CN110533573B - MRI image parallel processing method and processing system based on JAVA language - Google Patents
MRI image parallel processing method and processing system based on JAVA language Download PDFInfo
- Publication number
- CN110533573B CN110533573B CN201910829802.6A CN201910829802A CN110533573B CN 110533573 B CN110533573 B CN 110533573B CN 201910829802 A CN201910829802 A CN 201910829802A CN 110533573 B CN110533573 B CN 110533573B
- Authority
- CN
- China
- Prior art keywords
- image
- mri image
- processing
- data
- processing method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
Abstract
The invention provides a JAVA-language-based MRI image parallel processing method and a JAVA-language-based MRI image parallel processing system, which are characterized in that the processing method comprises the following steps: starting; uploading a DICOM data compression package of the MRI image of the patient; decompressing the compression package to obtain MRI image brain structure file data; extracting brain structure file data of an MRI image, and analyzing and processing; collecting analysis processing results, correcting standard template data in stages according to a large amount of statistical data, and giving statistical analysis results; and (5) ending. The invention has the advantages that: according to the processing method and the processing system, the MRI image processing method of parallel processing of a plurality of machines is rewritten based on an open source image processing function library, and a large amount of MRI data is learned by the machines to form a Chinese MRI image processing model, so that the purposes of improving the processing speed, efficiency and precision are finally achieved, and the high-precision MRI image processing for Chinese is achieved.
Description
Technical Field
The invention belongs to the technical field of MRI image processing, and particularly relates to an MRI image parallel processing method and an MRI image parallel processing system based on JAVA language.
Background
The existing method for processing the MRI image is mostly based on a MATLAB platform under a single machine, most of image processing templates are European and American and southeast Asia race targets, and the image processing capability and the calculation model are only used in the research field.
However, the prior art processing method has the problems of low processing efficiency and low MRI image processing precision, and has no high-precision MRI image processing method aiming at Chinese people.
Disclosure of Invention
In order to overcome the problems in the prior art, the invention provides an MRI image parallel processing method and an MRI image parallel processing system based on JAVA language, which are based on an open source image processing function library, re-write a plurality of MRI image processing methods for parallel processing of machines, and form a Chinese MRI image processing model by learning a large amount of MRI data through the machines, so that the purposes of improving the processing speed, efficiency and precision are finally realized, and the high-precision MRI image processing for Chinese is realized.
The invention provides an MRI image parallel processing method based on JAVA language, which comprises the following steps:
(1) Starting;
(2) Uploading a DICOM data compression package of the MRI image of the patient;
(3) Decompressing the compression package to obtain MRI image brain structure file data, and specifically comprising the following 6 steps:
1) Non-uniform intensity normalization
In order to facilitate the unified processing of a computer, machine features need to be eliminated, the image intensity is normalized to be in the range of (0, 255), and a gray scale normalization method based on the image content features is used;
obtaining, from a priori knowledge, a marker range image represented by a tensor:
X∈R p×q×r×m alpha < X < beta, where alpha, beta are the range boundaries of the machine feature;
directly setting the calculated values as 0 and 255 after the calculation is finished, mapping the partial ranges in alpha and beta respectively corresponding to the content characteristics to (0, 255) and obtaining the content characteristics in X 1 ，X 2 Normalizing … Xn respectively;
the contrast of the marker range image can then be expressed as
Wherein X is i，j，k，l Intensity, X, of a first feature representing a position in three-dimensional space (i, j, k) i，j，k，l =1 represents red, X i，j，k，l =2 represents green, X i，j，k，l =3 represents blue, X i，j，k，l =4 represents transparency;
Using m=1, the above equation becomes
2) Unified precision of three directions using convolution tri-interpolation algorithm
f(i+u，j+v)＝[A]×[B]×[C]
[A]＝[S(u+1) S(u+0) S(u-1) S(u-2)]
Interpolation variables in three directions of X, Y and Z: x=abs (1- Φx), y=abs (1- Φy), z=abs (1- Φz);
wherein, phix, phiy and phiz respectively represent the precision in three directions;
3) Performing coordinate system space transformation
The spatial location of Patient Position is determined in DICOM standard by two fields, image Position (Patient) - (0020,0032) and Image Orientation (Patient) - (0020,0037); wherein Image Position (event) represents the (x, y, z) coordinates of the upper left corner of the Image in a spatial coordinate system in millimeters; image Orientation (Patent) represents the coordinate angle between the image coordinates and the corresponding coordinate of the anatomical coordinate systemA cosine value; to obtain displacements DeltaX, deltaY, deltaZ corresponding to a voxel position point (X, Y, Z) in the hypothetical dicom image, a (X i ,y i ,z i ) For the ith mark point coordinate, N is the number of mark points, m is 2-order spline, if there is
Δy (x, y, z), Δz (x, y, z) are calculated using the methods described above;
4) Smoothing using mean-shift algorithm
N sample points x in a given d-dimensional space Rd i I=1, 2,3, the terms, n, the vector at x is in the basic form:
wherein S is k Representing data points of the dataset having a distance x less than the sphere radius h, i.e
S h (x)＝{y：(y-x i )T(y-x i )＜h 2 }
The drift vector is obtained through calculation, and the position of the center x of the updated sphere is as follows:
x：＝x+M h
each iteration is carried out, the circle center is always moved towards the direction with the maximum data set density, the average position of the points in the circle is found to be used as a new circle center position, and the calculation formula of the drift vector is obtained by adding Gaussian weights:
calculating center coordinates updated each time as follows:
smoothing with Gaussian kernel
5) Region growing segmentation algorithm segments gray matter and white matter
Let R denote the entire 3d image matrix, then the segmentation can be seen as dividing the region R into n sub-regions R 1 ,R 2 ,......R n And the following conditions are required to be satisfied:
a:U(R i )＝R；
b:R i is a connected region, i=1, 2,3,..;
c:R i ∩R j =empty set, for any i, j; all have i+.j;
d:P(R i ) =cure, for i=1, 2,..;
e:R(P i U R j )＝False,i≠j；
the above-mentioned segmentation must be performed in units of voxels, otherwise, continuity of the segmented object cannot be ensured, resulting in incomplete segmentation; in order to improve the efficiency of the segmentation process, the processing process can be switched to cuda, and the result is taken out after the processing is finished;
6) Registering segmentation results to standard segmented brains
Taking the axial vector data f (x, y), registering it to the standard axial vector data f (x ', y'), transforming the model into:
wherein α is a scaling parameter; θ is a rotation angle parameter, and (Δx, Δy) is a translation parameter;
the image wavelet transform function is:
set h= { H n The low pass filter is }, g= { G n And the frequency is a high-pass filter, and the decomposition number is calculated by a Mallat algorithm, wherein the frequency is as follows
Registering the sampled image to the standard image by three transformations, telescoping, translating, rotating, wherein the telescoping transformation computes:
translational transformation calculation: let p=k-2 m, q=l-2 n, have
Rotation transformation calculation: let p= -kcosθ+lsinθ, q= -ksinθ+lcosθ, there is
c M+1 (2m cosθ+2n sinθ，-2m sinθ+2n cosθ)
The wavelet low pass filtering becomes:
after registration transformation, the detailed brain structure file data can be counted according to the standard;
(4) Extracting MRI image data and analyzing and processing brain structure file data;
(5) Collecting analysis processing results, correcting standard template data in stages according to a large amount of statistical data, and giving statistical analysis results;
(6) And (5) ending.
Preferably, the processing procedure of the region growing segmentation algorithm in the step 5) for segmenting gray matter and white matter can be transferred to cuda, and the result is taken out after the processing is completed.
The MRI image parallel processing system based on JAVA language adopts the method for processing MRI images based on JAVA language through JDK, combines with a server CPU and a memory to establish a plurality of subprograms, and simultaneously processes a plurality of image data.
The invention has the advantages that:
(1) the parallel processing method of the MRI image based on the JAVA language can realize the accurate analysis processing of the MRI image by an open source technology;
(2) according to the JAVA language-based MRI image parallel processing system, a networking technology is adopted to process MRI images in parallel by a plurality of machines, so that the processing speed is remarkably improved;
(3) according to the JAVA language-based MRI image parallel processing method and system, an open source image processing function library is used as a basis, a plurality of machines are rewritten to process MRI images in parallel, a large amount of MRI data is learned by the machines to form a Chinese MRI image processing model, the purposes of improving processing speed, efficiency and precision are finally achieved, and high-precision MRI image processing for Chinese is achieved.
Drawings
Fig. 1 is a flow chart diagram of an MRI image parallel processing method based on JAVA language according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
Referring to fig. 1, a JAVA language-based MRI image parallel processing method includes the steps of:
(1) Starting;
(2) Uploading a DICOM data compression package of the MRI image of the patient;
(3) Decompressing and compressing the package to obtain MRI image brain structure file data, which specifically comprises the following 6 steps:
1) Non-uniform intensity normalization
In order to facilitate the unified processing of a computer, machine features need to be eliminated, the image intensity is normalized to be in the range of (0, 255), and a gray scale normalization method based on the image content features is used;
obtaining, from a priori knowledge, a marker range image represented by a tensor:
X∈R p×q×r×m a < X < beta, where alpha, beta are the range boundaries of the machine feature;
directly setting the calculated values as 0 and 255 after the calculation is finished, mapping the partial ranges in alpha and beta respectively corresponding to the content characteristics to (0, 255) and obtaining the content characteristics in X 1 ，X 2 Normalizing … Xn respectively;
the contrast of the marker range image can then be expressed as:
wherein X is i，j，k，l Intensity, X, of a first feature representing a position in three-dimensional space (i, j, k) i，j，k，l =1 represents red, X i，j，k，l =2 represents green, X i，j，k，l =3 represents blue, X i，j，k，l =4 represents transparency;
using m=1, the above equation becomes:
2) Unified precision of three directions using convolution tri-interpolation algorithm
f(i+u，j+v)＝[A]×[B]×[C]
[A]＝[S(u+1 )S(u+0 )S(u-1) S(u-2)
Interpolation variables in three directions of X, Y and Z: x=abs (1- Φx), y=abs (1- Φy), z=abs (1- Φz);
wherein, phix, phiy and phiz respectively represent the precision in three directions;
3) Performing coordinate system space transformation
The spatial location of Patient Position is determined in DICOM standard by two fields, image Position (Patient) - (0020,0032) and Image Orientation (Patient) - (0020,0037);wherein Image Position (event) represents the (x, y, z) coordinates of the upper left corner of the Image in a spatial coordinate system in millimeters; image Orientation (Patent) represents the cosine value of the coordinate included angle between the image coordinates and the corresponding coordinate of the anatomical coordinate system; to obtain displacements DeltaX, deltaY, deltaZ corresponding to a voxel position point (X, Y, Z) in the hypothetical dicom image, a (X i ,y i ,z i ) For the ith mark point coordinate, N is the number of mark points, m is a 2-order spline, and there are:
θy (x, y, z), Δz (x, y, z) are calculated using the method described above;
4) Smoothing using mean-shift algorithm
N sample points x in a given d-dimensional space Rd i I=1, 2,3, the terms, n, the vector at x is in the basic form:
wherein S is k Representing data points of the dataset having a distance x less than the sphere radius h, i.e
S h (x)＝{y：(y-x i ) T (y-x i )＜h 2 }
The drift vector is obtained through calculation, and the position of the center x of the updated sphere is as follows:
x：＝x+M h
each iteration is carried out, the circle center is always moved towards the direction with the maximum data set density, the average position of the points in the circle is found to be used as a new circle center position, and the calculation formula of the drift vector is obtained by adding Gaussian weights:
calculating center coordinates updated each time as follows:
smoothing using gaussian kernels, there are:
5) Region growing segmentation algorithm segments gray matter and white matter
Let R denote the entire 3d image matrix, then the segmentation can be seen as dividing the region R into n sub-regions R 1 ,R 2 ,......R n And the following conditions are required to be satisfied:
a:U(R i )＝R；
b:R i is a connected region, i=1, 2,3,..;
c:R i ∩R j =empty set, for any i, j; all have i+.j;
d:P(R i ) =cure, for i=1, 2,..;
e:R(P i U R j )＝False,i≠j；
the above-mentioned segmentation must be performed in units of voxels, otherwise, continuity of the segmented object cannot be ensured, resulting in incomplete segmentation; in order to improve the efficiency of the segmentation process, the processing process can be switched to cuda, and the result is taken out after the processing is finished;
6) Registering segmentation results to standard segmented brains
Taking the axial vector data f (x, y), registering it to the standard axial vector data f (x ', y'), transforming the model into:
wherein α is a scaling parameter; θ is a rotation angle parameter, and (Δx, Δy) is a translation parameter;
the image wavelet transform function is:
set h= { H n The low pass filter is }, g= { G n And (3) calculating the decomposition number by a Mallat algorithm if the filter is a high-pass filter, wherein the decomposition number is as follows:
registering the sampled image to the standard image by three transformations, telescoping, translating, rotating, wherein the telescoping transformation computes:
translational transformation calculation:
let p=k-2 m, q=l-2 n, have:
rotation transformation calculation:
let p=kcosθ+lsinθ, q= -ksinθ+lcosθ, there are:
c M+1 (2m cosθ+2n sinθ，-2m sinθ+2n cosθ)
the wavelet low pass filtering becomes:
after registration transformation, detailed MRI image brain structure file data can be counted according to the standard;
(4) Extracting brain structure file data of an MRI image, and analyzing and processing;
(5) Collecting analysis processing results, correcting standard template data in stages according to a large amount of statistical data, and giving statistical analysis results;
(6) And (5) ending.
The preferred embodiments of the invention disclosed above are intended only to assist in the explanation of the invention. The preferred embodiments are not intended to be exhaustive or to limit the invention to the precise form disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and the practical application, to thereby enable others skilled in the art to best understand and utilize the invention. The invention is limited only by the claims and the full scope and equivalents thereof.
Claims (2)
1. A JAVA language-based MRI image parallel processing method, characterized in that the processing method comprises the steps of:
(1) Starting;
(2) Uploading a DICOM data compression package of the MRI image of the patient;
(3) Decompressing the compressed package to obtain MRI image brain structure file data, and specifically comprising the following 6 steps:
1) Non-uniform intensity normalization
In order to facilitate the unified processing of a computer, machine features need to be eliminated, the image intensity is normalized to be in the range of (0, 255), and a gray scale normalization method based on the image content features is used;
obtaining, from a priori knowledge, a marker range image represented by a tensor:
directly setting the calculated values as 0 and 255 after the calculation is finished, mapping the partial ranges in alpha and beta respectively corresponding to the content characteristics to (0, 255) and obtaining the content characteristics in X 1 ，X 2 Normalizing … Xn respectively;
the contrast of the marker range image can then be expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,
intensity of the first feature representing the position of the three-dimensional space (i, j, k), +.>
=1 represents a red color and,
=2 represents green, ++>
=3 represents blue, ++>
=4 represents transparency;
using m=1, the above equation becomes:
2) The precision unification of the three directions is performed by using a convolution tri-interpolation algorithm,
interpolation variables in three directions of X, Y and Z: x=abs (1- Φx), y=abs (1- Φy), z=abs (1- Φz);
wherein, phix, phiy and phiz respectively represent the precision in three directions;
3) Performing coordinate system space transformation
The spatial location of Patient Position is determined in DICOM standard by two fields, image Position (Patient) - (0020,0032) and Image Orientation (Patient) - (0020,0037); wherein Image Position (event) represents the (x, y, z) coordinates of the upper left corner of the Image in a spatial coordinate system in millimeters; image Orientation (Patent) represents the cosine value of the coordinate included angle between the image coordinates and the corresponding coordinate of the anatomical coordinate system; in order to obtain displacements DeltaX, deltaY, deltaZ corresponding to a voxel position point (X, Y, Z) in the hypothetical dicom image, a value of (,
) For the ith mark point coordinate, N is the number of mark points, m is a 2-order spline, and there are:
1) Smoothing using mean-shift algorithm
N sample points x in a given d-dimensional space Rd i I=1, 2,3, the terms, n, the vector at x is in the basic form:
wherein S is k Representing data points of the dataset having a distance x less than the sphere radius h, i.e
The drift vector is obtained through calculation, and the position of the center x of the updated sphere is as follows:
each iteration is carried out, the circle center is always moved towards the direction with the maximum data set density, the average position of the points in the circle is found to be used as a new circle center position, and the calculation formula of the drift vector is obtained by adding Gaussian weights:
calculating center coordinates updated each time as follows:
smoothing using gaussian kernels, there are:
2) Region growing segmentation algorithm segments gray matter and white matter
Let R denote the entire 3d image matrix, then the segmentation can be seen as dividing the region R into n sub-regions R 1 ,R 2 ,......R n And the following conditions are required to be satisfied:
a: U(R i ) = R;
b: R i is a connected region, i=1, 2,3,..;
c: R i ∩ R j =empty set, for any i, j; all have i+.j;
d: P(R i ) =cure, for i=1, 2,..;
e: R(P i U R j ) = False, i≠j；
the above-mentioned segmentation must be performed in units of voxels, otherwise, continuity of the segmented object cannot be ensured, resulting in incomplete segmentation; in order to improve the efficiency of the segmentation process, the processing process can be switched to cuda, and the result is taken out after the processing is finished;
the processing process of dividing gray matter and white matter by the region growing dividing algorithm can be transferred to cuda, and the result is taken out after the processing is finished;
3) Registering segmentation results to standard segmented brains
Taking the axial vector data f (x, y), registering it to the standard axial vector data f (x ', y'), transforming the model into:
wherein the method comprises the steps of
Is a scaling parameter; />
For the rotation angle parameter, (-)>
) Is a translation parameter;
the image wavelet transform function is:
setting up
Is a low-pass filter, ">
As a high-pass filter, the number of decompositions is calculated by the Mallat algorithm, with:
registering the sampled image to the standard image by three transformations, telescoping, translating, rotating, wherein the telescoping transformation computes:
translational transformation calculation:
let p=k-2 m, q=l-2 n, have:
rotation transformation calculation:
let p=
,q=/>
The method comprises the following steps:
the wavelet low pass filtering becomes:
after registration transformation, detailed MRI image brain structure file data can be counted according to the standard;
extracting brain structure file data of an MRI image, and analyzing and processing;
collecting analysis processing results, correcting standard template data in stages according to a large amount of statistical data, and giving statistical analysis results;
and (5) ending.
2. The JAVA-language-based MRI image parallel processing system for realizing the JAVA-language-based MRI image parallel processing method as claimed in claim 1, wherein the JAVA-language-based MRI image parallel processing method is adopted to build a plurality of subroutines through JDK in combination with a server CPU and a memory, and simultaneously process a plurality of image materials.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910829802.6A CN110533573B (en) | 2019-09-04 | 2019-09-04 | MRI image parallel processing method and processing system based on JAVA language |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910829802.6A CN110533573B (en) | 2019-09-04 | 2019-09-04 | MRI image parallel processing method and processing system based on JAVA language |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110533573A CN110533573A (en) | 2019-12-03 |
| CN110533573B true CN110533573B (en) | 2023-07-14 |
Family
ID=68666699
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910829802.6A Active CN110533573B (en) | 2019-09-04 | 2019-09-04 | MRI image parallel processing method and processing system based on JAVA language |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110533573B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111784664B (en) * | 2020-06-30 | 2021-07-20 | 广州柏视医疗科技有限公司 | Method for generating distribution map of tumor lymph nodes |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103942780A (en) * | 2014-03-27 | 2014-07-23 | 北京工业大学 | Fuzzy-connectedness-algorithm-based segmentation method of thalamus and substructures of thalamus |
| CN104376569A (en) * | 2014-11-28 | 2015-02-25 | 成都影泰科技有限公司 | Method for processing DICOM (digital imaging and communications in medicine) medical images on basis of versions |
| CN107507212A (en) * | 2017-08-18 | 2017-12-22 | 中国科学院深圳先进技术研究院 | Digital brain method for visualizing, device, computing device and storage medium |
| CN108122263A (en) * | 2017-04-28 | 2018-06-05 | 上海联影医疗科技有限公司 | Image re-construction system and method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EA024855B1 (en) * | 2012-07-10 | 2016-10-31 | Закрытое Акционерное Общество "Импульс" | Method for obtaining subtraction angiographic image |
-
2019
- 2019-09-04 CN CN201910829802.6A patent/CN110533573B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103942780A (en) * | 2014-03-27 | 2014-07-23 | 北京工业大学 | Fuzzy-connectedness-algorithm-based segmentation method of thalamus and substructures of thalamus |
| CN104376569A (en) * | 2014-11-28 | 2015-02-25 | 成都影泰科技有限公司 | Method for processing DICOM (digital imaging and communications in medicine) medical images on basis of versions |
| CN108122263A (en) * | 2017-04-28 | 2018-06-05 | 上海联影医疗科技有限公司 | Image re-construction system and method |
| CN107507212A (en) * | 2017-08-18 | 2017-12-22 | 中国科学院深圳先进技术研究院 | Digital brain method for visualizing, device, computing device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110533573A (en) | 2019-12-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tong et al. | Improved U-NET network for pulmonary nodules segmentation | |
| Sandor et al. | Surface-based labeling of cortical anatomy using a deformable atlas | |
| CN112950651B (en) | Automatic delineation method of mediastinal lymph drainage area based on deep learning network | |
| Ruppert et al. | A new symmetry-based method for mid-sagittal plane extraction in neuroimages | |
| CN109300113B (en) | Pulmonary nodule auxiliary detection system and method based on improved convex hull method | |
| DE102005003605A1 (en) | Previous knowledge, level representation and visual grouping | |
| US20110064289A1 (en) | Systems and Methods for Multilevel Nodule Attachment Classification in 3D CT Lung Images | |
| Abdel-Nasser et al. | Temporal mammogram image registration using optimized curvilinear coordinates | |
| Maitra et al. | A novel edge detection algorithm for digital mammogram | |
| CN110533573B (en) | MRI image parallel processing method and processing system based on JAVA language | |
| CN113706514B (en) | Focus positioning method, device, equipment and storage medium based on template image | |
| Afzali et al. | Inter-patient modelling of 2D lung variations from chest X-ray imaging via Fourier descriptors | |
| Hsu | A hybrid approach for brain image registration with local constraints | |
| Gallea et al. | Three-dimensional fuzzy kernel regression framework for registration of medical volume data | |
| Zhou et al. | Automatic segmentation of liver from CT scans with CCP–TSPM algorithm | |
| Awate et al. | Kernel methods for riemannian analysis of robust descriptors of the cerebral cortex | |
| Qin et al. | Joint-saliency structure adaptive kernel regression with adaptive-scale kernels for deformable registration of challenging images | |
| Shen et al. | Fuzzy clustering based applications to medical image segmentation | |
| Undeman et al. | Fully automatic segmentation of MRI brain images using probabilistic anisotropic diffusion and multi-scale watersheds | |
| CN111583286B (en) | Abdomen MRI (magnetic resonance imaging) image contour extraction method based on Flow-XDoG operator | |
| Staib et al. | An integrated approach for locating neuroanatomical structure from MRI | |
| Petrou | Image registration: An overview | |
| Wei et al. | An algorithm for segmentation of lung ROI by mean-shift clustering combined with multi-scale HESSIAN matrix dot filtering | |
| CN110363738B (en) | Retina image registration method and device with affine invariance | |
| Macho et al. | Segmenting Teeth from Volumetric CT Data with a Hierarchical CNN-based Approach. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |