CN103985092A - Post-processing noise elimination method for performing ICA analysis of plural f MRI data - Google Patents
Post-processing noise elimination method for performing ICA analysis of plural f MRI data Download PDFInfo
- Publication number
- CN103985092A CN103985092A CN201410191416.6A CN201410191416A CN103985092A CN 103985092 A CN103985092 A CN 103985092A CN 201410191416 A CN201410191416 A CN 201410191416A CN 103985092 A CN103985092 A CN 103985092A
- Authority
- CN
- China
- Prior art keywords
- voxel
- tested
- space
- phase
- brain district
- 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
Landscapes
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention provides a post-processing noise elimination method for performing ICA analysis of plural fMRI data, and belongs to the field of plural fMRI data analysis. As for phase-free ambiguous space activation encephalic region components (please see the representation in the specification), firstly, a phase value range of a phase image (please see the representation in the specification) of the component is -pi to pi, the -pi is not included, voxel corresponding to the phase value within the range from -pi/4 to pi/4 is defined as the activation voxel, so that a single-tested-set noise elimination mask and a multi-tested-set average noise elimination mask are built and used for a single test (please see the representation in the specification) and the set average component, the voxel with small amplitude is removed, and a final noise elimination result is obtained. The method can guarantee that the ICA performs analysis of plural fMRI data, and further solves the problems of data loss and brain information loss caused by pre-processing noise elimination. As for the plural fMRI data collected under activity stimulation, only 26 independent components can be estimated through the preprocessing noise eliminating method, and 49 independent components can be estimated through the method.
Description
Technical field
The ICA that the present invention relates to plural fMRI data analyzes, and particularly relates to a kind of aftertreatment noise-eliminating method that plural fMRI data is carried out to ICA analysis.
Background technology
Brain function research is worldwide Focal point and difficult point, and functional mri (functional magnetic resonance imaging, fMRI) rely on its not damaged and high spatial resolution advantage, become one of important means of brain function research.By adopting the analytical approach of model-driven as GLM (general linear model), or the analytical approach of data-driven is as independent component analysis (independent component analysis, ICA), people can activate brain district information from the abundant space of fMRI extracting data.
FMRI data are plural numbers, comprise amplitude data and phase data.But most of fMRI analytical approach is only analyzed amplitude data, and abandons phase data completely.Its reason is, the noise ratio amplitude data of phase data are large and characteristic is unknown, and this causes the analysis of plural fMRI data to have challenge.But, in fMRI phase data, containing unique brain function composition, in order to disclose complete brain function information, must be used.For this reason, started to explore from the beginning of, people the ICA method of utilizing plural fMRI data from 2000.2002, first the people such as Calhoun were used for plural fMRI data analysis (Calhoun, V.D. by ICA, Adali, T., Pearlson, G., Van Zijl, P., Pekar, J., 2002.Independent component analysis of fMRI data in the complex domain.Magnetic Resonance in Medicine48,180-192).In order to eliminate phase noise, the method had abandoned the voxel of brain first half before ICA.In other words, ICA now has only utilized approximately 50% plural fMRI data.Even like this, compared with only analyzing the ICA method of fMRI amplitude data, the continuous activation number of voxels that Calhoun method is extracted task Related Component also on average exceeds 12-23%.2009, the people such as Rodriguez have proposed a kind of new phase place noise-eliminating method-phase place Quality Map (phase quality map) Denoising Method (Rodriguez, P.A., Correa, N.M., Eichele, T., Calhoun, V.D., Adali, T., 2009.Quality map thresholding for de-noising of complex-valued fMRI data and its application to ICA of fMRI.IEEE International Workshop on Machine Learning for Signal Processing (MLSP), 1-6).The method changes phase place in fMRI phase data higher than the artificial voxel of setting thresholding and is defined as noise voxel, before ICA, it is removed from plural fMRI data.Experiment shows, this part is removed data and accounts for 30% of fMRI total amount of data, that is to say, in the time utilizing phase place Quality Map method to carry out de-noising, ICA can utilize approximately 70% plural fMRI data, and data user rate is improved, the more horn of plenty of brain function information recovering.Therefore, phase place Quality Map Denoising Method is applied in the ICA of many sections of plural fMRI data analyzes document.
But, to be not difficult to find out, above-mentioned two kinds of methods of eliminating phase noise had all been removed the plural fMRI data of part (about at least 30%) before ICA, and these are removed in data and also contain abundant brain function information.With the plural fMRI data instance gathering under a kind of Motor stimulation, the independent component number that plural fMRI data comprise before de-noising is not less than 40, but reduces to 26 after phase place de-noising, that is to say, about 14 brain function compositions lack because of the disappearance of carrier data.Visible, in ICA analyzes, utilizing plural fMRI data is to extract the key of complete brain function information.
Summary of the invention
The object of the invention is to, cancel ICA phase place de-noising pre-service before, the substitute is a kind of method that activation brain district, space composition is carried out to effective de-noising is provided after ICA, ensure that ICA can, to single tested and how tested plural fMRI data analysis, solve the data that cause because of de-noising pre-service and abandon and brain information loss problem.
Technical scheme of the present invention is, for single tested plural fMRI data, directly carries out ICA, obtains activation brain district, K space composition
k=1 ..., K, the number that K is independent component.Right
carry out phase correction, obtain activating brain district composition without the space of phase ambiguity
owing to there is no pre-service de-noising,
contain a large amount of high amplitude noise voxels, cannot eliminate with threshold method (removing amplitude lower than the voxel of a certain threshold value).For this reason, the present invention utilizes source phase image to carry out de-noising.Particularly, brain district composition is activated in space
for complex signal, comprise magnitude image and phase image, note phase image is
for being different from original plural fMRI phase data (belonging to blended data), claim
for source phase image, its phase place span be (π, π].?
in, definition [π/4, π/4] phase value is corresponding in scope voxel is for activating voxel, other phase values, comprise (π ,-π/4) and (π/4, π] phase value in scope, corresponding voxel is for disturbing voxel, and the mask that is suitable for single tested space activation brain district composition aftertreatment de-noising builds as follows:
L is the pointer that voxel in brain district composition is activated in space.BM1 is acted on noisy
brain district composition is activated in the space that obtains removing much noise voxel, only retaining activation voxel:
finally, remove
middle amplitude small as
voxel, obtain the tested space of the list of final de-noising and activate brain district composition.
For how tested plural fMRI data, carry out respectively the tested ICA of list, obtain each tested space and activate brain district composition
p=1 ..., P, P is tested number.Right
carry out phase correction, obtain activating brain district composition without the space of phase ambiguity
from
in obtain source phase image
and build respectively single tested de-noising mask, be designated as BM1
p, by BM1
pact on noisy
brain district composition is activated in the tested space of list that obtain and remove much noise voxel, only retains activation voxel
again by
the group mean space that obtains preliminary de-noising activates brain district composition
in order to eliminate the not voxel in the tested middle activation of half, at single tested BM1
pon basis, how tested the de-noising mask that builds group mean space activation brain district composition be as follows:
BM2 is acted on
the group mean space that obtains further de-noising activates brain district composition
remove
middle amplitude small as
voxel, the how tested group mean space that obtains final de-noising activates brain district composition.
Effect and benefit that the present invention reaches are, with ICA pre-service Denoising Method, compare as phase place Quality Map Denoising Method, before ICA, plural fMRI data are removed without any voxel, ensure in ICA analyzes and utilize plural fMRI data, to obtain more brain function information.For example, for the plural fMRI data that gather under Motor stimulation, pre-service Denoising Method is only estimated 26 independent brain function compositions, and the present invention can estimate 49 independent brain function compositions.Activate the problem that has high amplitude noise in brain district composition for thing followed space, the invention provides the aftertreatment Denoising Method based on source phase image, remove noise voxel, retain and activate voxel according to phase range instead of amplitude size.Therefore, the present invention can ensure that ICA is to single tested and how tested plural fMRI data analysis, and then the data that solution causes because of pre-service de-noising abandon and brain information loss problem.
Brief description of the drawings
Fig. 1 activates to the task correlation space of single tested plural fMRI data the concrete steps that brain district composition carries out aftertreatment de-noising.
Fig. 2 activates to the average task correlation space of the group of 16 tested plural fMRI data the concrete steps that brain district composition carries out aftertreatment de-noising.
Embodiment
Below in conjunction with technical scheme and accompanying drawing, describe two specific embodiments of the present invention in detail.
Specific embodiment one: supposing the tested plural fMRI data of the list gathering under existing Motor stimulation, is 165 to the scanning times of tested full brain.This plural number fMRI data are carried out to ICA, and the task correlation space that obtains containing a large amount of high amplitude noises activates brain district composition, and the concrete steps that it is carried out to aftertreatment de-noising as shown in Figure 1.
The first step, according to the independent component number of the tested plural fMRI data of information theory criterion estimate sheet, result is 30, adopts PCA (principle component analysis) method that plural fMRI data are down to 30 dimensions from 165 dimensions.
Second step, the plural fMRI data after adopting plural EBM (entropy bound minimization) algorithm to PCA dimensionality reduction are carried out ICA, obtain 30 plural ICA and estimate compositions, therefrom select task correlation space to activate brain district composition
in this composition, contain a large amount of high amplitude noise voxels.
The 3rd step, selects the method for correcting phase pair based on time course composition
carry out phase correction, obtain activating brain district composition without the space of phase ambiguity
The 4th step, calculates source phase image
?
phase image.
The 5th step, adopts formula (1) to build single tested de-noising mask:BM1.
The 6th step, acts on BM1 noisy
the task correlation space that obtains removing much noise voxel, only retaining activation voxel activates brain district composition:
The 7th step, removes
in
voxel, the task correlation space that obtains final de-noising activates brain district composition.
Specific embodiment two: the 16 tested plural fMRI data that gather under existing Motor stimulation, the scanning times of each tested full brain is 165.16 tested plural fMRI data are carried out respectively to ICA, obtain concrete steps that the average task correlation space of the group of final de-noising activates brain district composition as shown in Figure 2.
The first step, adopts PCA that 16 tested plural fMRI data are down to 30 dimensions from 165 dimensions respectively.
Second step, 16 tested plural fMRI data after adopting plural EBM algorithm to PCA dimensionality reduction are carried out respectively ICA, select each tested task correlation space to activate brain district composition
in these compositions, contain a large amount of high amplitude noise voxels.
The 3rd step, selects the method for correcting phase based on time course composition right respectively
carry out phase correction, obtain activating brain district composition without the space of phase ambiguity
The 4th step, calculates source phase image
?
phase image.
The 5th step, adopts formula (1) to build each tested de-noising mask:BM1
1..., BM1
16.
The 6th step, by BM1
1..., BM1
16act on respectively noisy
the task correlation space that obtains denoising activates brain district composition:
The 7th step, the group mean space that obtains preliminary de-noising activates brain district composition
The 8th step, adopts formula (2) to build how tested de-noising mask:BM2.
The 9th step, acts on BM2
the group mean space that obtains further de-noising activates brain district composition
The tenth step, removes
voxel, obtain the average task correlation space of the group of final de-noising and activate brain district composition.
Claims (2)
1. an aftertreatment noise-eliminating method that plural fMRI data is carried out to ICA analysis, is characterized in that, for single tested plural fMRI data, directly carries out ICA, obtains K space and activates brain district composition
k=1 ..., K, the number that K is independent component; Right
carry out phase correction, obtain activating brain district composition without the space of phase ambiguity
obtain source phase image
?
phase image, phase place span be (π, π]; ?
in, definition [π/4, π/4] phase value is corresponding in scope voxel is for activating voxel, other phase values, comprise (π ,-π/4) and (π/4, π] phase value in scope, corresponding voxel is for disturbing voxel, build the mask that is suitable for activating the composition aftertreatment de-noising of brain district in single tested space as follows:
L is the pointer that voxel in brain district composition is activated in space; BM1 is acted on noisy
brain district composition is activated in the space that obtains removing much noise voxel, only retaining activation voxel:
finally, remove
the voxel that middle amplitude is small, brain district composition is activated in the tested space of list that obtains final de-noising;
For how tested plural fMRI data, carry out respectively the tested ICA of list, obtain each tested space and activate brain district composition
p=1 ..., P, P is tested number; Right
carry out phase correction, obtain activating brain district composition without the space of phase ambiguity
from
in obtain source phase image
and build respectively single tested de-noising mask, be designated as BM1
p, by BM1
pact on noisy
brain district composition is activated in the tested space of list that obtain and remove much noise voxel, only retains activation voxel
again by
the group mean space that obtains preliminary de-noising activates brain district composition
at single tested BM1
pon basis, build the de-noising mask of how tested group mean space activation brain district composition, eliminate the not voxel in the tested middle activation of half:
BM2 is acted on
the group mean space that obtains further de-noising activates brain district composition
remove
middle amplitude small as
voxel, the how tested group mean space that obtains final de-noising activates brain district composition.
2. aftertreatment noise-eliminating method according to claim 1, is characterized in that, the small voxel of amplitude is defined as the voxel that amplitude is less than 0.5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410191416.6A CN103985092B (en) | 2014-05-07 | 2014-05-07 | A kind of post processing noise-eliminating method that plural number fMRI data are carried out ICA analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410191416.6A CN103985092B (en) | 2014-05-07 | 2014-05-07 | A kind of post processing noise-eliminating method that plural number fMRI data are carried out ICA analysis |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103985092A true CN103985092A (en) | 2014-08-13 |
CN103985092B CN103985092B (en) | 2017-01-04 |
Family
ID=51277050
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410191416.6A Active CN103985092B (en) | 2014-05-07 | 2014-05-07 | A kind of post processing noise-eliminating method that plural number fMRI data are carried out ICA analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103985092B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105760700A (en) * | 2016-03-18 | 2016-07-13 | 大连理工大学 | Adaptive fixed-point IVA algorithm applicable to analysis on multi-subject complex fMRI data |
CN105912851A (en) * | 2016-04-07 | 2016-08-31 | 大连理工大学 | Method for estimating model order of complex fMRI data by utilization of PCA and non-annular characteristics |
CN106875366A (en) * | 2017-03-01 | 2017-06-20 | 大连理工大学 | Tranquillization state plural number fMRI data are carried out with the phase exact extension detection method that ICA post-processes de-noising |
CN108903942A (en) * | 2018-07-09 | 2018-11-30 | 大连理工大学 | A method of utilizing plural number fMRI spatial source phase identification spatial diversity |
CN114176518A (en) * | 2021-12-06 | 2022-03-15 | 大连理工大学 | Complex fMRI data space component phase inverse correction method for improving CNN classification performance |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103345749A (en) * | 2013-06-27 | 2013-10-09 | 中国科学院自动化研究所 | Method for detecting brain network function connectivity lateralization based on modality fusion |
-
2014
- 2014-05-07 CN CN201410191416.6A patent/CN103985092B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103345749A (en) * | 2013-06-27 | 2013-10-09 | 中国科学院自动化研究所 | Method for detecting brain network function connectivity lateralization based on modality fusion |
Non-Patent Citations (2)
Title |
---|
QIU-HUA LIN: "SEMI-BLIND KURTOSIS MAXIMIZATION ALGORITHM APPLIED TO COMPLEX-VALUED FMRI DATA", 《2011 IEEE INTERNATIONAL WORKSHOP ON MACHINE LEARNING FOR SIGNAL PROCESSING》 * |
QIU-HUA LIN: "Semiblind Spatial ICA of fMRI Using Spatial Constraints", 《HUMAN BRAIN MAPPING》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105760700A (en) * | 2016-03-18 | 2016-07-13 | 大连理工大学 | Adaptive fixed-point IVA algorithm applicable to analysis on multi-subject complex fMRI data |
CN105760700B (en) * | 2016-03-18 | 2018-06-08 | 大连理工大学 | A kind of adaptive fixed point IVA algorithms for being suitable for more subject plural number fMRI data analyses |
CN105912851A (en) * | 2016-04-07 | 2016-08-31 | 大连理工大学 | Method for estimating model order of complex fMRI data by utilization of PCA and non-annular characteristics |
CN105912851B (en) * | 2016-04-07 | 2019-04-16 | 大连理工大学 | A method of utilizing PCA and other than ring type characteristic estimating plural number fMRI data model order |
CN106875366A (en) * | 2017-03-01 | 2017-06-20 | 大连理工大学 | Tranquillization state plural number fMRI data are carried out with the phase exact extension detection method that ICA post-processes de-noising |
CN106875366B (en) * | 2017-03-01 | 2019-06-21 | 大连理工大学 | Tranquillization state plural number fMRI data are carried out with the phase exact extension detection method of ICA post-processing de-noising |
CN108903942A (en) * | 2018-07-09 | 2018-11-30 | 大连理工大学 | A method of utilizing plural number fMRI spatial source phase identification spatial diversity |
CN114176518A (en) * | 2021-12-06 | 2022-03-15 | 大连理工大学 | Complex fMRI data space component phase inverse correction method for improving CNN classification performance |
CN114176518B (en) * | 2021-12-06 | 2023-10-10 | 大连理工大学 | Complex fMRI data space component phase anti-correction method for improving CNN classification performance |
Also Published As
Publication number | Publication date |
---|---|
CN103985092B (en) | 2017-01-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhang et al. | Joint image denoising using adaptive principal component analysis and self-similarity | |
Boudraa et al. | EMD-based signal noise reduction | |
CN105913393B (en) | A kind of adaptive wavelet threshold image de-noising method and device | |
CN103985092A (en) | Post-processing noise elimination method for performing ICA analysis of plural f MRI data | |
CN103020916B (en) | Image denoising method combining two-dimensional Hilbert transform and BEMD | |
CN102663695B (en) | DR image denoising method based on wavelet transformation and system thereof | |
CN103116873B (en) | Image denoising method | |
Ahmed et al. | Medical image denoising system based on stacked convolutional autoencoder for enhancing 2-dimensional gel electrophoresis noise reduction | |
Rajeesh et al. | Noise reduction in magnetic resonance images using wave atom shrinkage | |
CN104103041B (en) | Ultrasonoscopy mixed noise Adaptive Suppression method | |
CN102901630A (en) | Adaptive redundant lifting wavelet noise reduction analysis-based bearing failure recognition method | |
CN104217448A (en) | A magnetic-resonance fast imaging method and system based on iterative feature correction | |
Denis et al. | From patches to deep learning: Combining self-similarity and neural networks for SAR image despeckling | |
Manjón et al. | MRI denoising using Deep Learning and Non-local averaging | |
CN107590785A (en) | A kind of Brillouin spectrum image-recognizing method based on sobel operators | |
CN102314675B (en) | Wavelet high-frequency-based Bayesian denoising method | |
Luisier et al. | Fast Haar-wavelet denoising of multidimensional fluorescence microscopy data | |
CN101930605B (en) | Synthetic Aperture Radar (SAR) image target extraction method and system based on two-dimensional mixing transform | |
CN102136134A (en) | Method for removing speckles from SAR (Synthetic Aperture Radar) image based on MRF (Markov Random Field) priority | |
CN101510943A (en) | Method for effectively removing image noise using ultra-perfection topological sparseness encode | |
Kalaiselvi et al. | A novel wavelet thresholding technique to denoise magnetic resonance images‖ | |
Bhonsle et al. | De-noising of CT images using combined bivariate shrinkage and enhanced total variation technique | |
Garg et al. | Improving segmentation by denoising brain MRI images through interpolation median filter in ADTVFCM | |
Zhang et al. | Joint image denoising using self-similarity based low-rank approximations | |
CN104517267A (en) | Infrared image enhancement and reestablishment method based on spectra inversion |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |