CN105069797B - Ice mapping three-dimensional density figure resolution detection method based on mask - Google Patents
Ice mapping three-dimensional density figure resolution detection method based on mask Download PDFInfo
- Publication number
- CN105069797B CN105069797B CN201510495697.9A CN201510495697A CN105069797B CN 105069797 B CN105069797 B CN 105069797B CN 201510495697 A CN201510495697 A CN 201510495697A CN 105069797 B CN105069797 B CN 105069797B
- Authority
- CN
- China
- Prior art keywords
- dimensional
- dimensional density
- density
- mask
- film
- 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
Classifications
-
- 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
-
- 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
- G06T2207/10056—Microscopic image
- G06T2207/10061—Microscopic image from scanning electron microscope
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
Abstract
A kind of Ice mapping three-dimensional density figure resolution detection method based on mask, the three-dimensional density figure that Ice mapping is reconstructed is split by three-dimensional two-value mask, radius power spectrum of the outer background noise portions of intramembranous particles structure division and film based on the three-dimensional spherical shell of Fourier space of three-dimensional density figure is calculated respectively, obtain the signal power and noise power of three-dimensional density figure, and then the estimate that the obtained spectral signal-noise ratio curve based on mask obtains global resolution ratio with selected threshold value is calculated by spectral signal-noise ratio, realize the evaluation and test to the reconstruction quality of Ice mapping；The three-dimensional density figure that the present invention can be reconstructed directly from Ice mapping estimates resolution ratio, and algorithm is simple and easy to apply, there is larger application space.
Description
Technical field
The present invention relates to a kind of technology of image processing field, specifically a kind of Ice mapping based on mask is three-dimensional
Density map resolution detection method.
Background technology
Ice mapping (Cryo-EM) technology has become one essential research means of structure biology.With traditional X
Ray (X-ray) is compared with nuclear magnetic resonance (NMR), and Ice mapping has specific advantage, such as：(1) crystal need not be grown, because
Molecular structure that this is solved is closer to its natural physiological status；(2) can be used for solving the huge compound of molecular weight, virus,
The even three-dimensional structure of organelle；(3) dynamic structure of active bio macromolecular can be studied.In recent years, with electron microscopic
Mirror, data acquisition camera, highly effective algorithm and program, the development of high-performance computer, Ice mapping technology are led in structure biology
Domain is just playing more and more important effect, there is the trend of the Fashion of Future structure biology research greatly.
The reconstruct of individual particle Ice mapping is the main application of Ice mapping technology, and its main flow is as follows：Sample preparation, electricity
Mirror imaging acquisition, the CTF corrections of sem image, particle are selected, classify and are averaged, generation initial model and model iteration.It is cold
The objective evaluation standard for freezing Electronic Speculum reconstruction quality is an important research topic in the field, and individual particle Ice mapping is come
Say, the resolution ratio for reconstructing obtained three-dimensional density figure is to evaluate the major criterion of reconstruction quality.Resolution ratio is divided into global resolution ratio
And local resolution, the quality of the whole model of global resolution ratio reflection, reflect that the model to what extent parses biological big point
The structure of son；It is the mark of evaluation model local quality quality and local resolution reflects each pixel or the resolution ratio in region
It is accurate.
The method of resolution ratio estimation has calculating Fourier's shell with the development of Ice mapping, at present conventional method
Related (Fourier shell correlation, FSC) method and calculating spectral signal-noise ratio (Spectral signal-to-
Noise ratio, SSNR) method.Calculate FSC methods and the data set collected is divided into two independent half data collection, respectively
Two half data collection are independently reconstructed, the FSC between two reconstructed density figures is then calculated, finally estimates from FSC curves
Meter obtains global resolution ratio.But this method has following defect：This method depends on restructuring procedure, easily by noise and the shadow of over-fitting
Ring；The architecture quality of two independent half data collection reconstruct is reconstructed less than whole set of data, is the calculating process of redundancy；From FSC
Curve estimation resolution requirement has threshold value standard, and there is presently no unified threshold value standard.
SSNR definition is the power proportions of domain space signal and noise.The key for calculating SSNR methods is how to estimate
The power of signal and noise, traditional method is to carry out back projection to the three-dimensional density figure of reconstruct to obtain one group of X-Y scheme image set,
And estimated in Fourier's domain space from the power of the image set estimating signal from the difference of the image set and original input picture collection
Noise power.Actually this method is to calculate the uniformity of reconstruction result and input data, and this method is obviously dependent on anti-throwing
The process of shadow thing and need raw data set.
The resolution ratio method of estimation of above two dependence is obviously not phase with the fast development of current Ice mapping technology
Symbol, it is therefore desirable to a kind of method that objective estimation resolution ratio can be only gone by reconstruction result.
Found by the retrieval to prior art, Chinese patent literature CN104732531A, day for announcing 2015.6.24,
A kind of high-resolution remote sensing image signal to noise ratio curve self-adapting acquisition methods are disclosed, including：Fourier is carried out to entire image
Conversion, by the mask that different frequency is interval, tries to achieve the frequency separation that power spectral density tends towards stability；Noise spectrum after mask
Fourier inversion carries out non-overlapped fritter to spatial domain, to signal pattern and noise image and divided, and tries to achieve respectively each small
The average value and variance of block；The outlier of signal averaging and noise variance is removed, progressive alternate fitting obtains noise variance and letter
The linear relationship of number average value, tries to achieve the slope and intercept of straight line；Signal divided by noise criteria in whole intensity value ranges is inclined
Difference obtains signal to noise ratio curve.But the technology needs first to try to achieve the average value and variance of noise, then tries to achieve the average value of signal according to this
And variance, the calculation error of noise is brought into the calculating of signal, and iterative fitting is needed, process is complicated, there is error risk.
The content of the invention
The present invention is directed to deficiencies of the prior art, proposes a kind of Ice mapping three-dimensional density figure based on mask
Resolution detection method, splits three-dimensional density figure by mask, the signal power and noise power outside film inner membrance, root is tried to achieve respectively
Estimate global resolution ratio according to signal to noise ratio curve selected threshold, be not required to the reconstruct of dependency graph picture, can directly be estimated from density map, and calculate
Method is simple.
The present invention is achieved by the following technical solutions：
The present invention is split by three-dimensional two-value mask to the three-dimensional density figure that Ice mapping is reconstructed, and calculates three-dimensional respectively
Radius power spectrum of the outer background noise portions of the intramembranous particles structure division and film of density map based on the three-dimensional spherical shell of Fourier space
(Radial Power Spectrum, RPS), obtains the signal power and noise power of three-dimensional density figure, and then is believed by spectrum
Spectral signal-noise ratio (mask-SSNR, the mSSNR) curve based on mask obtained than calculating of making an uproar obtains the overall situation with selected threshold value and divided
The estimate of resolution, realizes the evaluation and test to the reconstruction quality of Ice mapping.
Described radius power spectrum refers to：Distribution of the power of density map in Fourier space frequency, i.e.,：RPS (s)=
∑k|Shell(s)|M(k)|2Wherein：S is the radius of the three-dimensional spherical shell of Fourier space, and k is the coordinate vector of Fourier space,
Shell (s) is the three-dimensional spherical shell that radius is s, and M (k) is the Fourier transformation of three-dimensional density figure.
Described spectral signal-noise ratio, which is calculated, to be referred to：Spectral signal-noise ratioWherein：
RPSiAnd RPSoThe RPS curves of the outer background noise portions of intramembranous particles structure division and film of respectively three-dimensional density figure, NiAnd No
The number of the length, i.e. non-zero pixels of background noise portions respectively outside intramembranous particles structure division and film.
Technique effect
Compared with prior art, the present invention calculates spectral signal-noise ratio with mask and calculates global resolution ratio according to this, is not required to
Data set is separated, intermediate data is not required to, you can global resolution ratio is individually estimated in the density map reconstructed from Ice mapping.
Brief description of the drawings
Fig. 1 is schematic diagram of the present invention；
Fig. 2 is embodiment schematic diagram；
In figure：1 is three-dimensional density figure, and 2 be three-dimensional two-value mask, and 3 be intramembranous particles structure, and 4 be the outer ambient noise of film, 5
For mSSNR curves.
Embodiment
Embodiments of the invention are elaborated below, the present embodiment is carried out lower premised on technical solution of the present invention
Implement, give detailed embodiment and specific operating process, but protection scope of the present invention is not limited to following implementations
Example.
Embodiment 1
As depicted in figs. 1 and 2, the present embodiment comprises the following steps：
Step 1, undressed three-dimensional density Fig. 1 is obtained from Ice mapping, one three is generated according to three-dimensional density Fig. 1
Tie up two-value mask 2.
Three-dimensional density Fig. 1 of the present embodiment is according to PDB entry：3J7H emulation data carry out individual particle reconstruct and obtained.
Value is that value is 0 outside 1, film in the film of described three-dimensional two-value mask 2；The mask is relaxation mask, and fully wrapped around
The firmly large biological molecule particle in three-dimensional density Fig. 1.
Step 2, by three-dimensional two-value mask 2 three-dimensional density Fig. 1 is split, the radius work(of partitioning portion is calculated respectively
Rate composes (RPS), obtains the RPS curves of the three-dimensional spherical shell respective radius based on Fourier space, and then pass through spectral signal-noise ratio meter
Calculation obtains SSNR (mSSNR) curve based on mask, specifically includes：
Step 2.1) two-value mask 2 is applied to three-dimensional density Fig. 1, three-dimensional density Fig. 1 is divided into two parts：In film
Grain structure portion 3 and the outer background noise portions 4 of film；Original three-dimensional density Fig. 1 is designated asIntramembranous particles structure division 3 is designated asThe outer background noise portions 4 of film are designated as
The three-dimensional density figure of described intramembranous particles structure division 3 and the outer background noise portions 4 of filmWithWith original three
Tie up density mapSize is identical, andFilm outer density value be 0,Film in density value be 0.
Step 2.2) it is right respectivelyWithThree-dimensional Fourier transform is carried out, the squared absolute value of Fourier transformation, and base is taken
The quadratic sum of Fourier's modulus value is sought in the three-dimensional spherical shell of Fourier space, corresponding curve RPS is obtainediAnd RPSo。
Described RPS is distribution of the power of density map in Fourier space frequency, i.e.,：RPS (s)=∑ k | Shell
(s)|M(k)|2, wherein：S is the radius of the three-dimensional spherical shell of Fourier space, and k is the coordinate vector of Fourier space, Shell (s)
The three-dimensional spherical shell for being s for radius, M (k) is the Fourier transformation of three-dimensional density figure.
Described three-dimensional spherical shell outer boundary is the sphere that radius is s+0.5, and inner boundary is the sphere that radius is s-0.5, ball
Thickness of the shell is 1.
Step 2.3) according to RPSiAnd RPSo, calculate original three-dimensional density figureSignal power and noise power, and lead to
Cross mSSNR (dB)=10log10SSNR is converted to the ratio of signal power and noise power, and mSSNR is obtained so as to calculate
Curve 5.
Described signal power exists only in intramembranous particles structure divisionNoise power is present in intramembranous particles structural portion
PointWith background noise portions outside film
Described noise power is in three-dimensional density figureMiddle random distribution, i.e. noise power and three-dimensional density figureSize
It is directly proportional, then can passes throughPower estimating noise power.
Described signal power can pass throughPower subtractThe corresponding noise power of size obtain.
Described SSNR calculation formula are：Wherein：RPSiAnd RPSoRespectively
The RPS curves of the outer background noise portions of the intramembranous particles structure division and film of three-dimensional density figure, NiAnd NoRespectively three-dimensional density figureIntramembranous particles structure divisionWith background noise portions outside filmLength, i.e. non-zero pixels number.
The radius length of the three-dimensional spherical shell of described Fourier space is frequency, and the inverse of the frequency is resolution ratio.
Step 3, selected threshold value, the estimate of global resolution ratio are obtained according to threshold value, so as to evaluate the reconstruct of Ice mapping
Quality.
Described SSNR is at threshold value, then signal power is equal with noise power；SSNR is square under the threshold value, then noise power
It is relatively serious more than signal power, i.e. noise pollution；SSNR is above threshold value, then signal power is more than noise power.
The present embodiment selected threshold is 1, and corresponding resolution ratio is three when to take SSNR critical values be 1 (i.e. SSNR (dB)=0)
Tie up density mapGlobal resolution ratio, then the global resolution ratio of the present embodiment be
Peak frequency on described RPS curves is nyquist frequency, therefore the minimum resolution reached in theory is
2 times of three-dimensional density Fig. 1 voxel lengths.
Claims (7)
1. a kind of Ice mapping three-dimensional density figure resolution detection method based on mask, it is characterised in that pass through three-dimensional two-value
Mask is split to the three-dimensional density figure that Ice mapping is reconstructed, respectively calculate three-dimensional density figure intramembranous particles structure division and
Radius power spectrum of the outer background noise portions of film based on the three-dimensional spherical shell of Fourier space, obtain three-dimensional density figure signal power and
Noise power, and then obtained spectral signal-noise ratio curve based on mask is calculated by spectral signal-noise ratio and selected threshold value is obtained
The estimate of global resolution ratio, realizes the evaluation and test to the reconstruction quality of Ice mapping；
Described radius power spectrum refers to：Distribution of the power of density map in Fourier space frequency, i.e.,：RPS (s)=∑ k |
Shell(s)|M(k)|2, wherein：S is the radius of the three-dimensional spherical shell of Fourier space, as frequency, and the inverse of the frequency is point
Resolution, k is the coordinate vector of Fourier space, and Shell (s) is the three-dimensional spherical shell that radius is s, and M (k) is Fu of three-dimensional density figure
In leaf transformation.
2. Ice mapping three-dimensional density figure resolution detection method according to claim 1, it is characterized in that, described three-dimensional
Large biological molecule particle in the three-dimensional density figure that the fully wrapped around residence of two-value mask is stated.
3. Ice mapping three-dimensional density figure resolution detection method according to claim 2, it is characterized in that, described point
Cut, i.e., three-dimensional density figure is divided into background noise portions outside intramembranous particles structure division and film, the intramembranous particles knot after segmentation
Structure part and the outer background noise portions of film are identical with initial three-dimensional density map size, and the film outer density of intramembranous particles structure division
Value is 0 with density value in the film of the outer background noise portions of film.
4. Ice mapping three-dimensional density figure resolution detection method according to claim 1, it is characterized in that, described three-dimensional
Spherical shell outer boundary is the sphere that radius is s+0.5, and inner boundary is the sphere that radius is s-0.5, and shell thickness is 1.
5. Ice mapping three-dimensional density figure resolution detection method according to claim 1, it is characterized in that, it is described based on
The spectral signal-noise ratio curve of mask, according to the respective radius power spectrum of background noise portions outside intramembranous particles structure division and film,
The signal power and noise power of three-dimensional density figure are calculated, and passes through mSSNR (dB)=10log10SSNR to signal power with
The ratio of noise power is converted.
6. Ice mapping three-dimensional density figure resolution detection method according to claim 1 or 5, it is characterized in that, it is described
Spectral signal-noise ratio, which is calculated, to be referred to：Spectral signal-noise ratioWherein：RPSiAnd RPSoRespectively
The RPS curves of the outer background noise portions of the intramembranous particles structure division and film of three-dimensional density figure, NiAnd NoRespectively intramembranous particles knot
The number of the length, i.e. non-zero pixels of structure part and the outer background noise portions of film.
7. Ice mapping three-dimensional density figure resolution detection method according to claim 6, it is characterized in that, described RPS
Peak frequency on curve is nyquist frequency, and corresponding minimum resolution is 2 times of the voxel length of three-dimensional density figure.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510495697.9A CN105069797B (en) | 2015-08-13 | 2015-08-13 | Ice mapping three-dimensional density figure resolution detection method based on mask |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510495697.9A CN105069797B (en) | 2015-08-13 | 2015-08-13 | Ice mapping three-dimensional density figure resolution detection method based on mask |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105069797A CN105069797A (en) | 2015-11-18 |
| CN105069797B true CN105069797B (en) | 2017-08-15 |
Family
ID=54499154
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510495697.9A Active CN105069797B (en) | 2015-08-13 | 2015-08-13 | Ice mapping three-dimensional density figure resolution detection method based on mask |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105069797B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106951918B (en) * | 2017-03-01 | 2020-04-28 | 上海交通大学 | Single-particle image clustering method for analysis of cryoelectron microscope |
| CN107092790B (en) * | 2017-04-19 | 2019-07-09 | 上海交通大学 | Electron cryo-microscopy three-dimensional density figure resolution detection method |
| CN110490883A (en) * | 2019-08-22 | 2019-11-22 | 南京信易达计算技术有限公司 | A kind of electron cryo-microscopy data analyzed pattern system and method based on web |
| CN112465067B (en) * | 2020-12-15 | 2022-07-15 | 上海交通大学 | Cryoelectron microscope single-particle image clustering implementation method based on image convolution self-encoder |
| CN112614170B (en) * | 2021-01-08 | 2022-08-05 | 上海交通大学 | Fourier power spectrum-based single particle image registration method for cryoelectron microscope |
| WO2023147706A1 (en) * | 2022-02-07 | 2023-08-10 | 清华大学 | Neural network model training method and resolution estimation method for cryo-electron microscope density map |
| CN114612501B (en) * | 2022-02-07 | 2024-02-13 | 清华大学 | Neural network model training method and frozen electron microscope density map resolution estimation method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102682223A (en) * | 2010-11-30 | 2012-09-19 | 中国科学院计算机网络信息中心 | Structure detection method of protein cryoelectron microscopy density map |
-
2015
- 2015-08-13 CN CN201510495697.9A patent/CN105069797B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102682223A (en) * | 2010-11-30 | 2012-09-19 | 中国科学院计算机网络信息中心 | Structure detection method of protein cryoelectron microscopy density map |
Non-Patent Citations (2)
| Title |
|---|
| 《Fourier shell correlation threshold criteria》;Marin van Heel et al;《Journal of Structural Biology》;20051231;第151卷;第250-262页 * |
| Pawel A. Penczek*.《Three-dimensional spectral signal-to-noise ratio for a class》.《Journal of Structural Biology》.2002,第138卷(第1期),第34-46页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105069797A (en) | 2015-11-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105069797B (en) | Ice mapping three-dimensional density figure resolution detection method based on mask | |
| Huang et al. | T2 mapping from highly undersampled data by reconstruction of principal component coefficient maps using compressed sensing | |
| WO2018099321A1 (en) | Generalized tree sparse-based weighted nuclear norm magnetic resonance imaging reconstruction method | |
| Küstner et al. | MR image reconstruction using a combination of compressed sensing and partial Fourier acquisition: ESPReSSo | |
| CN107230197B (en) | Tropical cyclone objective strength determination method based on satellite cloud image and RVM | |
| CN105825496A (en) | Method for improving image quality of magnetic resonance image dataset, computing device, and computer program | |
| WO2014198239A1 (en) | Ct imaging method and system | |
| CN105787895B (en) | Statistics compressed sensing image reconstructing method based on Hierarchical GMM | |
| CN109472841B (en) | CBCT three-dimensional reconstruction method based on Gaussian mixture/Poisson maximum likelihood function | |
| Yao et al. | Accelerated dynamic MRI reconstruction with total variation and nuclear norm regularization | |
| CN109247939B (en) | Self-adaptive high-undersampled hyperpolarized gas lung dynamic MRI reconstruction method | |
| Usman et al. | Compressive manifold learning: Estimating one‐dimensional respiratory motion directly from undersampled k‐space data | |
| Gan et al. | Deep image reconstruction using unregistered measurements without groundtruth | |
| CN109188327B (en) | Magnetic resonance image fast reconstruction method based on tensor product complex small compact framework | |
| Cooper et al. | Patch based reconstruction of undersampled data (PROUD) for high signal‐to‐noise ratio and high frame rate contrast enhanced liver imaging | |
| Bahri et al. | ESPIRIT-based coil compression for cartesian sampling | |
| CN112651930A (en) | Medical image quality evaluation method and system based on image detail fidelity | |
| Wang et al. | Motion correction and noise removing in lung diffusion-weighted MRI using low-rank decomposition | |
| Wang et al. | Intravoxel incoherent motion magnetic resonance imaging reconstruction from highly under-sampled diffusion-weighted PROPELLER acquisition data via physics-informed residual feedback unrolled network | |
| Xu et al. | Mapping urbanization dynamic of mainland china using dmsp/ols night time light data | |
| Goossens et al. | Objectively measuring signal detectability, contrast, blur and noise in medical images using channelized joint observers | |
| Wang et al. | Objective resolution measurement in single particle reconstructions based on a new spectral signal-to-noise ratio estimation | |
| Li et al. | Swing golden angle–A navigator‐interleaved golden angle trajectory with eddy current suppression–Application in free‐running cardiac MRI | |
| Deng et al. | Motion-compensated orthonormal expansion ℓ1-minimization for reference-driven MRI reconstruction using Augmented Lagrangian methods | |
| Chen et al. | Magnetic resonance image reconstruction via L0-norm minimization |
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 |