin-plate spline technique. Thin-plate splines are a good choice for interpolation techniques because they originally model the deformation of a thin two-dimensional metal plate. Image intensity at each coordinate of the deformed two-dimensional manifold plane is sampled from the three-dimensional diagnostic image space. Thus, a reformatted image is generated from the sampled image intensity as a manifold slice, which displays a continuous and linearized visual representation of the thoracic cage and spine.

[0020] In one aspect of the present invention, the sampled image intensity at each position of the deformed two-dimensional manifold plane corresponds to at least one of the regions of the intercostal tissue and the tissue adjacent to the rib.

[0021] In one aspect of the present invention, a stack of interpolated two-dimensional multiplanar reconstructions (MPRs) is generated by calculating the normal direction of each point on the two-dimensional plane and shifting or moving the manifold surface a plurality of times in the inner and outer directions of the manifold plane. Then, three-dimensional images of the rib cage and spine are generated based on the stack of two-dimensional MPRs.

Brief Description of the Drawings

[0022] [Figure 1] Diagram of an exemplary imaging system. [Figure 2] Diagram showing a view of the rib cage in a "fillet view" or "fishbone view". [Figure 3] Diagram showing the visceral cavity of the rib cage. [Figure 4] Flowchart showing a method for generating a composite rib and spine view according to some embodiments. [Figure 5] Diagram showing a mapping from three-dimensional space to two-dimensional space according to some embodiments. [Figure 6] Diagram showing a processing workflow according to some embodiments. [Figure 7] Diagram showing three-dimensional views of the ribs and spine according to some embodiments.

Modes for Carrying Out the Invention

[0023] FIG. 1 shows an imaging system 100 such as a computed tomography (CT) imaging system. The imaging system 100 generally includes a stationary gantry 102 and a rotating gantry 104. The rotating gantry 104 is rotatably supported by the stationary gantry 102 and rotates around the inspection area 106 about a longitudinal axis or z-axis relative to the stationary gantry 102.

[0024] A patient support 112, such as a treatment table, supports an object or subject, such as a human patient, in the inspection area 106. The patient support 112 is configured to move the object or subject for loading, scanning, and unloading.

[0025] A radiation source 108, such as an X-ray tube, is rotatably supported by the rotating gantry 104. The radiation source 108 rotates with the rotating gantry 104 and emits radiation that traverses the inspection area 106.

[0026] The radiation-sensitive detector array 110 forms an angular arc across the inspection area 106 on the opposite side of the radiation source 108. The detector array 110 has one or more rows of detectors that extend along the z-axis direction, detect radiation traversing the inspection area 106, and generate projection data indicative of the detected radiation.

[0027] A general-purpose computing system or computer functions as an operator console 114 and has an input device 116, such as a mouse, keyboard, and / or the like, and an output device 120, such as a display monitor, film, etc. The console 114 enables an operator to control the operation of the system 100. This includes control of the image processing device 118. The image processing device 118 can receive data representing a three-dimensional (3D) diagnostic image generated by the imaging system 100 and generate a reformatted image that displays a continuous and straightened visual representation of the chest wall and spine. The reformatted image can be output to the output device 120 for reading by medical personnel.

[0028] It is understood that the processing of the image processing apparatus 118 can be achieved through a processor and / or processing circuit that executes computer-readable instructions, such as executable code encoded or embedded on a computer-readable storage medium, such as physical memory and other non-temporary media. Additionally or alternatively, one or more processors and / or processing circuits can execute computer-readable instructions propagated by carrier waves, signals, and other temporary (or non-temporary) media.

[0029] Figure 2 shows a diagram of the rib cage using the "fillet view" or "fishbone view" visualization scheme. This view allows medical professionals to accurately examine the costal midline in a normalized view (reformatted view), where all ribs are straightened and positioned in unique locations on the examination canvas.

[0030] One drawback of this view is that, due to its nature of treating each rib independently, discontinuities appear between ribs, which can lead to image artifacts that appear in the rib shape from adjacent ribs, as identified in 202. Another drawback is that the field of view is limited to ribs only.

[0031] Figure 3 shows a diagram of the visceral cavity. In this visual scheme, for example, a segmentation algorithm using a model-based approach is applied to segment the inside of the thoracic cavity with respect to the deformation of a cylindrical manifold. Once segmentation is performed, the manifold can be unwrapped, and the maximum intensity projection (MIP) near the surface proximity can be calculated.

[0032] This view allows the user to examine the entire thoracic cavity in relation to a continuous visual representation on the examination canvas. One drawback of the visceral cavity view shown in Figure 3 is that relative rib lengths are not maintained. For example, the first rib appears much longer than the other ribs. In addition, due to the nature of MIP visualization, microfractures may not be detectable. For example, MIP visualization may make small rib fractures invisible and undetectable in the generated view. Another drawback is that this view adds significant unrealistic distortion (wavy ribs) that limits the clinical reliability of the generated visceral cavity view.

[0033] Figure 4 is a flowchart illustrating a method for generating reformatted medical images. The type of medical imaging performed may be, for example, a CT scan that generates a 3D CT image of a human patient.

[0034] In block 402, a 3D diagnostic image is received. For example, the 3D diagnostic image may be a CT imaging system scan. In block 404, the scanned patient's ribs and spine are automatically segmented. Segmentation can be performed using machine learning or deep learning techniques. For example, a convolutional neural network (CNN) is a deep learning technique that can be used to segment the ribs and spine. CNN architectures such as U-Net and V-Net can also be adapted for the segmentation task. Algorithms can be developed that demonstrate fast and accurate segmentation of the ribs and spine using deep learning-based training with datasets from hundreds of scans.

[0035] As part of the process, in block 406, the ribs are detected, the costal midlines are extracted and labeled, and in block 408, the vertebral body central landmarks are detected and labeled. In block 410, a two-dimensional (2D) manifold plane is defined. The 2D manifold plane is defined to represent the visualization canvas of the subsequently generated reformatted image. The 2D manifold plane defines the visualization canvas of the reformatted image as a standardized visualization of the thoracic cage. Thus, the standardized visual representation of the thoracic cage can be used for comparison with follow-up scans without requiring image registration, since the standardized view is already registered. In other words, a one-to-one mapping between coordinates in one space and coordinates in another space is determined so that points in the two spaces correspond to the same anatomical point.

[0036] In block 412, the three-dimensional (3D) positions of the rib centerlines and vertebral center landmarks corresponding to the received scan image data representing a three-dimensional diagnostic image are mapped to 2D positions on a defined two-dimensional manifold plane. Since the positions of the ribs and vertebrae are known in 3D space, correlations can be constructed for all rib centerlines and all vertebral center landmarks. That is, the coordinate correspondence from 3D space to 2D space can be defined as a mapping function. Once the 3D-to-2D correspondence is established or set up, there may be missing correspondences or correlations from the remaining pixels in the defined 2D space.

[0037] Therefore, in areas where this information is missing, in block 414, the interpolation scheme is implemented to interpolate the x / y / z coordinate correspondence from 3D space to a defined 2D space. This information may include interribular tissue regions and tissue regions adjacent to the ribs. This information may also include parts of the lung that can be extremely important when pneumothorax and hemothorax coexist. The interpolation scheme deforms the defined 2D manifold so that it aligns with the detected rib centerline and vertebral center landmarks. Thin plate splines are a good choice for use as an interpolation method because they originally model the deformation of thin 2D metal plates. Elastic body splines (EBS) are another mapping or coordinate transformation method suitable for medical image analysis.

[0038] In block 416, image intensity from the 3D diagnostic image space is sampled at each coordinate in the deformed 2D manifold plane. Sampling image intensity from the 3D diagnostic image space at each coordinate in the deformed 2D manifold plane ultimately unwraps the thoracic cage and results in a reformatted image. The sampling of image intensity can be based on a coordinate correspondence from 3D space to 2D space as the mapping function. That is, sampling is performed according to a coordinate mapping function corresponding to the mapped and interpolated positions.

[0039] In block 418, the reformatted image is generated as a manifold slice from the sampled image intensity. The reformatted image generated as a manifold slice displays a continuous and linearized visual representation of the thoracic cage and spine. The reformatted image combines the advantages of both fillet view and visceral cavity view, displaying a normalized and linearized rib view while maintaining relative rib lengths, and avoids the disadvantages of both fillet view and visceral cavity view, including discontinuities between ribs, imaging artifacts appearing from adjacent rib positions, inaccurate rib lengths, and invisible fractures and distortions due to MIP. The reformatted image generated in block 418 can be output to medical professionals for diagnostic purposes.

[0040] To facilitate further examination and diagnosis, a stack of 2D manifold slices can be generated. In block 420, the 2D manifold plane is shifted along its normal direction inward and outward from the manifold plane, and in block 422, sampling is repeated for each increment shift along the normal direction to generate a stack of manifold slices. The stack of manifold slices covers a complete 3D visual representation of the rib cage and spine.

[0041] In block 424, a 3D visual representation covering the entire rib cage and vertebral centers is generated as a stack of manifolds covering the entire rib cage. The stack of manifold slices generated in block 422 may be a stack of interpolated 2D multi-planar reconstructions (MPRs), and in block 424, a 3D image of the rib cage and spine can be generated based on the stack of 2D MPRs. The generated 3D visual representation covering the entire rib cage and vertebral centers allows for the inclusion of all ribs as a whole. This allows medical professionals to easily browse the rib cage for a quick visual inspection and to look for rib cage and vertebral fractures.

[0042] Figure 5 shows the mapping from 3D space to 2D space. The positions of the ribs and vertebrae are known in 3D space. Each position of the rib centerline and vertebral center landmark in 3D space is mapped to a corresponding position on a defined 2D manifold plane. The defined 2D manifold plane is the visualization canvas for the 2D manifold image that is later generated. Essentially, the ribs are traced, and each rib centerline and vertebral center landmark is mapped to a pixel position that appears on the 2D manifold. For ease of explanation, only three rib and vertebral center landmarks are shown in Figure 5. However, it should be understood that all rib and vertebral center landmarks corresponding to the rib and vertebral center landmarks known in 3D space are mapped to positions within the defined 2D manifold plane.

[0043] Figure 6 shows the processing workflow according to several embodiments. In 602, the rib centerline and vertebral center landmarks are detected and labeled according to an automated segmentation scheme. The image segmentation scheme can be based on machine learning or deep learning techniques. A two-dimensional manifold is defined that shows the resulting reformatted view. In 604, an interpolation scheme deforms the two-dimensional manifold to align with the detected rib centerline and vertebral center landmarks. A thin spline or any suitable interpolation scheme can be used. As shown in 604, the defined 2D manifold is deformed to align with the detected rib centerline and vertebral center landmarks, which are mapped from 3D space to the 2D space of the deformed 2D manifold.

[0044] For example, image intensity from a 3D image space, such as a 3D image from a CT scan, is sampled at each coordinate of the deformed 2D manifold to complete or fill in missing 3D coordinates, as shown in 606. The missing 3D coordinates may correspond to regions of intercostal tissue and regions of tissue adjacent to the ribs, or any combination of intercostal tissue and regions of tissue adjacent to the ribs. This allows the deformed 2D manifold to be unfolded.

[0045] An unfolded 2D manifold is illustrated in 608. The new manifold view, shown in 608 as a reformatted image, combines the advantages of both fillet and visceral views while avoiding the drawbacks of both. In 608, it can be seen that a normalized and linearized view of the ribs is displayed while maintaining relative rib lengths. Discontinuities between ribs, image artifacts arising from the position of adjacent ribs, inaccurate rib lengths, and invisible fractures and distortions caused by MIP are avoided.

[0046] Figure 7 shows 3D views of the ribs and spine according to several embodiments. 3D visual representations of the thoracic cage and vertebral centers can be generated by repeatedly sampling by shifting or moving a deformed 2D manifold plane multiple times inward and outward along its normal direction, thereby generating a stack of 2D manifold slices. This results in a 3D image as a stack of manifolds covering the entire thoracic cage. Figure 7 shows that the resulting 3D image allows for the inclusion of all ribs as a whole.

[0047] A stack or manifold slice can be a stack of interpolated 2D multi-section reconstructions (MPR), and 3D images of the thoracic cage and spine can be generated based on a stack of 2D MPRs. The generated 3D rendering is shown in 702. The view in 704 is rotated by approximately 30 degrees, and the view in 706 is rotated by an angle slightly less than 90 degrees.

[0048] Although the present invention has been illustrated and described in detail in the drawings and the foregoing description, such illustrations and descriptions should be considered descriptive or illustrative and not limiting. The present invention is not limited to the disclosed embodiments.

[0049] Other variations of the disclosed embodiments can be understood and implemented by those skilled in the art in carrying out the claimed invention, based on a review of the drawings, disclosures, and appended claims.

[0050] In the claims, the words "comprising" do not exclude other components or steps, and the indefinite articles "a" or "an" do not exclude plurality.

[0051] A single processor, device, or other unit may fulfill the functions of several items enumerated in the claims. The mere fact that certain means are described in different dependent claims does not imply that combinations of these means cannot be used advantageously.

[0052] Operations such as acquisition, decision, retrieve, output, provide, store, calculate, simulate, receive, warn, and stop can be implemented as program code means of a computer program and / or as dedicated hardware. Computer programs may be stored and / or distributed on suitable media such as optical storage media or solid-state media supplied together with or as part of other hardware, but may also be distributed in other forms such as the internet or other wired or wireless telecommunications systems.

Claims

1. An image processing device, Memory that stores executable computer instructions, A processor that executes computer executable instructions, The computer executable instructions are provided to the image processing device, The steps include receiving data representing a three-dimensional diagnostic image including the target ribs and spine, The steps include segmenting the ribs and spine according to the received data representing the three-dimensional diagnostic image, Steps include detecting and labeling the rib midline from rib segmentation, The steps include detecting and labeling central vertebral body landmarks from spinal segmentation, A step of defining a two-dimensional manifold plane that represents a visualization canvas for displaying a reformatted image, The steps include mapping the three-dimensional positions of the rib centerline and the vertebral body center landmark, which correspond to the received data representing the three-dimensional diagnostic image, to two-dimensional positions on the defined two-dimensional manifold plane, A step of interpolating missing three-dimensional position coordinates on a defined two-dimensional manifold plane, comprising deforming the two-dimensional manifold plane so that it aligns with the detected rib centerline and vertebral center landmarks, The steps include sampling the image intensity at each coordinate of the deformed two-dimensional manifold plane from the three-dimensional diagnostic image space, A step of generating a manifold slice from the sampled image intensity, wherein the manifold slice displays a continuous and linearized visual representation of the rib cage and spine. The steps include shifting the deformed two-dimensional manifold plane along its normal direction, repeating the sampling, and generating a stack of manifold slices that cover a complete three-dimensional visual representation of the thoracic cage and spine, The steps include generating a three-dimensional visual representation of the thoracic cage and vertebral bodies as a stack of manifolds covering the entire thoracic cage, An image processing device that executes an image.

2. The image processing apparatus according to claim 1, wherein the segmentation of the ribs and the vertebrae is performed by machine learning or deep learning techniques including at least one of neural networks, logistic regression, random forests, nearest neighbor analysis, and cluster or multivariate analysis.

3. The image processing apparatus according to claim 1, wherein the interpolation is performed by a thin-plate spline technique.

4. The image processing apparatus according to claim 1, wherein the processor further executes the computer-executable instructions to cause the image processing apparatus to calculate the normal direction for each point on the two-dimensional manifold plane.

5. The image processing apparatus according to claim 1, wherein the stack of manifold slices generated is a stack of interpolated two-dimensional multi-section reconstructions.

6. The image processing apparatus according to claim 5, wherein the processor is further configured to execute the computer executable instructions to generate three-dimensional images of the thoracic cage and spine based on the stack of two-dimensional multi-sectional reconstructions.

7. The image processing apparatus according to claim 1, wherein the sampled image intensity at each position in the deformed two-dimensional manifold plane corresponds to at least one of the regions of tissue between the ribs and the regions of tissue adjacent to the ribs.

8. A computer implementation method for processing three-dimensional images, The steps include receiving data representing a three-dimensional diagnostic image including the target ribs and spine, The steps include segmenting the ribs and spine according to the received data representing the three-dimensional diagnostic image, Steps include detecting and labeling the rib midline from rib segmentation, The steps include detecting and labeling central vertebral body landmarks from spinal segmentation, A step of defining a two-dimensional manifold plane that represents a visualization canvas for displaying a reformatted image, The steps include mapping the three-dimensional positions of the rib centerline and the vertebral center landmarks corresponding to the received data representing the three-dimensional diagnostic image to two-dimensional positions on the defined two-dimensional plane, A step of interpolating missing three-dimensional position coordinates on a defined two-dimensional manifold plane, comprising deforming the two-dimensional manifold plane so that it aligns with the detected rib centerline and vertebral center landmarks, The steps include sampling image intensity from the three-dimensional diagnostic image space at each coordinate of the modified two-dimensional manifold plane, A step of generating a reformatted image as a manifold slice from the sampled image intensity, wherein the manifold slice displays a continuous and linearized visual representation of the rib cage and spine. A step of shifting the deformed two-dimensional manifold plane along its normal direction and repeating the sampling to generate a stack of manifold slices covering a complete three-dimensional visual representation of the thoracic cage and spine, wherein the three-dimensional visual representation of the thoracic cage and vertebral center is generated as a stack of manifolds covering the complete thoracic cage. A method of having.

9. The method according to claim 8, wherein the segmentation is performed using machine learning or deep learning techniques, and the machine learning or deep learning techniques include at least one of neural networks, logistic regression, random forests, nearest neighbors, and cluster or multivariate analysis.

10. The method according to claim 8, wherein the interpolation is performed by thin-plate spline technology.

11. The method according to claim 8, further comprising the step of calculating the normal direction for each point on the two-dimensional plane.

12. The method according to claim 8, wherein the stack of generated manifold slices is a stack of interpolated two-dimensional multi-section reconstructions.

13. The method according to claim 12, further comprising the step of generating three-dimensional images of the thoracic cage and spine based on the aforementioned stack of interpolated two-dimensional multi-sectional reconstructions.

14. The method according to claim 8, wherein the sampled image intensity at each position in the deformed two-dimensional manifold plane corresponds to at least one of the tissue regions between the ribs and the tissue regions adjacent to the ribs.

15. A non-temporary computer-readable medium storing instructions for a processing circuit to execute a process, wherein the process is: The steps include receiving data representing a three-dimensional diagnostic image including the target ribs and spine, The steps include segmenting the ribs and spine according to the received data representing the three-dimensional diagnostic image, Steps include detecting and labeling the rib midline from rib segmentation, The steps include detecting and labeling central vertebral body landmarks from spinal segmentation, A step of defining a two-dimensional manifold plane representing a visualization canvas for displaying a reformatted image, wherein the two-dimensional manifold plane is deformed so as to align with the detected rib centerline and vertebral center landmarks, The steps include mapping the three-dimensional positions of the rib centerline and vertebral center landmarks corresponding to the received data representing the three-dimensional diagnostic image to two-dimensional positions on the defined two-dimensional plane, A step of interpolating missing three-dimensional position coordinates on a defined two-dimensional manifold plane, comprising deforming the two-dimensional manifold plane so that it aligns with the detected rib centerline and vertebral center landmarks, The steps include sampling image intensity from the three-dimensional diagnostic image space at each coordinate of the modified two-dimensional manifold plane, A step of generating a manifold slice from the sampled image intensity, wherein the manifold slice displays a continuous and linearized visual representation of the rib cage and spine. The steps include shifting the deformed two-dimensional manifold plane along its normal direction, repeating the sampling, and generating a stack of manifold slices that cover a complete three-dimensional visual representation of the entire rib cage, A step of generating a three-dimensional visual representation of the thoracic cage and vertebral bodies as a stack of manifolds covering the entire thoracic cage, A computer-readable medium having [a certain characteristic].

Images