CN105205810A - Distance-field-fusion-based hippocampus segmentation method of MR image - Google Patents
Distance-field-fusion-based hippocampus segmentation method of MR image Download PDFInfo
- Publication number
- CN105205810A CN105205810A CN201510539859.4A CN201510539859A CN105205810A CN 105205810 A CN105205810 A CN 105205810A CN 201510539859 A CN201510539859 A CN 201510539859A CN 105205810 A CN105205810 A CN 105205810A
- Authority
- CN
- China
- Prior art keywords
- image
- point
- training set
- dictionary
- distance field
- 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.)
- Granted
Links
Classifications
-
- 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]
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
Landscapes
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention relates to a distance-field-fusion-based hippocampus segmentation method of an MR image. The method comprises: (1), normalization processing is carried out on an initial to-be-segmented MR test image and a skull and a biased field are removed, thereby obtaining an MR test image after normalization processing; (2), registering of an MR training set image and a training set label image to the MR test image is carried out; (3), distance conversion is carried out on the registered label image to obtain a distance field DF; (4), for a point X in the MR test image, an image block X<MR> is taken and is converted into a column vector; (5) a searching window is defined and an MR dictionary and a DF dictionary are selected; (6), the MR dictionary is used for expressing an MR test sample locally and linearly and a dictionary weight coefficient vector is solved; (7), a DF prediction image block vector of the test sample is obtained and is converted into an image block X <DF>; (8), the steps from (4) to (7) are repeated to obtain a DF value of each point; and (9), a label corresponding to each point in the test image is obtained. According to the invention, a target can be segmented accurately in an MR image.
Description
Technical field
The present invention relates to medical image analysis technical field, be specifically related to a kind of MR image hippocampus dividing method merged based on distance field.
Background technology
Hippocampus belongs to the grey matter structure of midbrain, and hippocampus atrophy is a kind of pathological manifestations of the mental illnesses such as alzheimer disease, schizophrenia, depression, can make diagnosis clinically by the volume and form analyzing hippocampus to these mental illnesses.MRI can provide high resolving power anatomical structure, clear contrast and image sequence widely, and this makes MRI technology very popular in clinical diagnosis.Doctor by the brain MR image of patient, can analyze all brain structures of patient, thus makes diagnosis to disease.Splitting the hippocampus of MR brain image, is the measurement of hippocampus volume size, the prerequisite of morphological analysis.Because the manual segmentation of hippocampus is very consuming time, uninteresting, and the segmentation result of different doctor also can difference to some extent, and the auto Segmentation of hippocampus seems particularly important in clinical practice.
Because the volume of hippocampus in MR image is very little, complex-shaped, obscurity boundary, partial volume effect obvious, the auto Segmentation of hippocampus is very challenging.
Up to now, MR image hippocampus dividing method mainly contains following two classes:
The first kind is the method for based upon activities skeleton pattern (ACM).These class methods develop to curve (curved surface) according to gradation of image demographic information or image gradient, and final curves (curved surface) converge on object boundary place.ACM method is more responsive to picture noise.
Equations of The Second Kind is the method based on multichannel chromatogram.These class methods to test pattern, to the training set labeled graph picture after deformation, merging atlas registration, obtaining final segmentation result by realizing label someway.Label merges the shape priors not using target to be split, and segmentation precision limits to some extent.
For the deficiencies in the prior art, the inventive method proposes a kind of MR image hippocampus dividing method merged based on distance field.
Summary of the invention
The inventive method provides a kind of MR image hippocampus dividing method merged based on distance field, and the inventive method is successfully applied in the hippocampus segmentation of MR brain image, can split the hippocampus in MR image exactly.
Above-mentioned purpose of the present invention is realized by following technological means.
Based on the MR image hippocampus dividing method that distance field merges, the method is based on two kinds of hypothesis:
I, MR image block and DF image block are positioned on two non-linearity manifolds, and neighbour's sample that any one MR image block can be flowed in the local space of shape by its place is linearly expressed;
II, under local constraint, the mapping that MR stream shape flows shape to DF is similar to a differomorphism mapping;
The MR image hippocampus dividing method that should merge based on distance field carries out as follows:
(1) to MR test pattern normalized initially to be split, remove skull and biased field, obtain the MR test pattern after normalized;
(2) the MR training set image in pre-prepd training set and the training set labeled graph picture corresponding with MR training set image are registrated to MR test pattern, make MR training set image, training set labeled graph picture aligns with MR test pattern on locus;
(3) range conversion is carried out to the training set labeled graph picture after step (2) registration, obtain the training set distance field DF of training set labeled graph picture, point
the value of corresponding training set distance field DF is:
……(1);
Wherein
crepresent the border of segmentation object,
brepresent distance
xnearest point, and
,
represent point
xand point
bbetween Euclidean distance;
(4) to the point in MR test pattern
, with
centered by get an image block
,
change into a column vector
, be used as a little
feature, wherein m representation vector
dimension;
(5) on training set MR image, training set distance field DF image respectively with point
centered by definition a search window choose image block
with
, build MR dictionary
with DF dictionary
, n is the size of dictionary;
(6) based on hypothesis I, MR dictionary is used
local linear represents MR test sample book
, dictionary
weights coefficient vector
solve by LAE method, expression is as follows:
,
……(2);
;
represent test sample book
at dictionary
in
kindividual neighbour;
(7) with the dictionary of step (6) gained
weights coefficient vector
linear combination DF dictionary
in sample, obtain test sample book
dF predicted picture block vector
, and handle
be converted into image block
; Specifically:
Based on hypothesis I, can obtain:
……(3);
Based on hypothesis II, can obtain:
……(4);
Due to
flocal linear, therefore:
……(5);
?
be converted into image block
can obtain
the distance field image block of prediction
;
(8) each repetition in MR test pattern is operated according to step (4)-(7), obtain the DF value of each point;
With
represent with point
xcentered by, size and
equal image block, for
in any point
u, its weight is:
……(6);
Point
xdF value be:
……(7);
represent with point
ucentered by the point predicted of image block
xdF value;
(9) to the DF predicted picture predicting out carry out threshold process, obtain the label corresponding to each point of test pattern, specifically:
The distance field defined from formula (1), point
xlabel can be derived by formula (8):
……(8);
Label is that this point of 1 expression belongs to segmentation object, and label is 0 expression background, obtains hippocampus segmentation image according to segmentation object.
Preferably, above-mentioned steps (1) specifically adopts gray scale normalization method to be normalized MR test pattern, removes skull with BET algorithm, removes biased field with N4 algorithm.
Preferably, above-mentioned steps (2) specifically uses DRAMMS Software tool that the MR training set image in training set and training set labeled graph picture are registrated to MR test pattern.
Preferably, above-mentioned steps (4) specifically extracts the feature of each point in MR test pattern, is used as test sample book.
Preferably, above-mentioned steps (5) specifically uses Local Search window to build MR dictionary and DF dictionary in MR training set image and DF training set image.
Preferably, above-mentioned steps (8), specifically to the DF predicted picture block weighted mean process of overlap, finally obtains the DF predicted value of each point.
The inventive method provides a kind of MR image hippocampus dividing method merged based on distance field, and the inventive method can split the hippocampus in MR image exactly, can be used for the diagnosis of the mental illnesses such as alzheimer disease.
Accompanying drawing explanation
The present invention is further illustrated to utilize accompanying drawing, but the content in accompanying drawing does not form any limitation of the invention.
Fig. 1 is the schematic flow sheet of the inventive method;
Fig. 2 is the frame diagram of the inventive method;
Fig. 3 is the details schematic flow sheet of the inventive method;
Fig. 4 is that the search window of the different size of the inventive method is to the segmentation result box traction substation of 15 test patterns; Wherein, Fig. 4 (a) is the segmentation result box traction substation for right hippocampus, and Fig. 4 (b) is the segmentation result box traction substation for left hippocampus;
Fig. 5 uses label fusion method and distance field fusion method to the segmentation result comparison diagram of 15 test patterns, and wherein, Fig. 5 (a) is the segmentation result comparison diagram for right hippocampus, and Fig. 5 (b) is the segmentation result comparison diagram for left hippocampus;
Fig. 6 is the coronal-plane schematic diagram adopting distance field fusion method mod sum label fusion method segmentation result on four test patterns, wherein, the first row represents manual segmentation result, second row is the segmentation result adopting distance field fusion method, the third line is the segmentation result adopting label fusion method, and a test pattern is shown in each list.
Embodiment
The invention will be further described with the following Examples.
Embodiment 1.
Based on the MR image hippocampus dividing method that distance field merges, the method is based on two kinds of hypothesis:
I, MR image block and DF image block are positioned on two non-linearity manifolds, and neighbour's sample that any one MR image block can be flowed in the local space of shape by its place is linearly expressed;
II, under local constraint, the mapping that MR stream shape flows shape to DF is similar to a differomorphism mapping.
The MR image hippocampus dividing method that should merge based on distance field carries out as follows:
(1) to MR test pattern normalized initially to be split, remove skull and biased field, obtain the MR test pattern after normalized.Specifically adopt gray scale normalization method to be normalized MR test pattern, remove skull with BET algorithm, remove biased field with N4 algorithm.
(2) the MR training set image in pre-prepd training set and the training set labeled graph picture corresponding with MR training set image are registrated to MR test pattern, make MR training set image, training set labeled graph picture aligns with MR test pattern on locus.Specifically use DRAMMS Software tool that the MR training set image in training set and training set labeled graph picture are registrated to MR test pattern in step (2).
(3) range conversion is carried out to the training set labeled graph picture after step (2) registration, obtain the training set distance field DF of training set labeled graph picture, point
the value of corresponding training set distance field DF is:
……(1);
Wherein
crepresent the border of segmentation object,
brepresent distance
xnearest point, and
,
represent point
xand point
bbetween Euclidean distance.
(4) to the point in MR test pattern
, with
centered by get an image block
,
change into a column vector
, be used as a little
feature, wherein m representation vector
dimension.
(5) on training set MR image, training set distance field DF image respectively with point
centered by definition a search window choose image block
with
, build MR dictionary
with DF dictionary
, n is the size of dictionary.
(6) based on hypothesis I, MR dictionary is used
local linear represents MR test sample book
, dictionary
weights coefficient vector
solve by LAE method, expression is as follows:
,
……(2);
;
represent test sample book
at dictionary
in
kindividual neighbour.
(7) with the dictionary of step (6) gained
weights coefficient vector
linear combination DF dictionary
in sample, obtain test sample book
dF predicted picture block vector
, and handle
be converted into image block
; Specifically:
Based on hypothesis I, can obtain:
……(3);
Based on hypothesis II, can obtain:
……(4);
Due to
flocal linear, therefore:
……(5);
?
be converted into image block
can obtain
the distance field image block of prediction
.
(8) each repetition in MR test pattern is operated according to step (4)-(7), obtain the DF value of each point;
With
represent with point
xcentered by, size and
equal image block, for
in any point
u, its weight is:
……(6);
Point
xdF value be:
……(7);
represent with point
ucentered by the point predicted of image block
xdF value.
(9) to the DF predicted picture predicting out carry out threshold process, obtain the label corresponding to each point of test pattern, specifically:
The distance field defined from formula (1), point
xlabel can be derived by formula (8):
……(8);
Label is that this point of 1 expression belongs to segmentation object, and label is 0 expression background, obtains hippocampus segmentation image according to segmentation object.
Preferably, above-mentioned steps (4) specifically extracts the feature of each point in MR test pattern, is used as test sample book.
Preferably, above-mentioned steps (5) specifically uses Local Search window to build MR dictionary and DF dictionary in MR training set image and DF training set image.
Preferably, above-mentioned steps (8), specifically to the DF predicted picture block weighted mean process of overlap, finally obtains the DF predicted value of each point.
The inventive method provides a kind of MR image hippocampus dividing method merged based on distance field, make use of the training set MR image of priori, training set distance field DF image, distance field is utilized to process, the hippocampus in MR image can be split exactly, can be used for the diagnosis of the mental illnesses such as alzheimer disease.
embodiment 2.
In order to verify the validity of the inventive method, with comprise in database 35 groups of MR brain data for basis is verified.The data often organizing MR brain data comprise the T1 weighted MR image of same patient and corresponding hippocampus labeled graph picture, and wherein random choose 20 groups of data are as training set, and all the other 15 groups of data are as test set.In experiment, test set object only adopts its MR image, does not adopt the hippocampus labeled graph picture in the database corresponding with it.
The present invention is based on the MR image hippocampus dividing method that distance field merges, the method is based on two kinds of hypothesis:
I, MR image block and DF image block are positioned on two non-linearity manifolds, and neighbour's sample that any one MR image block can be flowed in the local space of shape by its place is linearly expressed;
II, under local constraint, the mapping that MR stream shape flows shape to DF is similar to a differomorphism mapping.
The MR image hippocampus dividing method that should merge based on distance field carries out as follows:
(1) to MR test pattern normalized initially to be split, remove skull and biased field, obtain the MR test pattern after normalized.Specifically adopt gray scale normalization method to be normalized MR test pattern, remove skull with BET algorithm, remove biased field with N4 algorithm.
(2) the MR training set image in training set and the training set labeled graph picture corresponding with MR training set image are registrated to MR test pattern, make MR training set image, training set labeled graph picture aligns with MR test pattern on locus.Specifically use DRAMMS Software tool that the MR training set image in training set and training set labeled graph picture are registrated to MR test pattern in step (2).
(3) range conversion is carried out to the training set labeled graph picture after step (2) registration, obtain the training set distance field DF of training set labeled graph picture, point
the value of corresponding training set distance field DF is:
……(1);
Wherein
crepresent the border of segmentation object,
brepresent distance
xnearest point, and
,
represent point
xand point
bbetween Euclidean distance.
(4) to the point in MR test pattern
, with
centered by get an image block
,
change into a column vector
, be used as a little
feature, wherein m representation vector
dimension.
(5) on training set MR image, training set distance field DF image respectively with point
centered by definition a search window choose image block
with
, build MR dictionary
with DF dictionary
, n is the size of dictionary.
(6) based on hypothesis I, MR dictionary is used
local linear represents MR test sample book
, dictionary
weights coefficient vector
solve by LAE method, expression is as follows:
,
……(2);
;
represent test sample book
at dictionary
in
kindividual neighbour.
(7) with the dictionary of step (6) gained
weights coefficient vector
linear combination DF dictionary
in sample, obtain test sample book
dF predicted picture block vector
, and handle
be converted into image block
; Specifically:
Based on hypothesis I, can obtain:
……(3);
Based on hypothesis II, can obtain:
……(4);
Due to
flocal linear, therefore:
……(5);
?
be converted into image block
can obtain
the distance field image block of prediction
.
(8) each repetition in MR test pattern is operated according to step (4)-(7), obtain the DF value of each point;
With
represent with point
xcentered by, size and
equal image block, for
in any point
u, its weight is:
……(6);
Point
xdF value be:
……(7);
represent with point
ucentered by the point predicted of image block
xdF value.
(9) to the DF predicted picture predicting out carry out threshold process, obtain the label corresponding to each point of test pattern, specifically:
The distance field defined from formula (1), point
xlabel can be derived by formula (8):
……(8);
Label is that this point of 1 expression belongs to segmentation object, and label is 0 expression background, obtains hippocampus segmentation image according to segmentation object.
By with DSC(Dicesimilaritycoefficient, likeness coefficient) evaluate the accuracy of segmentation result, consider two factors: the overall performance of (1) this method; (2) compared with label fusion method (being called for short LF), the validity of distance field.
The inventive method is as follows hereinafter referred to as the experimental result of DFF method:
Fig. 4 is that the search window of the different size of the inventive method is to the segmentation result box traction substation of 15 test patterns; Wherein, Fig. 4 (a) is the segmentation result box traction substation for right hippocampus, and Fig. 4 (b) is the segmentation result box traction substation for left hippocampus.
Fig. 5 uses label fusion method and distance field fusion method to the segmentation result comparison diagram of 15 test patterns, and wherein, Fig. 5 (a) is the segmentation result comparison diagram for right hippocampus, and Fig. 5 (b) is the segmentation result comparison diagram for left hippocampus.Can find out that the DSC of the result that DFF is split is obviously good than LF method.
Fig. 6 is the coronal-plane schematic diagram adopting distance field fusion method mod sum label fusion method segmentation result on four test patterns, wherein, the first row represents manual segmentation result, second row is the segmentation result adopting distance field fusion method, the third line is the segmentation result adopting label fusion method, and a test pattern is shown in each list.Can find out that the result of DFF method is higher than the result precision of LF method.
Above result shows, uses the MR image hippocampus dividing method merged based on distance field can split hippocampus in MR brain image exactly, and then can be used for the diagnosis of the mental illnesses such as alzheimer disease.
Finally should be noted that; above embodiment is only in order to illustrate technical scheme of the present invention but not limiting the scope of the invention; although be explained in detail the present invention with reference to preferred embodiment; those of ordinary skill in the art is to be understood that; can modify to technical scheme of the present invention or equivalent replacement, and not depart from essence and the scope of technical solution of the present invention.
Claims (6)
1., based on the MR image hippocampus dividing method that distance field merges, it is characterized in that:
The MR image hippocampus dividing method that should merge based on distance field is based on two kinds of hypothesis:
I, MR image block and DF image block are positioned on two non-linearity manifolds, and neighbour's sample that any one MR image block can be flowed in the local space of shape by its place is linearly expressed;
II, under local constraint, the mapping that MR stream shape flows shape to DF is similar to a differomorphism mapping;
The MR image hippocampus dividing method that should merge based on distance field carries out as follows:
(1) to MR test pattern normalized initially to be split, remove skull and biased field, obtain the MR test pattern after normalized;
(2) the MR training set image in pre-prepd training set and the training set labeled graph picture corresponding with MR training set image are registrated to MR test pattern, make MR training set image, training set labeled graph picture aligns with MR test pattern on locus;
(3) range conversion is carried out to the training set labeled graph picture after step (2) registration, obtain the training set distance field DF of training set labeled graph picture, point
the value of corresponding training set distance field DF is:
……(1);
Wherein
crepresent the border of segmentation object,
brepresent distance
xnearest point, and
,
represent point
xand point
bbetween Euclidean distance;
(4) to the point in MR test pattern
, with
centered by get an image block
,
change into a column vector
, be used as a little
feature, wherein m representation vector
dimension;
(5) on training set MR image, training set distance field DF image respectively with point
centered by definition a search window choose image block
with
, build MR dictionary
with DF dictionary
, n is the size of dictionary;
(6) based on hypothesis I, MR dictionary is used
local linear represents MR test sample book
, dictionary
weights coefficient vector
solve by LAE method, expression is as follows:
,
……(2);
;
represent test sample book
at dictionary
in
kindividual neighbour;
(7) with the dictionary of step (6) gained
weights coefficient vector
linear combination DF dictionary
in sample, obtain test sample book
dF predicted picture block vector
, and handle
be converted into image block
; Specifically:
Based on hypothesis I, can obtain:
……(3);
Based on hypothesis II, can obtain:
……(4);
Due to
flocal linear, therefore:
……(5);
?
be converted into image block
can obtain
the distance field image block of prediction
;
(8) each repetition in MR test pattern is operated according to step (4)-(7), obtain the DF value of each point;
With
represent with point
xcentered by, size and
equal image block, for
in any point
u, its weight is:
……(6);
Point
xdF value be:
……(7);
represent with point
ucentered by the point predicted of image block
xdF value;
(9) to the DF predicted picture predicting out carry out threshold process, obtain the label corresponding to each point of test pattern, specifically:
The distance field defined from formula (1), point
xlabel can be derived by formula (8):
……(8);
Label is that this point of 1 expression belongs to segmentation object, and label is 0 expression background, obtains hippocampus segmentation image according to segmentation object.
2. a kind of MR image hippocampus dividing method merged based on distance field according to claim 1, it is characterized in that: described step (1) specifically adopts gray scale normalization method to be normalized MR test pattern, remove skull with BET algorithm, remove biased field with N4 algorithm.
3. a kind of MR image hippocampus dividing method merged based on distance field according to claim 2, is characterized in that: described step (2) specifically uses DRAMMS Software tool that the MR training set image in training set and training set labeled graph picture are registrated to MR test pattern.
4. a kind of MR image hippocampus dividing method merged based on distance field according to claims 1 to 3 any one, is characterized in that: described step (4) specifically extracts the feature of each point in MR test pattern, is used as test sample book.
5. a kind of MR image hippocampus dividing method merged based on distance field according to claim 4, is characterized in that: described step (5) specifically uses Local Search window to build MR dictionary and DF dictionary in MR training set image and DF training set image.
6. a kind of MR image hippocampus dividing method merged based on distance field according to claim 5, is characterized in that: described step (8), specifically to the DF predicted picture block weighted mean process of overlap, finally obtains the DF predicted value of each point.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510539859.4A CN105205810B (en) | 2015-08-30 | 2015-08-30 | MR image hippocampus dividing method based on distance field fusion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510539859.4A CN105205810B (en) | 2015-08-30 | 2015-08-30 | MR image hippocampus dividing method based on distance field fusion |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105205810A true CN105205810A (en) | 2015-12-30 |
CN105205810B CN105205810B (en) | 2018-12-11 |
Family
ID=54953472
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510539859.4A Active CN105205810B (en) | 2015-08-30 | 2015-08-30 | MR image hippocampus dividing method based on distance field fusion |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105205810B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106530321A (en) * | 2016-10-28 | 2017-03-22 | 南方医科大学 | Multi-graph image segmentation based on direction and scale descriptors |
CN106780532A (en) * | 2016-12-30 | 2017-05-31 | 陕西海泰电子有限责任公司 | A kind of intermodulation frequency equation method for quick towards DF figures |
CN107292346A (en) * | 2017-07-05 | 2017-10-24 | 四川大学 | A kind of MR image hippocampus partitioning algorithms learnt based on Local Subspace |
CN107392907A (en) * | 2017-09-01 | 2017-11-24 | 上海理工大学 | Parahippocampal gyrus function division method based on tranquillization state FMRI |
CN107424145A (en) * | 2017-06-08 | 2017-12-01 | 广州中国科学院软件应用技术研究所 | The dividing method of nuclear magnetic resonance image based on three-dimensional full convolutional neural networks |
CN108596924A (en) * | 2018-05-17 | 2018-09-28 | 南方医科大学 | A kind of MR prostate image partition methods based on distance field fusion and ellipsoid priori |
CN108937934A (en) * | 2018-05-07 | 2018-12-07 | 武汉科技大学 | A kind of detection of Brain MRI hippocampus and dividing method based on deep learning |
CN109886972A (en) * | 2019-01-24 | 2019-06-14 | 山西大学 | A kind of brain magnetic resonance image partition method based on multilayer dictionary |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1952981A (en) * | 2005-07-13 | 2007-04-25 | 西门子共同研究公司 | Method for knowledge based image segmentation using shape models |
US20120087561A1 (en) * | 2010-10-12 | 2012-04-12 | Siemens Corporation | Interaction method for regions of-interest in time series images |
CN104794739A (en) * | 2015-05-03 | 2015-07-22 | 南方医科大学 | Method for predicting CT (computerized tomography) image from MR (magnetic resonance) image on the basis of combination of corresponding partial sparse points |
-
2015
- 2015-08-30 CN CN201510539859.4A patent/CN105205810B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1952981A (en) * | 2005-07-13 | 2007-04-25 | 西门子共同研究公司 | Method for knowledge based image segmentation using shape models |
US20120087561A1 (en) * | 2010-10-12 | 2012-04-12 | Siemens Corporation | Interaction method for regions of-interest in time series images |
CN104794739A (en) * | 2015-05-03 | 2015-07-22 | 南方医科大学 | Method for predicting CT (computerized tomography) image from MR (magnetic resonance) image on the basis of combination of corresponding partial sparse points |
Non-Patent Citations (3)
Title |
---|
QIAN ZHENG 等: "Adaptive Segmentation of Vertebral Bodies from Sagittal MR Images Based on Local Spatial Information and Gaussian Weighted Chi-Square Distance", 《SOCIETY FOR IMAGING INFORMATICS IN MEDICINE》 * |
XABIER ARTAECHEVARRIA 等: "Combination Strategies in Multi-Atlas Image Segmentation: Application to Brain MR Data", 《IEEE TRANSACTIONS ON MEDICAL IMAGING》 * |
张雷 等: "基于多权重概率图谱的脑部图像分割", 《南方医科大学学报》 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106530321A (en) * | 2016-10-28 | 2017-03-22 | 南方医科大学 | Multi-graph image segmentation based on direction and scale descriptors |
CN106530321B (en) * | 2016-10-28 | 2019-07-12 | 南方医科大学 | A kind of multichannel chromatogram image partition method based on direction and scale description |
CN106780532A (en) * | 2016-12-30 | 2017-05-31 | 陕西海泰电子有限责任公司 | A kind of intermodulation frequency equation method for quick towards DF figures |
CN107424145A (en) * | 2017-06-08 | 2017-12-01 | 广州中国科学院软件应用技术研究所 | The dividing method of nuclear magnetic resonance image based on three-dimensional full convolutional neural networks |
CN107292346A (en) * | 2017-07-05 | 2017-10-24 | 四川大学 | A kind of MR image hippocampus partitioning algorithms learnt based on Local Subspace |
CN107292346B (en) * | 2017-07-05 | 2019-11-15 | 四川大学 | A kind of MR image hippocampus partitioning algorithm based on Local Subspace study |
CN107392907A (en) * | 2017-09-01 | 2017-11-24 | 上海理工大学 | Parahippocampal gyrus function division method based on tranquillization state FMRI |
CN108937934A (en) * | 2018-05-07 | 2018-12-07 | 武汉科技大学 | A kind of detection of Brain MRI hippocampus and dividing method based on deep learning |
CN108596924A (en) * | 2018-05-17 | 2018-09-28 | 南方医科大学 | A kind of MR prostate image partition methods based on distance field fusion and ellipsoid priori |
CN109886972A (en) * | 2019-01-24 | 2019-06-14 | 山西大学 | A kind of brain magnetic resonance image partition method based on multilayer dictionary |
Also Published As
Publication number | Publication date |
---|---|
CN105205810B (en) | 2018-12-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105205810A (en) | Distance-field-fusion-based hippocampus segmentation method of MR image | |
Cheng et al. | Classification of MR brain images by combination of multi-CNNs for AD diagnosis | |
Liu et al. | Anatomical landmark based deep feature representation for MR images in brain disease diagnosis | |
CN109035263B (en) | Automatic brain tumor image segmentation method based on convolutional neural network | |
Zhang et al. | MI-UNet: multi-inputs UNet incorporating brain parcellation for stroke lesion segmentation from T1-weighted magnetic resonance images | |
CN101666865B (en) | Method for registering diffusion tensor nuclear magnetic resonance image in local quick traveling mode | |
Zhang et al. | Multiview feature learning with multiatlas-based functional connectivity networks for MCI diagnosis | |
US20170069086A1 (en) | Method and device for constructing brain templates | |
CN104700120A (en) | fMRI feature extraction and classification method based on adaptive entropy projection clustering algorithm | |
CN109087298B (en) | Alzheimer's disease MRI image classification method | |
CN108171697B (en) | WMH automatic extraction system based on cluster | |
Jackson et al. | The graded change in connectivity across the ventromedial prefrontal cortex reveals distinct subregions | |
CN112862805B (en) | Automatic auditory neuroma image segmentation method and system | |
Jitsuishi et al. | Searching for optimal machine learning model to classify mild cognitive impairment (MCI) subtypes using multimodal MRI data | |
Jog et al. | Pulse sequence resilient fast brain segmentation | |
Wu et al. | A multi-atlas guided 3D fully convolutional network for MRI-based subcortical segmentation | |
Dai et al. | Transfer learning from partial annotations for whole brain segmentation | |
Zhang et al. | Deep white matter analysis: fast, consistent tractography segmentation across populations and dMRI acquisitions | |
Liu et al. | Anatomical-landmark-based deep learning for Alzheimer’s disease diagnosis with structural magnetic resonance imaging | |
Wang et al. | Bact-3D: A level set segmentation approach for dense multi-layered 3D bacterial biofilms | |
CN104463885B (en) | A kind of Multiple Sclerosis lesions region segmentation method | |
Jin et al. | Unsupervised detection of individual atrophy in Alzheimer's disease | |
Andica et al. | Automated three-dimensional major white matter bundle segmentation using diffusion magnetic resonance imaging | |
CN112200810B (en) | Multi-modal automated ventricle segmentation system and method of use thereof | |
Wagstyl et al. | Automated segmentation of cortical layers in BigBrain reveals divergent cortical and laminar thickness gradients in sensory and motor cortices |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |