Method and system for removing bed image of CT image and application

Abstract

The invention relates to a method and a system for removing a bed image of a CT (Computed Tomography) image and application, and the method sequentially comprises the following steps: obtaining a continuous CT image sequence of a patient, and constructing three-dimensional body data; carrying out projection analysis on the three-dimensional body data along the z-axis direction, and respectively calculating to obtain a maximum value projection image and a minimum value projection image; performing first threshold segmentation based on the minimum projection image to generate a first mask; performing second threshold segmentation based on the maximum projection image to generate a second mask; and based on the first mask and the second mask, generating an accurate bed area mask which does not contain human tissues. According to the method, the maximum projection (reflecting the upper limit of the tissue) and the minimum projection (reflecting the characteristics of the bed) are fused, so that the segmentation ambiguity of a single threshold value method in an overlapped CT value interval is overcome, and the accurate separation of the bed and the human tissue is realized.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a method, system, and application for removing bedside images from CT images. Background Technology

[0002] During CT scans, patients typically lie on a scanning table for data acquisition. When performing volume rendering (VRT) reconstruction, the table (usually made of high-strength, low-density materials such as carbon fiber) appears as a solid structure in the image, severely interfering with the doctor's observation of the patient's anatomical structures, especially in 3D images of areas such as the abdomen, pelvis, and lower limbs.

[0003] Currently, common methods for removing bed pedestals mainly rely on simple global thresholding. However, because the CT value (Henness unit, HU) range of bed pedestal materials overlaps with some soft tissues (such as muscles and organs) and fluids in the human body, a single threshold method is difficult to accurately distinguish. If the threshold is set too high, the bed pedestal structure may be mistakenly preserved or incompletely removed; if the threshold is set too low, it may incorrectly erode or remove patient tissues, resulting in incomplete 3D models and affecting diagnostic accuracy.

[0004] In addition, some methods based on region growing or edge detection are prone to segmentation leakage or insufficiency because the boundary between the bed and the human body is blurred and the bed has an irregular shape in the cross section (XY plane).

[0005] Therefore, there is an urgent need for a robust, automated, and precise method to separate the bed from human tissue.

[0006] In view of the above-mentioned shortcomings, the designer has actively researched and innovated in order to create a method, system and application for removing bedside images from CT images, making it more industrially valuable. Summary of the Invention

[0007] To address the aforementioned technical problems, the present invention aims to provide a method, system, and application for removing bedside images from CT scans. This invention relates to the field of medical image processing technology, and particularly to a three-dimensional visualization preprocessing method for computed tomography (CT) images. Specifically, it is a method for automatically and accurately segmenting and removing bedside images based on image morphological projection and logical operations.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: One of the objectives of this invention is: A method for removing bedside images from CT scans includes the following steps: Step 1: Obtain a sequence of continuous CT tomographic images of the patient and construct three-dimensional volume data V(x, y, z), where the x-axis and y-axis represent pixel coordinates and the z-axis represents the head-to-toe direction of the patient; Step 2: Perform projection analysis on the 3D volume data along the z-axis to calculate the maximum value projection image MIP(x,y) and the minimum value projection image MinIP(x,y). Step 3: Perform first threshold segmentation based on the minimum value projection image to generate a first mask. The first mask is designed to completely cover the scanning bed area and allow the inclusion of first type of interfering tissue (i.e., "partial non-bed tissue structures" in the first mask). Step 4: Perform a second threshold segmentation based on the maximum projection image to generate a second mask, which is designed to completely cover the human tissue region; Step 5: Based on the first and second masks, generate a precise bed area mask that does not contain human tissue through a series of predetermined morphological operations and logical operations; Step 6: Extend the precise bed area mask to three-dimensional space, and remove or make the corresponding voxels transparent from the original three-dimensional volume data for subsequent three-dimensional visualization rendering.

[0009] As a further improvement of the present invention, step 5 includes the following steps in sequence: Step 51: Subtract the first mask from the second mask to generate the third mask; Step 52: Fill the air outside the boundary of the human tissue region image in the third mask to obtain the filled fourth mask; Step 53: Perform arithmetic and logical operations on the first mask and the fourth mask, and subtract the defined non-bed area represented by the fourth mask from the first mask to generate an accurate bed area mask.

[0010] As a further improvement of the present invention, step 51 further includes performing at least one expansion erosion operation on the generated third mask to eliminate fine lines caused by mask alignment issues.

[0011] As a further improvement of the present invention, the filling operation in step 52 is specifically as follows: starting from the image border, reverse filling is performed to fill all background areas connected to the border as foreground.

[0012] As a further improvement of the present invention, when the CT value of the bed is lower than the CT value of human skin, step 5 includes the following steps in sequence: Step 51: Fill the holes in the discontinuous areas inside the human tissue region in the second mask to obtain the filled third mask; Step 52: Perform a logical AND-NOT operation on the first mask and the third mask, and remove the specific human tissue area represented by the third mask from the first mask, thereby generating a precise bed area mask.

[0013] As a further improvement of the present invention, the filling operation in step 51 is specifically as follows: taking the foreground as the seed area, reverse filling is started inside the human tissue, and all background areas connected to the border are filled as the foreground.

[0014] As a further improvement of the present invention, the second threshold used for the second threshold segmentation.

[0015] in, Figure 13 In a CT image, the area indicated by the arrow in the image below represents the bed. Based on the CT value distribution corresponding to the vertical line above, we can see that the actual CT values ​​of the bed are mostly between -400 and +300. On the MIP (Multi-Input Perimeter) image, the grayscale value of the bed is between 100 and 500, and on the MinIP image, it is between -200 and -600. This situation determines the selection of the threshold in the following section.

[0016] Although theoretically "the typical CT value of carbon fiber bed boards is in the range of approximately -200 HU to +50 HU, and most often appears in the region close to 0 HU", due to manufacturing processes and the resolution and errors of CT imaging, the actual CT value of the bed board in CT images is very large (in the CT data of this invention, the actual CT value of the bed board is mostly between -400 and +300).

[0017] The second objective of this invention is: A system for removing bedside images from CT scans includes: Memory, used to store computer programs; A processor for executing computer programs to implement a method for removing bedside images from CT images, as described above.

[0018] The third objective of this invention: A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for removing CT image bedside images as described above.

[0019] The fourth objective of this invention: The application of any of the above methods for removing bedside images from CT images in CT 3D visualization.

[0020] By means of the above-described solution, the present invention has at least the following advantages: High-precision segmentation: By fusing the maximum value projection (reflecting the upper limit of tissue) and the minimum value projection (reflecting the characteristics of the bed), the segmentation ambiguity of the single threshold method in the overlapping CT value range is overcome, and the precise separation of the bed and human tissue is achieved.

[0021] Strong robustness: The method is not dependent on the specific shape of the bed or the complex boundaries of contact with the human body, and is applicable to scanning beds of different models and postures.

[0022] High degree of automation: The entire process is based on global thresholding and morphological operations of the projected image, eliminating the need for manual drawing or complex parameter adjustments, and enabling batch automatic processing.

[0023] Good integrity: By combining the logical strategy of "broad inclusion" and "strict exclusion", it can ensure that the bed is completely removed and protect human tissue from being accidentally deleted to the greatest extent, thus ensuring the anatomical integrity of the final 3D model.

[0024] Computationally efficient: The main operations are performed on two two-dimensional projection images, avoiding complex iterative calculations in three-dimensional space. The processing speed is fast and suitable for the needs of real-time or near-real-time clinical preview.

[0025] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the following are preferred embodiments of the present invention described in detail with reference to the accompanying drawings. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart illustrating the first embodiment of the present invention; Figure 2 This is a schematic diagram of the minimum projection image (MinIP) in the first embodiment of the present invention; Figure 3 It is by Figure 2 A schematic diagram of the generated "mask containing a complete bed and a small amount of tissue" (Mask_Table_Min); Figure 4 This is a schematic diagram of the maximum value projection image (MIP) in the first embodiment of the present invention; Figure 5 It is by Figure 4A schematic diagram of the generated "mask containing the complete bed and a large amount of tissue" (Mask_Table_Max); Figure 6 This is a schematic diagram of Mask_NoTable_Hole, which is a "mask that does not contain a bed table" in the first embodiment of the present invention; Figure 7 This invention is by Figure 6 A diagram illustrating Mask_NoTable_Filled after the fill operation; Figure 8 This is a schematic diagram of the "precise bed mask" (Mask_FinalTable) obtained by arithmetic logic operations in the first embodiment of the present invention; Figure 9 This invention is by Figure 8 A schematic diagram of the result after expansion; Figure 10 This is a rendering of the 3D VRT image before the application of the bedside mask in this invention; Figure 11 This is a rendering of the 3D VRT image after applying the bed mask according to the present invention; Figure 12 This is a flowchart illustrating the second embodiment of the present invention.

[0028] Figure 13 These are CT images illustrating the threshold range of the present invention. Detailed Implementation

[0029] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0030] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0031] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method for automatically, accurately, and completely removing scanned bed images from CT image sequences. This method effectively solves the problem of difficulty in separating the bed and human tissues due to their similar CT values ​​by combining multi-planar projection analysis along the Z-axis with morphological logic operations. This results in a clean three-dimensional image containing only the patient's anatomical structures, improving the visual effect and clinical diagnostic value of 3D VRT.

[0032] Considering that the position of the CT table is relatively fixed in each frame of the CT sequence, it is possible to detect and generate the table mask on the Z-axis projection of the CT sequence. However, directly using threshold segmentation will result in the segmentation of soft tissue parts of the human body that fall within the CT value range of the table, resulting in an incomplete table mask.

[0033] This innovative approach simultaneously performs maximum value projection MIP and minimum value projection MinIP along the Z-axis. In the minimum projection, threshold segmentation and binarization are performed based on the CT value of the bed to obtain the first mask MASK1, which contains the complete bed. This mask also contains some non-bed tissue structures, but the gap between the two is large. In the maximum value projection, threshold segmentation and binarization are performed based on the CT value of the bed to obtain a second mask MASK2 containing the complete bed. This mask also contains complete tissue structures with very small gaps between them, making them difficult to separate. This mask is used for subsequent separation of "non-bed tissue structures" in the first mask.

[0034] The second mask is subtracted from the first mask to obtain the third mask, MASK3. This mask does not include the bed frame, but the subtraction results in intact edges of the tissue structure, but with voids inside (negative images of a small amount of human tissue contained in the first mask).

[0035] The third mask fills the air outside the image boundary of the human tissue region, starting from the air area and filling the non-bed area, leaving only a small amount of negative human tissue contained in the first mask. After the third mask is filled, it is further subjected to arithmetic and logical operations with the first mask to generate the "precise bed mask" MASK4.

[0036] First embodiment of the present invention: like Figures 1-11 As shown, this embodiment provides a method for removing CT image bed table images based on image morphology processing, including the following steps: Step S1: 3D data preparation. Obtain a set of consecutive CT tomographic image sequences and stack them to form a 3D volume data V(x, y, z), where x and y represent pixel coordinates and z represents the slice number (i.e., the Z-axis, the head-to-toe direction of the patient).

[0037] Step S2: Projection analysis in the Z-axis direction.

[0038] S21: Calculate the maximum intensity projection image MIP(x, y). Perform a maximum intensity projection on the volume data V in the Z-axis direction, that is, for each (x, y) position, take the maximum value of the CT values in all z slices: MIP(x, y) = max( V(x, y, z) ), z. This image highlights the densest structures throughout the scanned range.

[0039] S22: Calculate the minimum intensity projection image MinIP(x, y). Perform a minimum intensity projection on the volume data V in the Z-axis direction, that is, for each (x, y) position, take the minimum value of the CT values in all z slices: MinIP(x, y) = min( V(x, y, z) ), z. This image highlights the structures with the lowest density throughout the scanned range, such as air and the bed.

[0040] Step S3: Generate a "mask containing the complete bed and a small amount of tissue" (MASK1, the first mask).

[0041] Perform threshold segmentation on the minimum intensity projection image MinIP. By selecting an appropriate first threshold T_low (e.g., -700HU, usually made configurable), binarize MinIP: Mask_Table_Min(x, y) = 1 if MinIP(x, y) < T_low else 0. The resulting binary mask Mask_Table_Min aims to encompass the bed area as completely as possible, allowing for the inclusion of a small amount of human tissue while ensuring no omission of the bed. Subsequently, it is necessary to find a way to remove the small amount of human tissue content based on this to obtain the final "accurate bed mask".

[0042] Step S4: Generate a "mask containing the complete bed and a large amount of tissue" (MASK2, the second mask).

[0043] Thresholding is performed on the maximum projection image MIP. In MIP, a second threshold T_high (e.g., 0HU, usually configurable) higher than T_low is selected, and MIP is binarized: Mask_Table_Max(x, y) = 1 if MIP(x, y) > T_high else 0. The resulting binary mask Mask_Table_Max aims to encompass the bed area as completely as possible, while including a large amount of human tissue, which covers the human tissue in step S3 (first mask), ensuring no part of the bed is missed.

[0044] Step S5: Separate from the first mask to generate a "mask that does not contain the bed at all" (MASK3, the third mask).

[0045] Arithmetic calculations (subtraction) are performed on the masks obtained in steps S3 and S4 to remove the bed-like portion of the mask. Since both the first and second masks contain the bed-like portion, and the human tissue in the second mask overlaps the human tissue in the first mask, we can obtain the negative superposition of the human tissue in the second mask and the human tissue in the first mask through arithmetic operations on the two masks: subtracting the first mask from the second mask. Mask_NoTable_Hole = Mask_Table_Max - Mask_Table_Min. This result needs to be subjected to dilation and erosion operations to eliminate fine lines caused by mask alignment issues.

[0046] Step S6: Morphological filling of the third mask.

[0047] Because the air outside the human tissue region image boundary in Mask_NoTable_Hole is connected, a filling operation is required. After the air outside the human tissue region image boundary is filled, only a small amount of negative human tissue contained in the first mask remains. The method for filling the air area is as follows: using the foreground (Mask_NoTable_Hole = 1) as the seed region, reverse filling (or air filling) is performed starting from the image border (representing the air outside the scanning range), filling all background (Mask_NoTable = 0) areas connected to the border as foreground. This step yields the filled mask Mask_NoTable_Filled. Its physical meaning is: connecting the human tissue region and the bed area of ​​the second mask, leaving only a small amount of negative human tissue contained in the first mask.

[0048] Step S7: Arithmetic and logical operations generate the "precise bed mask" (MASK4, the fourth mask).

[0049] Perform arithmetic and logical operations on Mask_Table_Min (broadly including the bed frame) obtained in step S1 and Mask_NoTable_Filled (determining the non-bed frame area) obtained in step S6: Mask_FinalTable(x, y) = Mask_Table_Min(x, y) - (NOT Mask_NoTable_Filled(x, y)).

[0050] Explanation of the operation: From the broad candidate bed table area (Mask_Table_Min), subtract those areas that are definitely part of the human body or the interior space of the human body (Mask_NoTable_Filled). The remaining part is the precisely segmented final mask Mask_FinalTable that contains only the bed table. This result needs to be dilated to obtain better coverage of the bed table area.

[0051] Step S8: Apply to 3D VRT display.

[0052] The final bedside mask Mask_FinalTable is extended to three dimensions: for each Z-axis position z, the two-dimensional mask Mask_FinalTable(x, y) is used as the bedside region identifier for that slice. Before 3D VRT rendering, the CT values ​​of all voxels in the volume data V(x, y, z) corresponding to Mask_FinalTable(x, y) = 1 are set to a completely transparent value (e.g., -1000 HU, air value), or they are directly removed from the rendering data. Subsequently, standard volumetric rendering is performed on the processed volume data to obtain a three-dimensional image that does not include the scanning bedside and only clearly shows the patient's anatomical structures.

[0053] Summary of the innovations in this embodiment: 1. Steps for "Dual Projection Analysis": This embodiment is the first to creatively "functionally pair" these two projections and apply them to the unique challenge of "bed table partitioning".

[0054] The function of MIP has been redefined: In this scheme, the primary purpose of MIP is not to observe blood vessels or bones, but to capture the "upper limit of density existence" of human tissue in the Z-axis direction, in order to ensure the spatial continuity of human tissue and provide a "deterministic human tissue template" for subsequent logical operations.

[0055] The function of MinIP has been redefined: in this approach, the primary purpose of MinIP is to capture the "density stability characteristics" of the bed in the Z-axis direction. Since the bed consistently exhibits low density in every slice, while human tissue only shows extremely low density in certain layers (such as air-containing chambers), MinIP can most stably highlight the overall shape of the bed.

[0056] This embodiment addresses the specific problems mentioned above by providing a novel and complementary functional definition and combination of the two general tools.

[0057] 2. Steps to "Generate MASK1 / MASK2": The innovation of this step lies in the adoption of a "dual and contradictory threshold strategy," which is the key invention to resolve the ambiguity in the segmentation of the overlapping area of ​​CT values ​​between the bed and the tissue.

[0058] The generation logic of MASK1 is as follows: a low threshold T_low is used for MinIP, aiming to unconditionally guarantee that the bed is 100% captured, even if it results in the accidental inclusion of some low-density tissue (Type I interference). This is a strategic tolerance.

[0059] The generation logic of MASK2 is to use a high threshold T_high on the MIP to obtain a high-confidence, continuous human tissue region. This is a strategic guarantee.

[0060] The generation of these two masks is not an independent event, but rather two intentionally biased intermediate products serving a higher-level segmentation strategy (i.e., a subsequent "include-exclude" strategy). This provides a novel design approach.

[0061] 3. Steps for "generating MASK_Final through logical and morphological operations": The creativity of this step lies in designing a "logical decision chain" that is interconnected and information is extracted step by step. The order and combination of this chain are specific and irreversible.

[0062] The chain reaction design is as follows: MASK2 - MASK1 → Filling → MASK1 - (NOT Filling Result). This step is the core transformation hub. It utilizes the "tissue integrity" of MASK2 to counteract the "tissue interference" in MASK1, generating a "negative image of human tissue with holes." This step cleverly transforms the difficult problem of "removing tissue from MASK1" into the more easily solved morphological problem of "filling holes (air areas)."

[0063] Specificity of the filling operation: The (air area) filling here is background filling starting from the image boundary, not ordinary hole filling. After thresholding the projection binarization segmentation of MIP, because the threshold value is <0, no voids will appear inside human tissue. This is because a normal scan covers a large Z-axis distance (including at least one complete organ), and the maximum density projection normally will not leave voids inside the tissue boundary. In a few special cases, the user can manually circle the bed frame.

[0064] This series of routine operations, their specific permutations and combinations, the underlying physical logic, and the final "indirect segmentation" effect constitute a complete, non-obvious technical concept.

[0065] 4. The synergistic effect among the above steps: Synergistic effect: Dual projection provides complementary and reliable input features for dual threshold masks; the difference between dual threshold masks provides the operational basis for logical decision chains; the logical decision chains ultimately realize the accurate extraction of the bed table from the 3D volume data and its application to 3D visualization.

[0066] Unexpected technical effects: This embodiment solves the problem of "CT value overlap ambiguity" that traditional methods cannot overcome: both the single threshold method and the region growing method fail in this region. This solution bypasses this problem by using the statistical properties of the projection space through the synergy of MASK1 (wide inclusion) and MASK2 (strict screening) without relying on complex 3D segmentation.

[0067] Robust segmentation of "blurred contact regions" is achieved: traditional edge detection or watershed algorithms are prone to failure in this area. This scheme transforms the complex boundary separation problem in 3D space into a logical judgment problem in 2D projection space through the synergy of logical subtraction and boundary filling, greatly improving stability.

[0068] This forms a complete closed loop from "noise tolerance" to "precise results": the solution begins with active tolerance of noise and interference (MASK1 contains interference), and through layers of filtering and correction in intermediate steps (logic chains), it finally outputs highly accurate results. This "retreating to advance" strategy is the essence of synergy.

[0069] The technical solution in this embodiment is not a series of discrete steps, but a "systematic solution" precisely designed to solve a specific technical problem. Each step is interdependent, information flows unidirectionally, and functions are tightly coupled, together realizing a completely new paradigm for bed table partitioning.

[0070] The second embodiment of the present invention: If the CT value of the bed frame is less than the CT value of human skin, then the above solution can be simplified to: As Figure 12 shown, a method for removing CT image table images based on image morphology processing in this embodiment includes the following steps: Step S1: Three-dimensional data preparation. Obtain a set of continuous CT tomographic image sequences and stack them to form a three-dimensional volume data V(x, y, z), where x and y represent pixel coordinates and z represents the slice number (i.e., the Z-axis, the patient's head-foot direction).

[0071] Step S2: Projection analysis in the Z-axis direction.

[0072] S21: Calculate the maximum intensity projection image MIP(x, y). Perform a maximum intensity projection on the volume data V in the Z-axis direction, that is, for each (x, y) position, take the maximum CT value among all z slices: MIP(x, y) = max(V(x, y, z)), z. This image highlights the densest structures throughout the scan range.

[0073] S22: Calculate the minimum intensity projection image MinIP(x, y). Perform a minimum intensity projection on the volume data V in the Z-axis direction, that is, for each (x, y) position, take the minimum CT value among all z slices: MinIP(x, y) = min(V(x, y, z)), z. This image highlights the lowest density structures throughout the scan range, such as air and the table.

[0074] Step S3: Generate a "mask containing the complete table" (the first mask).

[0075] Perform threshold segmentation on the minimum intensity projection image MinIP. Since the table appears as a stable low-density area (CT value < 0 HU) in MinIP, while human tissues are denser in at least some slices, by selecting an appropriate first threshold T_low (for example, -800 HU, which is a hypothetical value), binarize MinIP: Mask_Table(x, y) = 1 if MinIP(x, y) < T_low else 0. The resulting binary mask Mask_Table aims to encompass the table area as completely as possible, such as Figure 3 , including the shadows left by a small amount of high-density substances such as bones through MinIP.

[0076] Step S4: Generate a "mask completely without the table" (the second mask).

[0077] Thresholding segmentation is performed on the maximum value projection image MIP. In the MIP, the bedside area, due to its low-density characteristics, still has a maximum value far lower than that of human soft tissue (skin). Therefore, a second threshold T_high (e.g., 0HU, theoretically the lower the better, so that the human tissue area is intact, and even avoids subsequent "reverse filling (or hole filling) within the human tissue") is selected, and the MIP is binarized: Mask_NoTable(x, y) = 1 if MIP(x, y) > T_high else 0. The resulting binary mask Mask_NoTable aims to confidently include only human tissue (complete skin edges), completely excluding the bedside area.

[0078] Step S5: Morphological filling of the second mask.

[0079] Because the human tissue region in Mask_NoTable may have discontinuities (such as inter-organ spaces, areas with CT values ​​lower than skin, etc.), a filling operation is required. Using the foreground (Mask_NoTable = 1) as the seed region, reverse filling (or hole filling) begins inside the human tissue, filling all background (Mask_NoTable = 0) regions connected to the border as the foreground. This step yields the filled mask Mask_NoTable_Filled. Its physical meaning is: marking the entire continuous human tissue region "definitely not a bedside table", which covers the small amount of human tissue included in step 3.

[0080] Step S6: Logical operations generate the "precise bed mask" (third mask).

[0081] Perform a logical AND-NOT operation between the Mask_Table obtained in step S3 (broadly including the bed frame) and the Mask_NoTable_Filled obtained in step S5 (determining the non-bed frame area): Mask_FinalTable(x, y) = Mask_Table(x, y) AND (NOT Mask_NoTable_Filled(x, y)).

[0082] Explanation of the operation: From the broad candidate bed area (Mask_Table), subtract those regions that are definitely part of the human body or the interior space of the human body (Mask_NoTable_Filled). The remaining part is the final mask (Mask_FinalTable) that is precisely segmented and contains only the bed.

[0083] Step S7: Apply to 3D VRT display.

[0084] The final bedside mask Mask_FinalTable is extended to three dimensions: for each Z-axis position z, the two-dimensional mask Mask_FinalTable(x, y) is used as the bedside region identifier for that slice. Before 3D VRT rendering, the CT values ​​of all voxels in the volume data V(x, y, z) corresponding to Mask_FinalTable(x, y) = 1 are set to a completely transparent value (e.g., -1000 HU, air value), or they are directly removed from the rendering data. Subsequently, standard volumetric rendering is performed on the processed volume data to obtain a three-dimensional image that does not include the scanning bedside and only clearly shows the patient's anatomical structures.

[0085] However, some bed frames use a high-density material to ensure strength, so the bed frame mask cannot be obtained directly through logical operations. In this case, it needs to be calculated according to the steps in the first embodiment.

[0086] The steps in the first embodiment calculate the cases where the CT value of the covered bed is greater than or equal to the CT value of human skin and the CT value of the bed in the second embodiment is less than the CT value of human skin. This is a more general solution that does not require the CT value of the bed to be less than the CT value of human skin, and is applicable to CT equipment data in more situations.

[0087] Furthermore, more morphological corrosion and expansion steps can be added to the steps of the above embodiments to achieve better results.

[0088] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0089] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0090] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for removing bedside images from CT scans, characterized in that: The steps are as follows: Step 1: Obtain a sequence of continuous CT tomographic images of the patient and construct three-dimensional volume data V(x, y, z), where the x-axis and y-axis represent pixel coordinates and the z-axis represents the head-to-toe direction of the patient; Step 2: Perform projection analysis on the three-dimensional volume data along the z-axis to calculate the maximum value projection image MIP(x, y) and the minimum value projection image MinIP(x, y). Step 3: Perform first threshold segmentation based on the minimum value projection image to generate a first mask. The first mask is designed to completely cover the scanning table area and allow the inclusion of first type of interfering tissue. Step 4: Perform a second threshold segmentation based on the maximum value projection image to generate a second mask, which is designed to completely cover the human tissue area; Step 5: Based on the first mask and the second mask, generate a precise bed area mask that does not contain human tissue through a series of predetermined morphological operations and logical operations; Step 6: Extend the precise bed area mask to three-dimensional space, and remove or make the corresponding voxels transparent from the original three-dimensional volume data for subsequent three-dimensional visualization rendering.

2. The method for removing bedside images from CT images as described in claim 1, characterized in that, Step 5 includes the following steps in sequence: Step 51: Subtract the first mask from the second mask to generate the third mask; Step 52: Fill the air outside the boundary of the human tissue region image in the third mask to obtain the filled fourth mask; Step 53: Perform arithmetic logic operations on the first mask and the fourth mask to subtract the defined non-bed area represented by the fourth mask from the first mask, thereby generating the precise bed area mask.

3. The method for removing bedside images from CT images as described in claim 2, characterized in that, Step 51 also includes performing at least one dilatational erosion operation on the generated third mask to eliminate fine lines caused by mask alignment issues.

4. The method for removing bedside images from CT images as described in claim 2, characterized in that, The filling operation in step 52 is specifically as follows: starting from the image border, perform reverse filling to fill all background areas connected to the border as foreground.

5. The method for removing bedside images from CT images as described in claim 1, characterized in that, When the CT value of the bedside table is lower than the CT value of human skin, step 5 includes the following steps in sequence: Step 51: Fill the holes in the discontinuous areas inside the human tissue region in the second mask to obtain the filled third mask; Step 52: Perform a logical AND-NOT operation on the first mask and the third mask to remove the defined human tissue region represented by the third mask from the first mask, thereby generating the precise bed area mask.

6. The method for removing bedside images from CT images as described in claim 5, characterized in that, The filling operation in step 51 is as follows: using the foreground as the seed area, reverse filling is performed inside the human tissue to fill all background areas connected to the border as the foreground.

7. The method for removing bedside images from CT images as described in claim 1, characterized in that, The second threshold used for the second threshold segmentation has a value higher than the first threshold.

8. A system for removing bedside images from CT scans, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement a method for removing bedside images from CT images as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of a method for removing CT image bedside images as described in any one of claims 1 to 7.

10. The application of the method for removing bedside images from CT images as described in any one of claims 1 to 7 in CT three-dimensional visualization.

Images