Medical image reconstruction methods

Abstract

The present invention discloses a medical image reconstruction method, characterized by comprising: obtaining an original image; pre-processing the original image to obtain a second image; obtaining a first head skin mask that matches the second image using a threshold segmentation method; performing edge detection on the second image to obtain a first edge mask; filtering the first edge mask using the first head skin mask, retaining only the portion of the first edge mask corresponding to the head contour, to obtain a filtered second edge mask; performing morphological closing and hole filling operations on the second edge mask to obtain a second head skin contour mask; and reconstructing the skin based on the second image and the head skin contour mask. The medical image reconstruction method provided by the present invention has strong noise immunity and resistance to interference from head attachments, and generates a high-precision, smooth mesh that is unaffected by internal head tissue.

Description

Technical Field

[0001] The present invention relates to the technical field of computer vision and graphic image processing, and more particularly to a medical image reconstruction method. Background Art

[0002] Surgical navigation applications require establishing a correspondence between 3D images captured by devices like MRI (magnetic resonance imaging) and CT (computed tomography) and actual human tissue in the real world. This correspondence is typically established by registering the 3D spatial position of human surface tissue captured by the navigation camera with a human skin model constructed from the scanned images. Therefore, establishing an accurate 3D human skin model from the images is essential for this process. The step of establishing the human skin contour from the 3D images is called skin reconstruction.

[0003] Skin reconstruction is a technique for obtaining a 3D mesh representation of the skin's outer surface from a 3D image. Existing methods primarily use isosurface extraction algorithms to obtain this mesh representation. This method connects image pixels with identical or similar pixel values ​​into a curved surface and represents it as a 3D mesh. This method has three problems:

[0004] 1. Poor noise immunity: This method is easily affected by noise with an intensity close to that of skin pixels, resulting in inaccurate skin contour structure. In addition, this method is also easily affected by motion artifacts and magnetic susceptibility artifacts in MRI imaging, resulting in unstable reconstruction results.

[0005] 2. Susceptible to interference from objects attached to the head: MRI images objects attached to the head, such as glasses and oxygen masks, which can interfere with the skin contour extraction using the isosurface method and be reconstructed as part of the skin.

[0006] 3. It will generate a complex internal tissue structure of the head: Using the isosurface extraction algorithm will indiscriminately extract the isosurfaces of the skin and the inside of the head to construct a three-dimensional mesh representation. However, during surgical navigation and positioning, only the three-dimensional structure of the outer surface of the skin is required, and the three-dimensional mesh inside the head is not needed. The three-dimensional mesh inside the head will only increase the amount of calculation during operation and the amount of data during storage and transmission, which is meaningless to the task objectives.

[0007] Therefore, there is an urgent need to provide a medical image reconstruction method that can resist noise, remove interference from head attachments, and remove internal tissue structures of the head. Summary of the Invention

[0008] In view of this, the present invention provides a medical image reconstruction method, comprising:

[0009] Get the original image;

[0010] Pre-processing the original image to obtain a second image; obtaining a first head skin mask matching the second image by a threshold segmentation method;

[0011] performing edge detection on the second image to obtain a first edge mask;

[0012] filtering the first edge mask using the first head skin mask, retaining only a portion of the first edge mask corresponding to the head contour, to obtain a filtered second edge mask;

[0013] Performing a morphological closing operation and a hole filling operation on the second edge mask to obtain a second head skin contour mask;

[0014] Skin reconstruction is performed according to the second image and the head skin contour mask.

[0015] Optionally, pre-processing the original image to obtain the second image includes:

[0016] Preprocessing the original image to obtain a first image;

[0017] Perform noise reduction processing on the first image to obtain the second image.

[0018] Optionally, performing noise reduction processing on the first image to obtain the second image includes: performing noise reduction processing on the first image by bilateral filtering using an automatic image processing tool to obtain the second image;

[0019] Optionally, obtaining a first head skin mask matching the second image by a threshold segmentation method includes:

[0020] Set the binarization threshold to M, set voxels larger than M in the second image to 1, and set voxels smaller than or equal to M in the second image to 0, to obtain a binarization mask of the same size as the second image;

[0021] The first head skin mask is obtained by performing a maximum connected domain, a morphological closing operation, and a hole filling operation on the binary mask.

[0022] Optionally, performing a maximum connected domain, a morphological closing operation, and a hole filling operation on the binary mask to obtain the first head skin mask includes:

[0023] A medical image processing software package is used to perform maximum connected domain, morphological closing operation and hole filling operation on the binary mask to obtain the first head skin mask.

[0024] Optionally, performing edge detection on the second image to obtain a first edge mask includes:

[0025] A medical image processing software package is used to perform edge detection on the second image using a Canny operator to obtain the first edge mask.

[0026] Optionally, performing a morphological closing operation and a hole filling operation on the second edge mask to obtain a second head skin contour mask includes:

[0027] A medical image processing software package is used to perform morphological closing and hole filling operations on the second edge mask to obtain the second head skin contour mask.

[0028] Optionally, performing skin reconstruction according to the second image and the second head skin contour mask includes:

[0029] Obtaining a head skin inner contour mask and a head skin outer contour mask according to the second head skin contour mask;

[0030] Skin reconstruction is performed according to the second image, the head skin inner contour mask, and the head skin outer contour mask.

[0031] Optionally, obtaining a head skin inner contour mask and a head skin outer contour mask according to the second head skin contour mask includes:

[0032] performing an erosion operation on the second head skin contour mask to obtain a head skin inner contour mask, and performing an expansion operation on the second head skin contour mask to obtain a head skin outer contour mask;

[0033] Optionally, performing skin reconstruction according to the second image, the head skin inner contour mask, and the head skin outer contour mask includes:

[0034] Setting the values ​​of the portion of the second image outside the head skin outer contour mask to 0, and setting the values ​​of the portion of the second image within the head skin inner contour mask to 255, to obtain a denoised image retaining only the skin contour;

[0035] Extracting the head skin from the second image retaining only the skin contour using an isosurface extraction algorithm to obtain an isosurface set;

[0036] A single-layer mesh surface is extracted from the isosurface set using a maximum continuous surface extraction method, which is a mesh representation of the head skin.

[0037] Optionally, extracting the head skin from the second image retaining only the skin contour using an isosurface extraction algorithm to obtain an isosurface set includes:

[0038] The isosurface extraction algorithm is executed using a 3D visualization software package to extract the obtained isosurface set.

[0039] Optionally, preprocessing the original image to obtain a first image includes:

[0040] The resolution of the original image is adjusted to a preset resolution using magnetic resonance data processing software, the voxels of the original image are set to preset values, and the pixel values ​​of the original image are converted to a range of 0 to 255.

[0041] Compared with the prior art, the medical image reconstruction method provided by the present invention achieves at least the following beneficial effects:

[0042] The medical image reconstruction method provided by the present invention has the characteristics of strong anti-noise ability, strong anti-interference ability of head attachments, high-precision and smooth generated grid, and no influence from the internal tissues of the head.

[0043] The medical image reconstruction method provided by the present invention first generates a regional mask containing the head skin boundary based on the original MRI image, then uses the regional mask to shield the image part except the skin, and then performs isosurface extraction on this basis to reconstruct a reliable head skin.

[0044] The medical image reconstruction method provided by the present invention uses a skin boundary area mask to shield most of the noise interference outside the skin and interference from the eyes, oxygen mask, and imaging artifacts. At the same time, this area mask also shields the influence of the complex tissue structure inside the head, making the reconstructed skin mesh a single-layer surface. Since there is no invalid internal head structure mesh, the amount of data required to represent the skin reconstruction result is less.

[0045] Of course, any product implementing the present invention does not necessarily need to achieve all of the technical effects described above at the same time.

[0046] Further features and advantages of the present invention will become apparent from the following detailed description of exemplary embodiments of the present invention with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention.

[0048] Figure 1 is a flow chart of the medical image reconstruction method provided by the present invention;

[0049] Figure 2 This is a flow chart of the medical image reconstruction method provided by the present invention;

[0050] Figure 3 is the first image obtained after preprocessing the original image;

[0051] Figure 4A second image is obtained after performing noise reduction processing on the first image;

[0052] Figure 5 This is the first head skin mask result image after threshold segmentation;

[0053] Figure 6 is a first edge mask obtained after edge detection on the second image;

[0054] Figure 7 is the second edge mask image;

[0055] Figure 8 is a schematic diagram of a head skin inner contour mask and a head skin outer contour mask obtained according to the second head skin contour mask;

[0056] Figure 9 It is a denoised image that only retains the skin contour;

[0057] Figure 10 yes Figure 9 A local magnified view of the boundary;

[0058] Figure 11 It is the result of skin reconstruction;

[0059] Figure 12 It is the skin reconstruction result directly using the isosurface extraction algorithm;

[0060] Figure 13 It is a cross-sectional image of the skin mesh directly using the isosurface extraction algorithm;

[0061] Figure 14 is the result of skin reconstruction using the method of the present invention;

[0062] Figure 15 This is a cross-sectional view of the skin mesh using the method of the present invention. DETAILED DESCRIPTION

[0063] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that unless otherwise specifically stated, the relative arrangement of components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention.

[0064] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.

[0065] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0066] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0067] It should be noted that like reference numerals and letters refer to like items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0068] Combine Figure 1 , Figure 1 is a flow chart of the medical image reconstruction method provided by the present invention, Figure 1 The medical image reconstruction method in the embodiment includes the following steps:

[0069] S101: Acquire original image;

[0070] S102: Pre-processing the original image to obtain a second image;

[0071] S103: Obtaining a first head skin mask that matches the second image by a threshold segmentation method;

[0072] S104: performing edge detection on the second image to obtain a first edge mask;

[0073] S105: filtering the first edge mask through the first head skin mask, retaining only the portion of the first edge mask corresponding to the head contour, to obtain a filtered second edge mask;

[0074] S106: performing a morphological closing operation and a hole filling operation on the second edge mask to obtain a second head skin contour mask;

[0075] S107: Perform skin reconstruction based on the second image and the head skin contour mask.

[0076] Specifically, both step S103 and step S104 process the second image. In terms of order, step S103 can be performed first and then step S104, or step S104 can be performed first and then step S103. Of course, steps S103 and S104 can also be processed simultaneously.

[0077] Combine Figures 3 to 11 , Figure 4 is the second image; Figure 5 A first head skin mask obtained by the threshold segmentation corresponding to step S103; Figure 6 A first edge mask obtained by performing edge detection corresponding to step S104; Figure 7 is a second edge mask result obtained by filtering the first edge mask through the first head skin mask in step S105; Figure 11This corresponds to the skin reconstruction result obtained in step S107.

[0078] Step S102 performs pre-processing on the original image to obtain a second image, which may include pre-processing and noise reduction processing.

[0079] The present invention first generates a mask containing the head skin boundary based on the original image. This mask is then used to block the image portion excluding the skin. Based on this mask, isosurface extraction is performed to reconstruct a reliable head skin. This method uses the skin boundary mask to block out most external noise interference from the skin, as well as interference from the eyes, oxygen mask, and imaging artifacts. It also blocks the influence of the complex internal tissue structure of the head, resulting in a reconstructed skin mesh with a single-layer curved surface. Because there is no invalid internal head structure mesh, the skin reconstruction requires less data.

[0080] In summary, the present invention has the characteristics of strong anti-noise ability, strong anti-interference ability of head attachments, high and smooth generated grid precision, and no influence from the internal tissues of the head.

[0081] Combine Figure 2 , Figure 2 This is a flow chart of the medical image reconstruction method provided by the present invention. Figure 2 The medical image reconstruction method in the embodiment includes the following steps:

[0082] S201: Acquire original image;

[0083] S202: Preprocess the original image to obtain a first image;

[0084] S203: Perform noise reduction processing on the first image to obtain a second image;

[0085] S204: Obtaining a first head skin mask that matches the second image by a threshold segmentation method;

[0086] S205: Perform edge detection on the second image to obtain a first edge mask;

[0087] S206: filtering the first edge mask using the first head skin mask, retaining only the portion of the first edge mask corresponding to the head contour, to obtain a filtered second edge mask;

[0088] S207: performing a morphological closing operation and a hole filling operation on the second edge mask to obtain a second head skin contour mask;

[0089] S208: Obtaining a head skin inner contour mask and a head skin outer contour mask according to the second head skin contour mask;

[0090] S209: Perform skin reconstruction based on the second image, the head skin inner contour mask, and the head skin outer contour mask.

[0091] Specifically, both step S204 and step S205 process the second image. In terms of order, step S204 can be performed first and then step S205, or step S205 can be performed first and then step S204. Of course, steps S204 and S205 can also be processed simultaneously.

[0092] Combine Figures 3 to 11 , Figure 3 is the first image obtained by performing preprocessing corresponding to step S202; Figure 4 The second image obtained by the noise reduction process in step S203; Figure 5 A first head skin mask obtained by the threshold segmentation corresponding to step S204; Figure 6 A first edge mask obtained by performing edge detection corresponding to step S205; Figure 7 is a second edge mask result obtained by filtering the first edge mask through the first head skin mask in step S206; Figure 8 The inner contour mask of the head skin and the outer contour mask of the head skin obtained in step S208 are obtained; Figure 11 This corresponds to the skin reconstruction result obtained in step S209.

[0093] In some optional embodiments, step S201: obtaining an original image. The original image here can be an MRI magnetic resonance imaging image or a CT electronic computed tomography image. In this embodiment, only the MRI magnetic resonance imaging image is used as an example.

[0094] Optionally, the preprocessing process of step S202 is to make the resolution of the original image and the voxel value of the original image fall within a specified range to ensure the execution effect of the subsequent algorithm.

[0095] Optionally, step S203 is a noise reduction process, which can reduce the impact of random noise on the subsequent edge detection step. The first image noise reduction process is an optional step, which can be skipped when the noise is weak to speed up the processing.

[0096] Optionally, step S204 is a threshold segmentation process, in which a head contour mask is obtained. The head contour mask provides an approximate position of the head and is combined with the edge detection result to generate an accurate head skin.

[0097] Optionally, step S205 mainly performs edge detection processing on the second image (the image after noise reduction). The edge detection is to locate the accurate position of the head skin.

[0098] Optionally, step S206 and step S207 are to synthesize the threshold segmentation result and the edge detection result and combine them with morphological operations to obtain the head skin contour mask. This step synthesizes the threshold segmentation result and the edge detection result and combines them with morphological operations to obtain the head skin contour mask. Figure 5 The first head skin mask is used as the ROI Figure 6 The first edge mask is selectively shielded to remove edge detection results outside the head, and the first edge mask is filtered, repaired, and the like to obtain an accurate head skin mask.

[0099] The regional mask in the present invention specifically means: if you are only interested in the signal of a certain area in the image, you can generate a matrix of the same size as the original image, set the matrix values ​​outside the area of ​​interest to 0, and set the matrix values ​​within the area of ​​interest to 1. This matrix is ​​called the mask of our area of ​​interest.

[0100] In some optional embodiments, combined with Figure 3 , Figure 3 The first image is obtained after preprocessing the original image. Step S202 preprocesses the original image to obtain the first image, including: using FreeSurfer to increase the resolution of the original image to a preset resolution, setting the voxels of the original image to preset values, and transforming the pixel values ​​of the original image to a range of 0 to 255.

[0101] This preprocessing process ensures that the original image's resolution and voxel values ​​fall within a specified range to ensure the effectiveness of subsequent algorithms. In other words, the purpose of preprocessing is to stabilize the voxel value range of the original image, facilitating subsequent threshold segmentation. A voxel is a volumetric pixel, the basic unit used to represent a display point in three-dimensional space.

[0102] In this embodiment, the MRI image (original image) is sampled to a resolution of 256×256×256, the voxel value size is 1×1×1mm, and the pixel value is converted to the range of 0-255. Of course, the specific numerical values ​​of the resolution, pixel value, and voxel value in this embodiment are only for schematic illustration. Specifically, the mri_convert command in the FreeSurfer software package can be used to implement the preprocessing process. Freesurfer is an open source software for MRI image processing and analysis, which can perform preprocessing of multiple modal data, cortical reconstruction, and mutual registration between volume and surface. The change in resolution has no effect on the subsequent process, but the change in voxel value will affect the subsequent threshold setting.

[0103] In some optional embodiments, referring to Figure 4 , Figure 4The second image is obtained by performing noise reduction processing on the first image. In this embodiment, step S203 performs noise reduction processing on the first image to obtain the second image. The image processing software package SimpleITK can be used to perform noise reduction processing on the first image by bilateral filtering to obtain the second image.

[0104] The noise reduction step can reduce the impact of random noise on the subsequent edge detection step. The first image noise reduction step is an optional step and can be skipped when the noise is weak to speed up the processing.

[0105] It is understandable that bilateral filtering is used for noise reduction. This noise reduction method considers both the spatial and range information of the image, and has the ability to reduce noise while preserving edges. This noise reduction method is used to reduce random noise in the first image and highlight the skin contour, making the subsequent edge detection results more stable and reliable. Specifically, this solution adopts the bilateral filtering implementation method in the SimpleITK software package, with the parameter settings of domainSigma=4.0, rangeSigma=50.0, and numberOfRangeGaussianSamples=100. Of course, the parameter settings here can be set according to actual needs and are not specifically limited.

[0106] In some optional embodiments, referring to Figure 5 , Figure 5 is a first head skin mask result image after threshold segmentation. In step S204, obtaining a first head skin mask that matches the second image by the threshold segmentation method includes:

[0107] Set the binarization threshold to M, set the voxels in the second image that are larger than M to 1, and set the voxels in the second image that are smaller than or equal to M to 0, and obtain a binarization mask of the same size as the second image;

[0108] The first head skin mask is obtained by performing maximum connected domain, morphological closing and hole filling operations on the binary mask.

[0109] Step S204 is to perform threshold segmentation processing on the second image after noise reduction. The head contour mask can be obtained through the threshold segmentation step. In the subsequent process, the head contour mask provides the approximate position of the head and is combined with the edge detection result to generate an accurate head skin mask.

[0110] Specifically, the denoised second image is converted into a binary mask by performing a binarization operation on the denoised second image. Optionally, the binarization threshold is between 10 and 50 (including endpoint values). Optionally, the binarization threshold is 30, that is, voxels in the second image that are greater than the binarization threshold are set to 1, and voxels that are less than or equal to the binarization threshold are set to 0, so that a binary mask equal to the size of the second image can be obtained; the binary mask obtained above is modified by extracting the maximum connected domain, performing morphological closing operations and filling holes to obtain a head contour mask.

[0111] In some optional embodiments, the medical image processing software package ANTsPy is used to perform maximum connected component, morphological closing, and hole filling operations on the binary mask to obtain the first head skin mask. Optionally, when using the ANTsPy software package, the radius parameter of the morphological closing operation is set to 4. Of course, other parameters can also be set, and are not specifically limited here.

[0112] In some optional embodiments, referring to Figure 6 , Figure 6 This is the first edge mask obtained after edge detection on the second image. This step S205 mainly performs edge detection on the second image (the denoised image). The edge detection is to locate the exact position of the head skin. The input is the denoised second image, and the output is the first edge mask.

[0113] Preferably, the canny operator is used to perform edge detection on the second image after noise reduction to obtain an edge mask. In a specific implementation, the ANTsPy software package can be used. The preferred canny operator edge detection parameters are sigma = 1, lower = 0.5, upper = 1. Of course, the parameters here are only for schematic illustration and can also be set to other parameters, which are not specifically limited here.

[0114] In this embodiment, the accurate position of the head skin can be obtained by performing edge detection using the Canny operator.

[0115] In some optional embodiments, referring to Figure 7 , Figure 7 is the second edge mask image, corresponding to step S206: filtering the first edge mask through the first head skin mask, retaining only the portion of the first edge mask corresponding to the head contour, to obtain a filtered second edge mask;

[0116] This step combines the threshold segmentation results and edge detection results and combines them with morphological operations to obtain the head skin contour mask. Figure 5 The first head skin mask is used as the ROI Figure 6The first edge mask is selectively masked to remove edge detection results outside the head, resulting in a second edge mask. This step primarily post-processes the edge detection results (the first edge mask). As will be appreciated, edge detection can easily generate a large number of non-skin edge detection results inside and outside the head in an image. Therefore, additional auxiliary information is needed to filter and patch the first edge mask to obtain an accurate head skin mask.

[0117] In some optional embodiments, step S207 performs a morphological closing and hole-filling operation on the second edge mask to obtain a second head skin contour mask. Specifically, the morphological closing and hole-filling operations are performed on the second edge mask using medical image processing software ANTsPy to obtain the second head skin contour mask. Optionally, the radius parameter of the morphological closing operation is set to 5. Of course, the radius parameter of the morphological closing operation can also be set to other parameters, which are not specifically limited here.

[0118] The first edge mask is subjected to filtering, patching and other operations to obtain a complete and accurate head skin mask (second head skin contour mask).

[0119] In some optional embodiments, referring to Figure 8 , Figure 8 Schematic diagram of a head skin inner contour mask and a head skin outer contour mask obtained based on the second head skin contour mask. In this embodiment, step S208 of obtaining the head skin inner contour mask and the head skin outer contour mask based on the second head skin contour mask includes: performing an erosion operation on the second head skin contour mask to obtain a head skin inner contour mask, and performing an expansion operation on the second head skin contour mask to obtain a head skin outer contour mask.

[0120] Alternatively, based on the obtained accurate skin contour mask (the second head skin contour mask), an erosion operation with a radius of 2 is performed to obtain an inner head skin contour mask; and a dilation operation with a radius of 2 is performed to obtain an outer head skin contour mask. The radii of the erosion and dilation operations can be set based on actual needs. The erosion operation reduces the second head skin contour mask, while the dilation operation expands it. The combination of these two operations results in a shell with its center plane at the skin.

[0121] In some optional embodiments, referring to Figure 9 and Figure 10 , Figure 9 is a denoised image that only retains the skin contour, Figure 10 yes Figure 9 A local enlarged view of the boundary. Figure 11is the skin reconstruction result. Step S209 of this embodiment performs skin reconstruction based on the second image, the head skin inner contour mask, and the head skin outer contour mask, including:

[0122] The values ​​of the part of the second image outside the head skin outer contour mask are set to 0, and the values ​​of the part of the second image inside the head skin inner contour mask are set to 255, so as to obtain a denoised image with only the skin contour retained. The denoised image with only the skin contour retained can be referred to Figure 9 and Figure 10 ,from Figure 10 The denoised image result with preserved skin contour can be clearly seen in the figure.

[0123] The head skin is extracted from the second image which only retains the skin contour using an isosurface extraction algorithm to obtain an isosurface set;

[0124] A single-layer mesh surface is extracted from the isosurface set using the maximum continuous surface extraction method, which is the mesh representation of the head skin.

[0125] Step S209 is skin reconstruction processing. The input of this step is the second image after noise reduction and the head skin inner contour mask and the head skin outer contour mask obtained in step S5. The output of this step is a grid representation of the head skin.

[0126] In some optional embodiments, extracting the head skin from the second image retaining only the skin contour using an isosurface extraction algorithm to obtain an isosurface set includes executing the isosurface extraction algorithm using the PyVista software package to extract the isosurface set. Specifically, the isosurface extraction algorithm can be executed using the PyVista software package with parameters such as a number of extracted isosurfaces of 4 and a voxel value range of 30 to 60. This step obtains the extracted isosurface set. The number of extracted isosurfaces can be greater than 1 and is not specifically limited here.

[0127] Comparative Example, combined with the figures respectively show the processing results of directly using the isosurface extraction algorithm in the related art (refer to Figure 12 and Figure 13 ) and the treatment results of the method of the present invention (refer to Figure 14 and Figure 15 ), Figure 12 It is the skin reconstruction result directly using the isosurface extraction algorithm. Figure 13 It is a cross-sectional view of the skin mesh directly using the isosurface extraction algorithm. Figure 14 is the result of skin reconstruction using the method of the present invention, Figure 15 The skin mesh results of the method of the present invention are compared with the processing results of the two methods. It can be found that the method of the present invention has the following advantages:

[0128] The method proposed in this paper has strong anti-noise ability. Figure 12 As shown in the figure, before the skin contour mask is used to limit the range, the skin reconstruction results generated by the mean surface extraction algorithm in the related art have some abnormal structures caused by image background noise, artifacts, etc., such as Figure 12 There are abnormal columnar structures on both sides of the middle head, but the reconstruction results of the present invention do not have this problem;

[0129] The skin mesh generated by the present invention is smoother and closer to the real skin position. Figure 12 and Figure 14 As shown, at the right ear and left eye positions, the skin reconstruction results of the present invention do not have the spur structures and unevenness problems generated by the original isosurface extraction method in the related art, and the skin surface generated by the present invention is better;

[0130] The skin mesh generated by the present invention does not have unnecessary internal tissue structure of the head, and the generated mesh is a single-layer mesh. Figure 13 and Figure 15 As shown, the skin mesh generated by the original isosurface extraction algorithm in related art contains multiple layers and isolated mesh cells, while the mesh generated by the present invention is a single, continuous surface. This is because the skin contour mask used in the present invention completely blocks the influence of internal head tissue. Calculating and representing this single, continuous surface requires less computation and data, resulting in more accurate and streamlined processing results for applications based on the skin reconstruction results.

[0131] It can be seen from the above embodiments that the medical image reconstruction method provided by the present invention achieves at least the following beneficial effects:

[0132] The medical image reconstruction method provided by the present invention has the characteristics of strong anti-noise ability, strong anti-interference ability of head attachments, high-precision and smooth generated grid, and no influence from the internal tissues of the head.

[0133] The medical image reconstruction method provided by the present invention first generates a regional mask containing the head skin boundary based on the original MRI image, then uses the regional mask to shield the image part except the skin, and then performs isosurface extraction on this basis to reconstruct a reliable head skin.

[0134] The medical image reconstruction method provided by the present invention uses a skin boundary area mask to shield most of the noise interference outside the skin and interference from the eyes, oxygen mask, and imaging artifacts. At the same time, this area mask also shields the influence of the complex tissue structure inside the head, making the reconstructed skin mesh a single-layer surface. Since there is no invalid internal head structure mesh, the amount of data required to represent the skin reconstruction result is less.

[0135] Although some specific embodiments of the present invention have been described in detail by way of examples, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It should be understood by those skilled in the art that modifications may be made to the above embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A medical image reconstruction method, characterized in that: include: Get the original image; Pre-processing the original image to obtain a second image; Obtaining a first head skin mask that matches the second image by a threshold segmentation method, including: setting a binarization threshold to M, setting voxels in the second image that are larger than M to 1, and setting voxels in the second image that are smaller than or equal to M to 0, to obtain a binary mask of the same size as the second image; performing a maximum connected domain, a morphological closing operation, and a hole filling operation on the binary mask to obtain the first head skin mask; performing edge detection on the second image to obtain a first edge mask; filtering the first edge mask using the first head skin mask, retaining only a portion of the first edge mask corresponding to the head contour, to obtain a filtered second edge mask; Performing a morphological closing operation and a hole filling operation on the second edge mask to obtain a second head skin contour mask; Performing skin reconstruction according to the second image and the head skin contour mask, comprising: Obtaining a head skin inner contour mask and a head skin outer contour mask according to the second head skin contour mask; Skin reconstruction is performed based on the second image, the head skin inner contour mask, and the head skin outer contour mask, including: setting the values ​​of the portion of the second image located outside the head skin outer contour mask to 0, and setting the values ​​of the portion of the second image located within the head skin inner contour mask to 255, to obtain a denoised image that only retains the skin contour; using an isosurface extraction algorithm to extract the head skin from the second image that only retains the skin contour to obtain an isosurface set; and extracting a single-layer mesh surface from the isosurface set using a maximum continuous surface extraction method, which is a mesh representation of the head skin.

2. The medical image reconstruction method according to claim 1, wherein: Pre-processing the original image to obtain a second image includes: Preprocessing the original image to obtain a first image; Perform noise reduction processing on the first image to obtain the second image.

3. The medical image reconstruction method according to claim 2, wherein: Performing noise reduction processing on the first image to obtain a second image includes: using an automatic image processing tool to perform noise reduction processing on the first image by bilateral filtering to obtain the second image.

4. The medical image reconstruction method according to claim 1, wherein: Performing a maximum connected domain, a morphological closing operation, and a hole filling operation on the binary mask to obtain the first head skin mask includes: A medical image processing software package is used to perform maximum connected domain, morphological closing operation and hole filling operation on the binary mask to obtain the first head skin mask.

5. The medical image reconstruction method according to claim 1, wherein: Performing edge detection on the second image to obtain a first edge mask includes: A medical image processing software package is used to perform edge detection on the second image using a Canny operator to obtain the first edge mask.

6. The medical image reconstruction method according to claim 1, wherein: Performing a morphological closing operation and a hole filling operation on the second edge mask to obtain a second head skin contour mask, including: A medical image processing software package is used to perform morphological closing and hole filling operations on the second edge mask to obtain the second head skin contour mask.

7. The medical image reconstruction method according to claim 1, wherein: Obtaining a head skin inner contour mask and a head skin outer contour mask according to the second head skin contour mask includes: An erosion operation is performed on the second head skin contour mask to obtain a head skin inner contour mask, and an expansion operation is performed on the second head skin contour mask to obtain a head skin outer contour mask.

8. The medical image reconstruction method according to claim 1, wherein: Extracting the head skin from the second image retaining only the skin contour using an isosurface extraction algorithm to obtain an isosurface set, including: The isosurface extraction algorithm is executed using a 3D visualization software package to extract the obtained isosurface set.

9. The medical image reconstruction method according to claim 1, wherein: Preprocessing the original image to obtain a first image includes: The resolution of the original image is adjusted to a preset resolution using magnetic resonance data processing software, the voxels of the original image are set to preset values, and the pixel values ​​of the original image are converted to a range of 0 to 255.

Images