CN111145280B - OCT image speckle suppression method - Google Patents
OCT image speckle suppression method Download PDFInfo
- Publication number
- CN111145280B CN111145280B CN201911241433.5A CN201911241433A CN111145280B CN 111145280 B CN111145280 B CN 111145280B CN 201911241433 A CN201911241433 A CN 201911241433A CN 111145280 B CN111145280 B CN 111145280B
- Authority
- CN
- China
- Prior art keywords
- image
- low
- resolution
- resolution image
- oct
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 230000001629 suppression Effects 0.000 title claims abstract description 15
- 230000003044 adaptive effect Effects 0.000 claims abstract description 8
- 238000004364 calculation method Methods 0.000 claims abstract description 5
- 230000006870 function Effects 0.000 claims description 14
- 238000012937 correction Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- 230000001788 irregular Effects 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 5
- 238000005259 measurement Methods 0.000 abstract description 3
- 238000012014 optical coherence tomography Methods 0.000 description 24
- 238000013507 mapping Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 238000006073 displacement reaction Methods 0.000 description 2
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 238000011158 quantitative evaluation Methods 0.000 description 2
- 210000001525 retina Anatomy 0.000 description 2
- 230000002207 retinal effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000011282 treatment Methods 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/003—Reconstruction from projections, e.g. tomography
- G06T11/008—Specific post-processing after tomographic reconstruction, e.g. voxelisation, metal artifact correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
- G06T3/4076—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution using the original low-resolution images to iteratively correct the high-resolution images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- 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
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- 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
- G06T2207/30041—Eye; Retina; Ophthalmic
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Health & Medical Sciences (AREA)
- Image Processing (AREA)
- Eye Examination Apparatus (AREA)
Abstract
The invention belongs to the technical field of image processing, and particularly relates to an OCT image speckle suppression method. Characterized by comprising the following steps: an image registration step of registering a first low-resolution image and a plurality of subsequent second low-resolution images to calculate a motion parameter of the second low-resolution image with respect to the first resolution image; an image reconstruction step of calculating coordinates of the low-resolution image in the high-resolution image based on the calculated motion parameters, and performing pixel difference calculation by using adaptive normalized convolution to reconstruct the high-resolution image; the motion parameters comprise a horizontal offset a, a vertical offset b and a rotation angle theta, and the resolution of the first low-resolution image is the same as that of the second low-resolution image. By increasing the entropy of the image by benefiting from the true measurement of multiple low resolution images, not only can speckle noise from OCT images be reduced, but also the ability to enhance structural characteristics is provided.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to an OCT image speckle suppression method.
Background
The optical coherence tomography technology can rapidly acquire the cross-sectional images of eye biological tissues with micron-scale resolution, is an important tool for retina imaging at present, and provides assistance for clinical ophthalmologists to diagnose diseases. Speckle noise caused by multiple forward and backward scattering of light waves is a major factor causing degradation of OCT image quality, and the presence of speckle noise often masks fine but important morphological details, as well as affecting the performance of automated analysis methods for objective accurate quantification.
The invention patent application of application publication number CN10934569A, application publication date 2019, 2 and 15 discloses a condition-based generation countermeasure network architecture, wherein a mapping model from an OCT image containing speckle noise to a noise-free OCT image is obtained through training, and then the mapping model is adopted to eliminate the speckle noise of the OCT image of the retina. The speckle noise removing model can effectively remove speckle noise and well retain image detail information. The invention patent of the grant publication number CN105976321B, 7 month and 23 days of the grant publication date 2019 discloses a method and a device for reconstructing super-resolution of an optical coherence tomography image. And processing the obtained average image block of each image block through a pre-constructed high-low resolution dictionary pair and a corresponding sparse coefficient mapping equation to obtain a high resolution image of the three-dimensional low resolution OCT image. These methods all rely on model setup, require a large number of virtual mat samples, and have long learning training times.
Disclosure of Invention
In order to improve the quality of OCT images and reduce speckle noise of OCT images so as to improve the detection, diagnosis and treatment effects of various ophthalmic diseases, the invention provides an OCT image speckle suppression method based on super-resolution reconstruction.
An OCT image speckle suppression method, comprising:
an image registration step of registering a first low-resolution image and a plurality of subsequent second low-resolution images to calculate a motion parameter of the second low-resolution image with respect to the first resolution image;
an image reconstruction step of calculating coordinates of the low-resolution image in the high-resolution image based on the calculated motion parameters, and performing pixel difference calculation by using adaptive normalized convolution to reconstruct the high-resolution image;
the motion parameters comprise a horizontal offset a, a vertical offset b and a rotation angle theta, and the resolution of the first low-resolution image is the same as that of the second low-resolution image.
In the above-described solution, the image registration is intended to estimate motion parameters between low resolution images, and the image reconstruction is intended to reconstruct a high resolution image by combining the registered images using the estimated motion parameters. They increase the entropy of the image by benefiting from the true measurement of multiple low resolution images, not only to reduce speckle noise from OCT images, but also to have the ability to enhance structural properties.
Preferably, the image registration step includes:
step a1, performing Gaussian filtering and subsampling on a low-resolution image with the size of M multiplied by N to obtain a lower-layer image with the size of M/2 multiplied by N/2;
a2, calculating a lower layer image motion parameter of a lower layer image corresponding to the second low resolution image relative to a lower layer image corresponding to the first low resolution image;
step a3, executing steps a1-a2 on the lower layer image obtained in the step a1, and repeating k times;
and a4, performing iterative correction based on the k calculated motion parameters of the lower image to obtain the motion parameters of the second low-resolution image relative to the first resolution image.
Preferably, k in the step a3 is equal to 1.
Preferably, in the step a2, the motion parameter of the pixel point (x, y) in the second low resolution image g relative to the pixel point (x, y) in the first low resolution image f is calculated by a linear equation
Solving for, wherein 。
Preferably, the step a4 iterates the corrected error function:
。
preferably, the image reconstruction step includes: step b-1, calculating the coordinates of the low-resolution image in the high-resolution grid according to the motion parameters estimated in the step a; and b-2, fusing irregular sampling data obtained from different low-resolution images by adopting a structure self-adaptive normalized convolution reconstruction method.
Preferably, in the step b-2, the local signal from the projection is approximated to a set of polynomial basis functions based on a structure-adaptive normalized convolution method.
Preferably, in the step b-2, the projection coefficient of the pixel s0= (x 0, y 0) on the polynomial basis function is set.
Preferably, the local structure information is obtained in the step b-2 by using a gradient structure tensor method.
Preferably, in step b-2, the certainty of each irregular sample is divided into its four nearest HR grid points in a bilinear weighted manner, and the certainty parameters of the samples on all the grids are accumulated on the high resolution grid to form a density image.
The invention has the following beneficial effects:
(1) Structural characteristics are enhanced while reducing OCT image speckle noise.
(2) High resolution images can be reconstructed and the poor signal to noise ratio of OCT images improved, with sub-pixel shifting to recover lost signals from multiple images to increase the entropy of the images.
(3) The MSE can be reduced while increasing PSNR and MSSIM, helping to expand subsequent OCT image processing, such as layer segmentation and lesion detection.
Drawings
Fig. 1 is an original image, a gold standard image and an exemplary image after reconstruction according to the present invention.
Fig. 2 is an experimental diagram of an original image, a gold standard image, and five noise reduction methods, and an example image after SR reconstruction according to the present invention.
Detailed Description
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. Unless defined otherwise, all terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It will be further understood that the terms used in the specification should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and this disclosure. The present disclosure is to be considered as an exemplification of the invention and is not intended to limit the invention to the specific embodiments illustrated.
Example 1
The invention relates to a step flow chart of an OCT image speckle suppression method, wherein the original image size is 992x512, and the resolution of the image is 4 times that of the original image, namely, the image with 1984x 1024.
The method of the invention comprises the following steps:
a. an image registration step of registering a first low resolution image and a plurality of subsequent second low resolution images to calculate a motion parameter of the second low resolution image with respect to the first resolution image.
b. And an image reconstruction step of calculating coordinates of the low-resolution image in the high-resolution image based on the calculated motion parameters, and performing pixel difference calculation by using adaptive normalized convolution to reconstruct the high-resolution image.
In this example, OCT images of volumes of 20 normal patients were selected, each volume containing 254 images as the dataset of the present invention.
In the image registration step, one OCT image (i.e., a first low resolution image) is read in and registered with three subsequent OCT images (i.e., a second low resolution image). In the registration stage, in order to increase speed and robustness, coarse to fine image structures are used. The registration step flow comprises the following steps:
step a-1, subjecting the low resolution image of size M x N to Gaussian filtering and subsampling to obtain a lower layer image of size M/2 x N/2.
A2, calculating a lower layer image motion parameter of a lower layer image corresponding to the second low resolution image relative to a lower layer image corresponding to the first low resolution image;
step a3, executing steps a1-a2 on the lower layer image obtained in step a1, and repeating k times. In particular, in this embodiment, k is 1 to use a three-layer gaussian pyramid. By repeating steps a1-a2, two images of M/4 XN/4 size are obtained.
And a4, performing iterative correction based on the k calculated motion parameters of the lower image to obtain the motion parameters of the second low-resolution image relative to the first low-resolution image. The iterative correction operation is as follows: calculating the motion parameters of each second low-resolution image relative to the first low-resolution image from the coarsest original OCT image, then carrying out rotation and translation correction on the second-layer image according to the motion parameters, obtaining a new second-layer image by interpolation, recalculating the new motion parameters, and the like, and finally calculating the motion parameters of the original image with high precision. Wherein the motion parameters are calculated as follows:
the relationship between the horizontal offset a, the vertical offset b, and the rotation angle θ between the first low-resolution image f and the second low-resolution image g can be written as:
the sin (θ) and cos (θ) are expanded by taylor expansion to give the following equations:
expanding f to the first term in its taylor expansion gives the first order equation as follows:
the error E that occurs after image g and image f are translated by a, b and θ is rotated can be approximated as:
where the summation is in the overlap region of f and g. The minimum value of E (a, b, θ) can be obtained by calculating its reciprocal relative to a and b and setting θ to zero. To minimize the difference between g and f distorted by (a, b, θ), it can be obtained by solving the following equations for a, b and θ:
wherein, the liquid crystal display device comprises a liquid crystal display device,。
by solving this set of linear equations, the motion parameters a, b, θ can be calculated.
Through the flow, the horizontal displacement and the vertical displacement of the image can be estimated, and the coordinates of the matching points can be found through parameters.
In step b a structure-adaptive normalized convolution (SANC) reconstruction method is used for fusing irregularly sampled data obtained from different low resolution images. Coordinates of the low resolution image in the high resolution grid are calculated from the estimated motion parameters resulting from the registration. And performing pixel difference calculation by using the self-adaptive normalized convolution, and reconstructing a high-resolution image. The method comprises the following steps:
and b-1, calculating the coordinates of the low-resolution image in the high-resolution grid according to the motion parameters estimated in the step a. The magnification factor in this embodiment is 2, and the size of the high resolution grid is 992×512, which is 2 times larger than the original image, i.e., 1984×1024, which is 4 times larger than the original image.
And b-2, modeling the neighborhood of the local signal by using the local signal as a center through a polynomial surface basis function. The adaptive applicability function is an anisotropic gaussian function for approximating the local signal from the projection onto a set of polynomial basis functions giving different weights to all data points in the image neighborhood. In this embodiment, the projection coefficient of the pixel s0= (x 0, y 0) on the polynomial basis function. And performing first-order robustness processing by using a deterministic function for signal fusion. The certainty of the signal needs to be known in advance before interpolation is performed by using the NC, so that abnormal signals can be prevented from participating in reconstruction. The projection coefficient P is solved using a least squares method. The structural adaptive function has a decisive role for interpolation, and the adaptive function is an anisotropic gaussian function which reduces blurring. To obtain parameters therein, an estimate of the local image structure and scale is required. In order to obtain an adaptive kernel at the reconstructed pixel, knowledge of the local image structure around the pixel has to be achieved, and the local structure information is obtained using a GTS (gradient structure tensor) method. In addition, the local sample density is known, the certainty of each irregular sample is divided into four nearest HR grid points in a bilinear weighting mode, the certainty parameters of the samples on all the grids are accumulated on the HR grid to form a density image, the scale space of the density image is constructed through quick separation and recursive filtering, and the estimation filtering result is equal to the Gaussian scale of C.
Referring to fig. 1, it can be observed that the proposed method significantly enhances most of the disk tissues (retinal layers), not only suppresses speckle noise, but also preserves structural details such as retinal layer boundaries. Referring to fig. 2, which is the result of applying different methods of speckle reduction to three example images with representative image blocks, the present invention and five other advanced OCT image speckle reduction methods are used to compare: NLM, NCDF, K-SVD, ASR and BM4D. Most filters, e.g., NLM, NCDF, K-SVD and BM4D, can be seen to reduce noise in the image, but without introducing additional information. At the same time, these filters further smooth the local features, thus creating the impression of blurring of the image. ASR achieves relatively better performance than the four methods described above because it estimates global and local motion of the registered OCT scans. In the experiment, qualitative and quantitative methods were used to evaluate the performance of the algorithm, respectively.
Because visual inspection is difficult to fully prove the superiority of the invention over other denoising methods, the invention adopts qualitative and quantitative evaluation results, and more objective quantitative evaluation is carried out. See the following table:
the method provided by the invention is superior to other methods, and is characterized by lower MSE, higher PSNR and MSSIM: the values were 0.040, 23.76 and 0.772, respectively. It can be seen that the super resolution method proposed by the invention can increase the entropy of an image by using the true measurements of multiple images and can retrieve missing local features from multiple images using sub-pixel shifting.
While embodiments of the present invention have been described, various modifications and adaptations may be made by one of ordinary skill in the art within the scope of the following claims.
Claims (9)
1. An OCT image speckle suppression method, comprising:
an image registration step of registering a first low resolution image and a plurality of subsequent second low resolution images to calculate a motion parameter of the second low resolution image with respect to the first low resolution image;
an image reconstruction step of calculating coordinates of the low-resolution image in the high-resolution image based on the calculated motion parameters, and performing pixel difference calculation by using adaptive normalized convolution to reconstruct the high-resolution image;
wherein the motion parameters comprise a horizontal offset a, a vertical offset b and a rotation angle theta, and the resolution of the first low-resolution image is the same as that of the second low-resolution image;
the motion parameter of the pixel point (x, y) in the second low resolution image g relative to the pixel point (x, y) in the first low resolution image f is calculated by a linear equation
Solving for, wherein
。
2. The OCT image speckle suppression method of claim 1, wherein the image registration step comprises:
step a1, gaussian filtering and subsampling the low-resolution image of size M×N to obtainA lower image of a size;
a2, calculating a lower layer image motion parameter of a lower layer image corresponding to the second low resolution image relative to a lower layer image corresponding to the first low resolution image;
step a3, executing steps a1-a2 on the lower layer image obtained in the step a1, and repeating k times;
and a4, performing iterative correction based on the k calculated motion parameters of the lower image to obtain the motion parameters of the second low-resolution image relative to the first low-resolution image.
3. The OCT image speckle suppression method of claim 2, wherein:
in the step a3, k is equal to 1.
4. A method of speckle reduction of OCT images according to claim 3, wherein the step a4 iterates the correction of the error function:
。
5. the OCT image speckle suppression method according to any one of claims 1 to 3, wherein the image reconstruction step includes:
step b-1, calculating the coordinates of the low-resolution image in the high-resolution grid according to the motion parameters estimated in the step a;
and b-2, fusing irregular sampling data obtained from different low-resolution images by adopting a structure self-adaptive normalized convolution reconstruction method.
6. The OCT image speckle suppression method of claim 5, wherein:
in step b-2, the local signal from the projection is approximated onto a set of polynomial basis functions based on a structure-adaptive normalized convolution method.
7. The OCT image speckle suppression method of claim 6, wherein:
the pixel point S in the step b-2 0 =（x 0 ，y 0 ) Projection coefficients on the polynomial basis functionP（S 0 ）=[P 0 P 1 …P m ] T (S 0 )。
8. The OCT image speckle suppression method of claim 6, wherein:
and (3) obtaining local structure information by using a gradient structure tensor method in the step b-2.
9. The OCT image speckle suppression method of claim 5, wherein:
in step b-2, the certainty of each irregular sample is divided into four nearest HR grid points in a bilinear weighting manner, and the certainty parameters of the samples on all grids are accumulated on a high-resolution grid to form a density image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911241433.5A CN111145280B (en) | 2019-12-06 | 2019-12-06 | OCT image speckle suppression method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911241433.5A CN111145280B (en) | 2019-12-06 | 2019-12-06 | OCT image speckle suppression method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111145280A CN111145280A (en) | 2020-05-12 |
| CN111145280B true CN111145280B (en) | 2023-09-15 |
Family
ID=70517763
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911241433.5A Active CN111145280B (en) | 2019-12-06 | 2019-12-06 | OCT image speckle suppression method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111145280B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112419157B (en) * | 2020-11-30 | 2024-04-02 | 浙江凌迪数字科技有限公司 | Cloth super-resolution method based on generation of countermeasure network |
| CN112785518B (en) * | 2021-01-11 | 2022-08-30 | 飞燕航空遥感技术有限公司 | Method and system for correcting circular dark spots in image |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101226634A (en) * | 2008-01-30 | 2008-07-23 | 哈尔滨工业大学 | Method for reestablishment of single frame image quick super-resolution based on nucleus regression |
| CN103136734A (en) * | 2013-02-27 | 2013-06-05 | 北京工业大学 | Restraining method on edge Halo effects during process of resetting projections onto convex sets (POCS) super-resolution image |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101375490B1 (en) * | 2012-04-27 | 2014-03-17 | 가천대학교 산학협력단 | Method for obtain PET images with super resolution using a movement of PET device |
| US10366480B2 (en) * | 2016-07-01 | 2019-07-30 | Analytical Mechanics Associates, Inc. | Super-resolution systems and methods |
-
2019
- 2019-12-06 CN CN201911241433.5A patent/CN111145280B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101226634A (en) * | 2008-01-30 | 2008-07-23 | 哈尔滨工业大学 | Method for reestablishment of single frame image quick super-resolution based on nucleus regression |
| CN103136734A (en) * | 2013-02-27 | 2013-06-05 | 北京工业大学 | Restraining method on edge Halo effects during process of resetting projections onto convex sets (POCS) super-resolution image |
Non-Patent Citations (1)
| Title |
|---|
| 孟庆武等.《单帧低分辨率图像的SFMAP高分辨率重建算法》.《计算机科学》.2004,第147-150. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111145280A (en) | 2020-05-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111081354B (en) | System and method for denoising medical images through deep learning network | |
| Muckley et al. | Training a neural network for Gibbs and noise removal in diffusion MRI | |
| CN108305230A (en) | A kind of blurred picture integrated conduct method and system | |
| JP6448313B2 (en) | Method for processing image data representing a three-dimensional volume | |
| CN109961411B (en) | Non-downsampling shear wave transformation medical CT image denoising method | |
| US20230360179A1 (en) | Machine learning processing of contiguous slice image data | |
| CN111145280B (en) | OCT image speckle suppression method | |
| CN108830791B (en) | Image super-resolution method based on self sample and sparse representation | |
| Rabbani et al. | Optical coherence tomography noise reduction using anisotropic local bivariate Gaussian mixture prior in 3D complex wavelet domain | |
| CN111815766A (en) | Processing method and system for reconstructing blood vessel three-dimensional model based on 2D-DSA image | |
| CN112037304B (en) | Two-stage edge enhancement QSM reconstruction method based on SWI phase image | |
| Ahn et al. | Adaptive template filtering for signal-to-noise ratio enhancement in magnetic resonance imaging | |
| CN106204475B (en) | Based on field map to the corrected method of geometric distortion artifact in fMRI | |
| CN114240797A (en) | OCT image denoising method, device, equipment and medium | |
| CN113962887A (en) | Training method and denoising method for two-dimensional cryoelectron microscope image denoising model | |
| CN111563843B (en) | Image super-resolution reconstruction method, system and related device | |
| CN114332283A (en) | Training method based on double-domain neural network and photoacoustic image reconstruction method | |
| Seetharaman et al. | Retracted: A novel approach in hybrid median filtering for denoising medical images | |
| CN117391955A (en) | Convex set projection super-resolution reconstruction method based on multi-frame optical coherence tomography | |
| CN116645283A (en) | Low-dose CT image denoising method based on self-supervision perceptual loss multi-scale convolutional neural network | |
| CN114565711A (en) | Heart image reconstruction method and system based on deep learning | |
| CN114187181A (en) | Double-path lung CT image super-resolution method based on residual information refining | |
| CN104504657B (en) | Method and device for de-noising magnetic resonance diffusion tensor | |
| CN117611453A (en) | Nuclear magnetic resonance image super-resolution recovery method and model construction method | |
| Fu et al. | ADGAN: An asymmetric despeckling generative adversarial network for unpaired OCT image speckle noise reduction |
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 |