CN114972565A - Image noise reduction method and device, electronic equipment and medium - Google Patents
Image noise reduction method and device, electronic equipment and medium Download PDFInfo
- Publication number
- CN114972565A CN114972565A CN202210579138.6A CN202210579138A CN114972565A CN 114972565 A CN114972565 A CN 114972565A CN 202210579138 A CN202210579138 A CN 202210579138A CN 114972565 A CN114972565 A CN 114972565A
- Authority
- CN
- China
- Prior art keywords
- time point
- image data
- echo time
- noise
- amplitude image
- 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.)
- Pending
Links
- 230000009467 reduction Effects 0.000 title claims abstract description 66
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 40
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 27
- 210000003007 myelin sheath Anatomy 0.000 claims abstract description 24
- 210000005013 brain tissue Anatomy 0.000 claims abstract description 13
- 238000004364 calculation method Methods 0.000 claims abstract description 11
- 238000009826 distribution Methods 0.000 claims description 41
- 238000004891 communication Methods 0.000 claims description 18
- 238000012549 training Methods 0.000 claims description 16
- 238000005259 measurement Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 14
- 241000209094 Oryza Species 0.000 claims description 11
- 235000007164 Oryza sativa Nutrition 0.000 claims description 11
- 235000009566 rice Nutrition 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 7
- 238000003860 storage Methods 0.000 claims description 7
- 230000000694 effects Effects 0.000 abstract description 4
- 102000006386 Myelin Proteins Human genes 0.000 description 19
- 108010083674 Myelin Proteins Proteins 0.000 description 19
- 210000005012 myelin Anatomy 0.000 description 19
- 239000012223 aqueous fraction Substances 0.000 description 12
- 238000010586 diagram Methods 0.000 description 11
- 238000012545 processing Methods 0.000 description 10
- 238000013528 artificial neural network Methods 0.000 description 7
- 230000005284 excitation Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000009792 diffusion process Methods 0.000 description 4
- 230000010355 oscillation Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000011946 reduction process Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000010832 independent-sample T-test Methods 0.000 description 2
- 230000005415 magnetization Effects 0.000 description 2
- 238000001454 recorded image Methods 0.000 description 2
- 210000001519 tissue Anatomy 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000012307 MRI technique Methods 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000003376 axonal effect Effects 0.000 description 1
- 238000002592 echocardiography Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000002595 magnetic resonance imaging Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000023105 myelination Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30008—Bone
- G06T2207/30012—Spine; Backbone
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Processing (AREA)
Abstract
The application provides a method and a device for reducing noise of an image, electronic equipment and a medium. The method is based on a preset multi-echo gradient echo sequence, myelin sheath water image data of human brain tissues are collected, and amplitude image data collected at each echo time point is obtained through calculation; determining a first noise level metric value of the multi-echo gradient echo sequence based on a noise value in amplitude image data acquired at a last echo time point and an amplitude of the amplitude image data corresponding to a first echo time point; based on an error value between the first noise level metric value and the second noise level metric value, a corresponding second noise level metric value is obtained. The amplitude image data of each echo time point passes through a preset noise reduction convolutional neural network to obtain the noise value of the amplitude image data of each echo time point, so that the noise reduction is performed on the amplitude image data collected by the corresponding echo time point to obtain the noise-free amplitude image data of each echo time point, and the noise reduction effect is improved.
Description
Technical Field
The present application relates to the field of image processing technologies, and in particular, to a method and an apparatus for reducing noise of an image, an electronic device, and a medium.
Background
Myelination imaging is a magnetic resonance imaging technique and is of great interest for diagnosing lesions that occur in the myelin sheath and for quantifying myelin integrity. The specific way of acquiring the T2 signal by using the multi-echo gradient echo sequence (mcre) is as follows: a series of alternating polarity readout gradients are used immediately after the small angle excitation pulse and the phase encoding gradient. A small angle excitation pulse is used to minimize the First Echo Time (TE 1). To further shorten the TE1 and echo interval, signals (echoes) are acquired during the readout gradient plateau, as well as on both the rising and falling edges. There are several advantages due to this approach, including low specific absorption rate of the sequence, short echo separation, 3D excitation and minimum phase radio frequency excitation. It is worth mentioning that the echo time of the first echo can be reduced by using the minimum phase radio frequency excitation mode, so that the accuracy of sampling the real information of the myelin sheath water can be improved.
The images acquired by the multi-echo gradient echo sequence for acquiring myelin sheath water image data in each echo time conform to Rician distribution. T2 is the transverse magnetization vector signal composed of amplitude values corresponding to voxels of the image corresponding to each echo time.
At present, the method for reducing the noise of an image acquired by mGRE mainly comprises the following steps: anisotropic diffusion filters and non-local mean filters. These two filters need to determine noise reduction parameters (noise variance) before use. The noise reduction parameter is typically the noise value of the first echo image in the mcre sequence. Then, each echo image of mcre is denoised by applying an anisotropic diffusion filter or a non-local mean filter.
However, for the images obtained by the mcre sequence, the noise parameters applied by the anisotropic diffusion filter and the non-local mean filter only consider the noise value of the first echo image, and the noise values of the images acquired due to different echo time points in the mcre sequence are not the same. Therefore, only selecting the noise reduction parameters based on the first echo image has poor noise reduction effect on the image, thereby causing inaccurate calculation of the myelin water fraction.
Disclosure of Invention
An object of the embodiments of the present application is to provide an image denoising method, apparatus, electronic device and medium, so as to solve the above problems in the prior art, improve a denoising effect of an image, and thus improve accuracy of myelin sheath water fraction calculation.
In a first aspect, a method for reducing noise of an image is provided, and the method may include:
acquiring myelin sheath water image data of voxels in human brain tissue based on a preset multi-echo gradient echo sequence to obtain amplitude image data acquired at each echo time point; the preset multi-echo gradient echo sequence comprises a plurality of echo time points;
determining a first noise level metric value of the multi-echo gradient echo sequence based on a noise value in amplitude image data acquired at a last echo time point and an amplitude of the amplitude image data corresponding to a first echo time point;
obtaining error values between the first noise level metric values and the stored second noise level metric values respectively; the second noise level metric is determined after adding a new noise value based on the initial recorded amplitude image data containing the noise value;
if the target error value is smaller than a preset error threshold value, acquiring a target second noise level metric value corresponding to the target error value; the target error value is one of the obtained error values;
inputting the amplitude image data acquired at each echo time point as input data into a preset noise reduction convolutional neural network corresponding to the target second noise level metric value to obtain a noise value of the amplitude image data of each echo time point output by the noise reduction convolutional neural network; the noise reduction convolutional neural network is obtained by training a convolutional neural network based on residual learning based on noise distribution corresponding to different second noise level measurement values;
and denoising the amplitude image data acquired at the corresponding echo time point by adopting the noise value corresponding to each echo time point to obtain noiseless amplitude image data of each echo time point.
In an optional implementation, before determining the first noise level metric of the multi-echo gradient echo sequence based on the amplitude of the amplitude image data corresponding to the first echo time point and the noise value in the amplitude image data acquired at the last echo time point, the method further includes:
determining the variance of background data in the amplitude image data of the last echo time point as a noise value in the amplitude image data of the last echo time point;
and determining the average value of the amplitudes in the amplitude image data of the first echo time point as the amplitude of the amplitude image data of the first echo time point.
In an optional implementation, determining a first noise level metric value of the multi-echo gradient echo sequence based on an amplitude of amplitude image data corresponding to a first echo time point and a noise value in amplitude image data acquired at a last echo time point includes:
and determining a first noise level metric of the multi-echo gradient echo sequence according to a result of dividing a noise value in the amplitude image data acquired at the last echo time point by the amplitude of the amplitude image data at the first echo time point.
In an alternative implementation, the training process of the noise reduction convolutional neural network includes:
based on a preset multi-echo gradient echo sequence, myelin sheath water image data of human brain tissues are collected, and amplitude image data of myelin sheath water images collected at all echo time points are obtained through calculation;
adding different preset noise values into the initial recorded amplitude image data of the corresponding echo time point according to the first echo time point and the last echo time point, acquiring different new recorded amplitude image data corresponding to the echo time point and second noise level measurement values corresponding to the different new recorded amplitude image data, and storing the second noise level measurement values;
determining noise distributions corresponding to different second noise level metrics based on the different newly recorded amplitude image data and corresponding initially recorded amplitude image data;
and training the convolutional neural network based on residual learning by adopting the noise distribution corresponding to the different second noise level metric values to obtain the noise reduction convolutional neural network corresponding to the different second noise level metric values.
In an optional implementation, adding different preset noise values to the initial recorded amplitude image data of the corresponding echo time point for the first echo time point and the last echo time point, and acquiring different new recorded amplitude image data corresponding to the echo time point and a second noise level metric value corresponding to the different new recorded image data, includes:
increasing the variance of background data in the initial recorded amplitude image data corresponding to the first echo time point and the last echo time point by the preset different noise values to obtain new recorded amplitude image data corresponding to the first echo time point and the last echo time point; wherein the newly recorded amplitude image data corresponding to the first echo time point and the last echo time point contains a new noise value;
and determining a second noise level metric of the multi-echo gradient echo sequence based on a new noise value in the newly recorded amplitude image data of the last echo time point and the amplitude of the newly recorded amplitude image data of the first echo time point.
In an optional implementation, the amplitude image data of each echo time point are subjected to a rice distribution;
increasing the variance of background data in the initial recorded amplitude image data sample corresponding to the first echo time point and the last echo time point by the preset different noise values to obtain a new recorded amplitude image data sample corresponding to the first echo time point and the last echo time point, including:
and increasing the variance in the Rice distribution obeyed by the initial recorded amplitude image data samples corresponding to the first echo time point and the last echo time point by the preset different noise values to obtain new recorded amplitude image data samples corresponding to the first echo time point and the last echo time point.
In an optional implementation, the noise reduction of the amplitude image data acquired at the corresponding echo time point by using the noise value corresponding to each echo time point to obtain the noiseless amplitude image data at each echo time point includes:
and subtracting the noise value corresponding to the corresponding echo time point from the amplitude image data of each echo time point to obtain the noise-free amplitude image data of each echo time point.
In a second aspect, there is provided an apparatus for reducing noise of an image, the apparatus may include:
the acquisition unit is used for acquiring myelin sheath water image data of voxels in human brain tissue based on a preset multi-echo gradient echo sequence to obtain amplitude image data acquired at each echo time point; the preset multi-echo gradient echo sequence comprises a plurality of echo time points;
the determining unit is used for determining a first noise level metric value of the multi-echo gradient echo sequence based on the amplitude of the amplitude image data corresponding to the first echo time point and the noise value in the amplitude image data acquired at the last echo time point;
an obtaining unit, configured to obtain error values between the first noise level metric values and the stored second noise level metric values; the second noise level metric is determined after adding a new noise value based on the initial recorded amplitude image data containing the noise value;
if the target error value is smaller than a preset error threshold value, acquiring a target second noise level metric value corresponding to the target error value; the target error value is one of the obtained error values;
inputting the amplitude image data acquired at each echo time point as input data into a preset noise reduction convolution neural network corresponding to the target second noise level metric value to obtain a noise value of the amplitude image data of each echo time point output by the noise reduction convolution neural network; the noise reduction convolutional neural network is obtained by training a convolutional neural network based on residual learning based on noise distribution corresponding to different second noise level measurement values;
and the noise reduction unit is used for reducing noise of the amplitude image data acquired at the corresponding echo time point by adopting the noise value corresponding to each echo time point to obtain noiseless amplitude image data of each echo time point.
In a third aspect, an electronic device is provided, which includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete communication with each other through the communication bus;
a memory for storing a computer program;
a processor adapted to perform the method steps of any of the above first aspects when executing a program stored in the memory.
In a fourth aspect, a computer-readable storage medium is provided, having stored therein a computer program which, when executed by a processor, performs the method steps of any of the above first aspects.
The noise reduction method of the image provided by the application considers the noise influence of the amplitude image data acquired by the mGRE sequence at different echo time points, the noise in the amplitude image data acquired by the first echo time point is not taken as reference, the noise level of the whole mGRE sequence is calculated, and the noise in the amplitude image data corresponding to each echo time point is identified by the trained noise reduction convolution neural network, so that the oscillation amplitude problem of the whole T2 attenuation curve is considered to the maximum extent, the noise reduction accuracy is improved, and the myelin sheath water fraction calculation accuracy is also improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
Fig. 1 is a schematic flowchart of a method for reducing noise of an image according to an embodiment of the present disclosure;
fig. 2 is a schematic diagram of a T2 attenuation curve provided in an embodiment of the present application;
FIG. 3 is a graphical illustration of a myelin sheath water content value distribution provided by an embodiment of the present application;
fig. 4 is a schematic structural diagram of an image noise reduction apparatus according to an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without any creative effort belong to the protection scope of the present application.
For convenience of understanding, terms referred to in the embodiments of the present application are explained below:
voxel (voxel) is an abbreviation of volume element (volume pixel). The method is the minimum unit of digital data on three-dimensional space segmentation, and is used in the fields of three-dimensional imaging, scientific data, medical images and the like. Some real three-dimensional displays use voxels to describe resolution, such as displays that can display 512 x 512 voxels. Voxels do not contain data of locations in space (i.e. their coordinates), but can be extrapolated from their location relative to other voxels, i.e. their location in the data structure that constitutes a single volumetric image.
Neural myelin content is generally achieved by estimating the amount of myelin water in tissue, since myelin contains water with different specific magnetic resonance properties than other surrounding tissues. Myelin image data typically contains more noise, the snr is lower, and the estimation error for direct estimation of myelin content is usually larger.
The noise level in images (or "amplitude image data") acquired by a magnetic resonance imaging apparatus based on a multi-echo gradient echo sequence (mcre) is determined by the number of receiving coils, and the noise acquired by a single-channel coil is generally regarded as rice (Rician) noise (more specifically, in low-signal regions, the noise is approximately in a rice distribution; in high-signal regions, the noise is approximately in a gaussian distribution, but in general, the noise is in a rice distribution), i.e., the noise in the image is regarded as complying with the Rician distribution, while the noise acquired by a multi-channel coil is generally complying with a non-central chi-square distribution. Due to the instability of the non-central chi-square distribution, the images acquired by the multi-channel coil are generally considered to have noise that still follows Rician distribution without special specification.
Images containing noise are generally abstracted into the following form:
y＝x+e (1)
where x and y represent the noiseless image data and the noisy image data, respectively, and e represents the noise value. The noise reduction process can be expressed as finding a specific transformation T (·) so that the noisy image becomes a noiseless image after transformation, i.e., T (y) x, or using the residual idea so that T (y) e, and thus x (y) y.
The distribution characteristics of Rician noise are as follows:
wherein N1 and N2 both conform to a normal distribution, i.e., N1-N (0, σ) 1 2 )，N2～N(0,σ 2 2 ). Therefore, when the conventional filter is used for noise reduction, noise variance detection needs to be performed on the image to be subjected to noise reduction.
The method for processing the service data provided by the embodiment of the application can be applied to a server and can also be applied to a terminal. The server may be an application server or a cloud server; in order to ensure the accuracy of the detection, the Terminal may be a Mobile phone with strong computing power, a smart phone, a notebook computer, a digital broadcast receiver, a Personal Digital Assistant (PDA), a User Equipment (UE) such as a tablet computer (PAD), a handheld device, a vehicle-mounted device, a wearable device, a computing device or other processing device connected to a wireless modem, a Mobile Station (MS), a Mobile Terminal (Mobile Terminal), and the like.
In the noise reduction process of the existing anisotropic diffusion filter and the non-local mean filter, only the noise reduction parameters based on the first echo image are selected, and the intrinsic correlation property between images corresponding to an mGRE sequence is not applied, so that the transverse magnetization vector (T2) attenuation curve formed by the amplitude values corresponding to the voxels of the image corresponding to each echo time point after noise reduction is not smooth, and the myelin water fraction calculation is deviated. The noise influence of the amplitude image data acquired at different echo time points of the mGRE sequence is considered, the noise in the amplitude image data acquired at the first echo time point is not taken as reference, the noise level of the whole mGRE sequence is calculated, and the noise in the amplitude image data corresponding to each echo time point is identified by the trained noise reduction convolution neural network, so that the oscillation amplitude problem of the whole T2 attenuation curve is considered to the maximum extent, and the noise reduction accuracy is improved. The attenuation curve of T2 is the intrinsic relation between the amplitude image data corresponding to each echo time point in the whole sequence.
The preferred embodiments of the present application will be described in conjunction with the drawings of the specification, it should be understood that the preferred embodiments described herein are only for illustrating and explaining the present application, and are not intended to limit the present application, and the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Fig. 1 is a schematic flowchart of a method for reducing noise of an image according to an embodiment of the present disclosure. As shown in fig. 1, the method may include:
step S110, based on a preset multi-echo gradient echo sequence, myelin sheath water image data of voxels in human brain tissue are collected, and amplitude image data collected at each echo time point is obtained.
The preset multi-echo gradient echo sequence comprises a plurality of echo time points, and each echo time point acquires myelin sheath water image data in human brain tissue to obtain amplitude image data acquired at each echo time point.
Because the acquired amplitude image data contains noise values, and the images acquired by the multi-echo gradient echo sequence in each echo time all accord with the Rician distribution, the amplitude image data obeys the Rician distribution, namely, the following conditions are met:
where x and y represent a noiseless image and a noisy image, respectively, and the noise values follow normal distributions of N1 and N2.
Step S120, determining a first noise level metric value of the multi-echo gradient echo sequence based on the noise value in the amplitude image data acquired at the last echo time point and the amplitude of the amplitude image data corresponding to the first echo time point.
In a specific implementation, the variance of the background data in the amplitude image data of the last echo time point may be determined as a noise value in the amplitude image data of the last echo time point; and determining the average value of all amplitudes in the amplitude image data of the first echo time point as the amplitude of the amplitude image data of the first echo time point.
It should be noted that, an average value of noise values in the amplitude image data corresponding to each echo time point may also be obtained, and the average value is used as a noise value of the amplitude image data of the last echo time point; or, the amplitude corresponding to the median of each amplitude in the amplitude image data at the first echo time point is determined as the amplitude of the amplitude image data at the first echo time point, the determining manner is only one determining manner, and the specific situation may be selected according to the actual service situation, which is not limited herein.
And then, determining a first noise level metric value of the multi-echo gradient echo sequence, namely a current noise level metric value, according to the result of dividing the noise value in the amplitude image data acquired at the last echo time point by the amplitude of the amplitude image data acquired at the first echo time point.
It should be noted that, other operations may also be performed on the noise value corresponding to the last echo time point and the amplitude value corresponding to the first echo time point, for example, a value obtained by dividing the noise value corresponding to the last echo time point by the sum of the noise value and the amplitude value corresponding to the first echo time point is determined as the current noise level metric value of the multi-echo gradient echo sequence, which is not limited herein.
It will be appreciated that the current noise level metric here can be analogized to the signal-to-noise ratio of the image: the ratio of the signal to the noise indicates the quality of the image, so that the current noise level metric value defines the ratio of the noise value in the amplitude image data acquired at the last echo time point to the amplitude of the amplitude image data at the first echo time point to represent the noise level of the whole multi-echo gradient echo sequence.
Step S130, based on the first noise level metric value and each stored second noise level metric value, a target second noise level metric value corresponding to the target error value is obtained.
In specific implementation, error values between the first noise level metric values and the stored second noise level metric values are obtained; wherein the second noise level measure is determined after adding a new noise value based on the initially recorded amplitude image data comprising the noise value. Wherein, the obtaining process of the second noise level metric value comprises the following steps:
based on a preset multi-echo gradient echo sequence, myelin sheath water image data of human brain tissues are collected, and amplitude image data of myelin sheath water images collected corresponding to all echo time points are obtained through calculation;
and adding preset different noise values to the variance of background data in the initial recorded amplitude image data corresponding to the first echo time point and the last echo time point in the initial recorded amplitude image data containing the noise values to obtain new recorded amplitude image data corresponding to the first echo time point and the last echo time point. Specifically, because the amplitude image data of each echo time point obeys rice distribution, the variance in the rice distribution obeyed by the initial recorded amplitude image data samples corresponding to the first echo time point and the last echo time point is increased by preset different noise values, and new recorded amplitude image data samples corresponding to the first echo time point and the last echo time point are obtained. That is, in combination with formula (2), it can be seen that only the value of at least one variance in the formula needs to be changed, so that the new recorded amplitude image data corresponding to the first echo time point and the last echo time point contains a new noise value.
And then, determining a second noise level metric value of the multi-echo gradient echo sequence based on a new noise value in the newly recorded amplitude image data sample of the last echo time point and the amplitude of the newly recorded amplitude image data of the first echo time point, and storing the second noise level metric value.
Further, detecting whether an error value smaller than a preset error threshold exists in the obtained error values;
and if the target error value is smaller than the preset error threshold value, acquiring a target second noise level metric value corresponding to the target error value.
Step S140, using the amplitude image data collected at each echo time point as input data, and inputting a preset residual learning convolutional neural network corresponding to the target second noise level metric value to obtain a noise value of the amplitude image data at each echo time point output by the noise reduction convolutional neural network.
The convolutional neural network for residual learning is obtained by training based on the noise distribution corresponding to different second noise level measurement values.
Before executing the step, a noise reduction convolutional neural network needs to be trained, and the training process of the noise reduction convolutional neural network comprises the following steps:
acquiring a second noise level metric corresponding to the initial recorded amplitude image data acquired in the step S130, and new recorded amplitude image data after adding different preset noise values;
determining noise distributions corresponding to different second noise level metrics based on different newly recorded image data and corresponding initial recorded amplitude image data samples; for example, the newly recorded amplitude image data is a data sample of the image SS, and the corresponding initially recorded amplitude image data sample is a data sample of the image S, and the data sample of the image SS is subtracted from the corresponding data sample of the image S by combining the formula (2), so as to obtain a noise distribution corresponding to the second noise level metric, that is, the noise distribution does not include image content data.
And performing iterative training on the convolutional neural network based on residual learning by adopting the obtained noise distribution corresponding to different second noise level measurement values, so as to obtain the noise reduction convolutional neural network corresponding to different second noise level measurement values.
The trained noise reduction convolutional neural network can identify the overall distribution of noise in the amplitude image data corresponding to each echo time point, instead of the numerical relationship between neighborhoods in the amplitude image, aiming at specific noise, namely noise meeting each second noise level metric value.
It can be understood that, if any error value is not smaller than the preset error threshold in step S130, the noise distribution corresponding to the first noise level metric may be obtained to train a noise reduction convolutional neural network satisfying the first noise level metric, or the noise reduction process may be directly ended, which is not limited herein.
And S150, denoising the amplitude image data acquired at the corresponding echo time point by adopting the noise value corresponding to each echo time point to obtain noiseless amplitude image data of each echo time point.
And subtracting the noise value corresponding to the corresponding echo time point from the amplitude image data of each echo time point to obtain the noise-free amplitude image data of each echo time point.
Based on the above embodiment of the present application, because the noise influence of the amplitude image data acquired at different echo time points of the mcre sequence is considered, that is, the internal relation between the amplitude image data acquired at different echo time points is considered, the T2 × attenuation curve is smoothed, that is, the noise reduction effect of the image is improved, thereby improving the accuracy of the myelin water content value calculation.
As shown in fig. 2, taking a T2 × attenuation curve corresponding to a voxel in mcre data as an example, after the noise reduction convolutional neural network with a specific noise level and corresponding noise reduction processing, an original T2 × attenuation curve, that is, an original attenuation curve with different oscillation amplitudes in the graph is smoothed, so as to obtain a smoothed noise-reduced curve, that is, a noise-reduced attenuation curve. The myelin water fraction Original MWF for the selected voxels was 0.27287 and Denoised MWF after denoising was 0.17078. In fig. 2, TE represents the echo time, and Signal represents the Signal quantity, i.e., the amplitude of the voxel.
Due to the signal features of the mGRE sequence S (r, TE) and its features fitted by the complex three-pool model S (t):
where r represents the spatial location of each voxel, A/M represents the amplitude value, TE represents the echo time, ω 0 represents the proton precession frequency, my, ax and ex represent myelin water, axonal water and extracellular water, respectively, f represents the frequency and
representing the initial phase value. The myelin water fraction can be expressed as:
MWF＝A my /(A my +A ax +A ex ) (5)
it can be seen that there is a link between the amplitude image data acquired at each echo time. By applying equation (4) to fit the result of equation (3), the myelin water fraction is finally obtained, and the accuracy of noise reduction will directly affect the estimation of the myelin water fraction.
As shown in fig. 3, taking the myelin water value corresponding to 30 voxels as an example, after the noise reduction convolutional neural network with a specific noise level and the corresponding noise reduction processing, the discretization range (unreasonable range) of the original myelin water value (solid circle in the figure) is adjusted to a reasonable range (hollow circle in the figure). To increase statistical accuracy, two sets of data can be differentially tested using a two-tailed based independent sample T-test. Fig. 3 shows the MWF values for 30 randomly selected voxels, which were data before (filled circles in the figure) and after (hollow circles in the figure) denoising, respectively. MWF: myelin water fraction; original mean: average of raw myelin water fraction; denoised mean: average value of myelin water fraction after noise reduction; SD: standard deviation; p: difference values of independent sample t-test.
The method considers the noise influence of the amplitude image data acquired by different echo time points of the mGRE sequence, calculates the noise level of the whole mGRE sequence without taking the noise in the amplitude image data acquired by the first echo time point as reference, and identifies the noise in the amplitude image data corresponding to each echo time point by the trained noise reduction convolution neural network so as to maximally consider the oscillation amplitude problem of the whole T2 attenuation curve, thereby improving the noise reduction accuracy and the myelin sheath water fraction calculation accuracy.
Corresponding to the above method, an embodiment of the present application further provides an image noise reduction device, as shown in fig. 4, where the image noise reduction device includes: the device comprises an acquisition unit 410, a determination unit 420, an acquisition unit 430 and a noise reduction unit 440;
the acquisition unit 410 is used for acquiring myelin sheath moisture image data of voxels in human brain tissue based on a preset multi-echo gradient echo sequence to obtain amplitude image data acquired at each echo time point; the preset multi-echo gradient echo sequence comprises a plurality of echo time points;
a determining unit 420, configured to determine a first noise level metric of the multi-echo gradient echo sequence based on a noise value in amplitude image data acquired at a last echo time point and an amplitude of the amplitude image data corresponding to a first echo time point;
an obtaining unit 430, configured to obtain error values between the first noise level metric values and the stored second noise level metric values respectively; the second noise level metric is determined after adding a new noise value based on the initial recorded amplitude image data containing the noise value;
if the target error value is smaller than a preset error threshold value, acquiring a target second noise level metric value corresponding to the target error value; the target error value is one of the obtained error values;
inputting the amplitude image data acquired at each echo time point as input data into a preset noise reduction convolution neural network corresponding to the target second noise level metric value to obtain a noise value of the amplitude image data of each echo time point output by the noise reduction convolution neural network; the noise reduction convolutional neural network is obtained by training a convolutional neural network based on residual learning based on the noise distribution corresponding to different second noise level measurement values;
and the noise reduction unit 440 is configured to reduce noise of the amplitude image data acquired at the corresponding echo time point by using the noise value corresponding to each echo time point, so as to obtain noise-free amplitude image data of each echo time point.
In an alternative implementation, the determining unit 420 is further configured to:
determining the variance of background data in the amplitude image data of the last echo time point as a noise value in the amplitude image data of the last echo time point;
and determining the average value of all amplitudes in the amplitude image data of the first echo time point as the amplitude of the amplitude image data of the first echo time point.
In an alternative implementation, the determining unit 420 is specifically configured to determine a first noise level metric of the multi-echo gradient echo sequence according to a result of dividing a noise value in the amplitude image data acquired at the last echo time point by an amplitude of the amplitude image data acquired at the first echo time point.
In an alternative implementation, the apparatus further comprises a training unit 450; a training unit 450 for:
based on a preset multi-echo gradient echo sequence, myelin sheath water image data of human brain tissues are collected, and amplitude image data of myelin sheath water images collected corresponding to all echo time points are obtained through calculation;
adding different preset noise values into the initial recorded amplitude image data of the corresponding echo time point according to the first echo time point and the last echo time point, acquiring different new recorded amplitude image data corresponding to the echo time point and second noise level measurement values corresponding to the different new recorded amplitude image data, and storing the second noise level measurement values;
determining noise distributions corresponding to different second noise level metrics based on the different newly recorded amplitude image data and corresponding initially recorded amplitude image data;
and training the convolutional neural network based on residual learning by adopting the noise distribution corresponding to the different second noise level metric values to obtain the noise reduction convolutional neural network corresponding to the different second noise level metric values.
In an optional implementation, the determining unit 420 is further specifically configured to:
increasing the variance of background data in the initial recorded amplitude image data corresponding to the first echo time point and the last echo time point by the preset different noise values to obtain new recorded amplitude image data corresponding to the first echo time point and the last echo time point; wherein the newly recorded amplitude image data corresponding to the first echo time point and the last echo time point contains a new noise value;
determining a second noise level metric for the multi-echo gradient echo sequence based on a new noise value in the newly recorded amplitude image data for the last echo time point and an amplitude of the newly recorded amplitude image data for the first echo time point.
In an optional implementation, the amplitude image data of each echo time point are subjected to a rice distribution;
the obtaining unit 430 is further configured to increase the variance in the rice distribution obeyed by the initial recorded amplitude image data corresponding to the first echo time point and the last echo time point by the preset different noise values, so as to obtain new recorded amplitude image data corresponding to the first echo time point and the last echo time point.
In an optional implementation, the noise reduction unit 440 is specifically configured to subtract the noise value corresponding to the corresponding echo time point from the amplitude image data at each echo time point to obtain noise-free amplitude image data at each echo time point.
The functions of the functional units of the image noise reduction apparatus provided in the above embodiments of the present application may be implemented by the above method steps, and therefore, detailed working processes and beneficial effects of the units in the image noise reduction apparatus provided in the embodiments of the present application are not repeated herein.
An electronic device is further provided in the embodiments of the present application, as shown in fig. 5, and includes a processor 510, a communication interface 520, a memory 530, and a communication bus 540, where the processor 510, the communication interface 520, and the memory 530 complete communication with each other through the communication bus 540.
A memory 530 for storing a computer program;
the processor 510, when executing the program stored in the memory 530, implements the following steps:
acquiring myelin sheath water image data of voxels in human brain tissue based on a preset multi-echo gradient echo sequence to obtain amplitude image data acquired at each echo time point; the preset multi-echo gradient echo sequence comprises a plurality of echo time points;
determining a first noise level metric value of the multi-echo gradient echo sequence based on a noise value in amplitude image data acquired at a last echo time point and an amplitude of the amplitude image data corresponding to a first echo time point;
obtaining error values between the first noise level metric values and the stored second noise level metric values respectively; the second noise level metric is determined after adding a new noise value based on the initial recorded amplitude image data containing the noise value;
if the target error value is smaller than a preset error threshold value, acquiring a target second noise level metric value corresponding to the target error value; the target error value is one of the obtained error values;
inputting the amplitude image data acquired at each echo time point as input data into a preset noise reduction convolutional neural network corresponding to the target second noise level metric value to obtain a noise value of the amplitude image data of each echo time point output by the noise reduction convolutional neural network; the noise reduction convolutional neural network is obtained by training a convolutional neural network based on residual learning based on noise distribution corresponding to different second noise level measurement values;
and denoising the amplitude image data acquired by the corresponding echo time point by adopting the noise value corresponding to each echo time point to obtain noiseless amplitude image data of each echo time point.
The aforementioned communication bus may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus.
The communication interface is used for communication between the electronic equipment and other equipment.
The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the processor.
The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but also Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components.
As the implementation manner and the beneficial effects of the problem solving of each device of the electronic device in the foregoing embodiment can be implemented by referring to each step in the embodiment shown in fig. 1, detailed working processes and beneficial effects of the electronic device provided in the embodiment of the present application are not repeated herein.
In yet another embodiment provided by the present application, a computer-readable storage medium is further provided, which stores instructions that, when executed on a computer, cause the computer to execute the method for reducing noise of an image according to any one of the above embodiments.
In yet another embodiment provided by the present application, there is also provided a computer program product containing instructions which, when run on a computer, cause the computer to perform the method for reducing noise in an image as described in any of the above embodiments.
As will be appreciated by one of skill in the art, the embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, embodiments of the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including the preferred embodiment and all changes and modifications that fall within the true scope of the embodiments of the present application.
It is apparent that those skilled in the art can make various changes and modifications to the embodiments of the present application without departing from the spirit and scope of the embodiments of the present application. Thus, if such modifications and variations of the embodiments of the present application fall within the scope of the claims of the embodiments of the present application and their equivalents, the embodiments of the present application are also intended to include such modifications and variations.
Claims (10)
1. A method for reducing noise in an image, the method comprising:
acquiring myelin sheath water image data of voxels in human brain tissue based on a preset multi-echo gradient echo sequence to obtain amplitude image data acquired at each echo time point; the preset multi-echo gradient echo sequence comprises a plurality of echo time points;
determining a first noise level metric value of the multi-echo gradient echo sequence based on a noise value in amplitude image data acquired at a last echo time point and an amplitude of the amplitude image data corresponding to a first echo time point;
obtaining error values between the first noise level metric values and the stored second noise level metric values respectively; the second noise level metric is determined after adding a new noise value based on the initial recorded amplitude image data containing the noise value;
if the target error value is smaller than a preset error threshold value, acquiring a target second noise level metric value corresponding to the target error value; the target error value is one of the obtained error values;
inputting the amplitude image data acquired at each echo time point as input data into a preset noise reduction convolutional neural network corresponding to the target second noise level metric value to obtain a noise value of the amplitude image data of each echo time point output by the noise reduction convolutional neural network; the noise reduction convolutional neural network is obtained by training a convolutional neural network based on residual learning based on noise distribution corresponding to different second noise level measurement values;
and denoising the amplitude image data acquired at the corresponding echo time point by adopting the noise value corresponding to each echo time point to obtain noiseless amplitude image data of each echo time point.
2. The method of claim 1, wherein prior to determining the first noise level metric for the multi-echo gradient echo sequence based on the magnitude of the magnitude image data corresponding to the first echo time point and the noise value in the magnitude image data acquired at the last echo time point, the method further comprises:
determining the variance of the background data in the amplitude image data of the last echo time point as a noise value in the amplitude image data of the last echo time point;
and determining the average value of the amplitudes in the amplitude image data of the first echo time point as the amplitude of the amplitude image data of the first echo time point.
3. The method of claim 1 or 2, wherein determining a first noise level metric value for the multi-echo gradient echo sequence based on the magnitude of the magnitude image data corresponding to the first echo time point and the noise value in the magnitude image data acquired at the last echo time point comprises:
and determining a first noise level metric value of the multi-echo gradient echo sequence according to a result of dividing the noise value in the amplitude image data acquired at the last echo time point by the amplitude of the amplitude image data at the first echo time point.
4. The method of claim 1, wherein the training process of the noise-reducing convolutional neural network comprises:
based on a preset multi-echo gradient echo sequence, myelin sheath water image data of human brain tissues are collected, and amplitude image data of myelin sheath water images collected corresponding to all echo time points are obtained through calculation;
adding different preset noise values into the initial recorded amplitude image data of the corresponding echo time point according to the first echo time point and the last echo time point, acquiring different new recorded amplitude image data corresponding to the echo time point and second noise level measurement values corresponding to the different new recorded amplitude image data, and storing the second noise level measurement values;
determining noise distributions corresponding to different second noise level metrics based on the different newly recorded amplitude image data and corresponding initially recorded amplitude image data;
and training the convolutional neural network based on residual learning by adopting the noise distribution corresponding to the different second noise level metric values to obtain the noise reduction convolutional neural network corresponding to the different second noise level metric values.
5. The method of claim 4, wherein adding different preset noise values to the initial recorded amplitude image data of the corresponding echo time point for the first echo time point and the last echo time point, and obtaining different new recorded amplitude image data corresponding to the echo time point and a second noise level metric value corresponding to the different new recorded amplitude image data samples comprises:
increasing the variance of background data in the initial recorded amplitude image data corresponding to the first echo time point and the last echo time point by the preset different noise values to obtain new recorded amplitude image data corresponding to the first echo time point and the last echo time point; wherein the newly recorded amplitude image data corresponding to the first echo time point and the last echo time point contains a new noise value;
and determining a second noise level metric of the multi-echo gradient echo sequence based on a new noise value in the newly recorded amplitude image data of the last echo time point and the amplitude of the newly recorded amplitude image data of the first echo time point.
6. The method according to claim 5, wherein the amplitude image data of each echo time point obeys a rice distribution;
increasing the variance of background data in the initial recorded amplitude image data corresponding to the first echo time point and the last echo time point by the preset different noise values to obtain new recorded amplitude image data corresponding to the first echo time point and the last echo time point, including:
and increasing the variance in the Rice distribution obeyed by the initial recorded amplitude image data corresponding to the first echo time point and the last echo time point by the preset different noise values to obtain new recorded amplitude image data corresponding to the first echo time point and the last echo time point.
7. The method of claim 1, wherein the denoising the amplitude image data collected at the corresponding echo time point using the noise value corresponding to each echo time point to obtain the noiseless amplitude image data at each echo time point comprises:
and subtracting the noise value corresponding to the corresponding echo time point from the amplitude image data of each echo time point to obtain the noise-free amplitude image data of each echo time point.
8. An apparatus for reducing noise in an image, the apparatus comprising:
the acquisition unit is used for acquiring myelin sheath water image data of voxels in human brain tissue based on a preset multi-echo gradient echo sequence to obtain amplitude image data acquired at each echo time point; the preset multi-echo gradient echo sequence comprises a plurality of echo time points;
the determining unit is used for determining a first noise level metric value of the multi-echo gradient echo sequence based on the amplitude of the amplitude image data corresponding to the first echo time point and the noise value in the amplitude image data acquired at the last echo time point;
an obtaining unit, configured to obtain error values between the first noise level metric values and the stored second noise level metric values; the second noise level metric is determined after adding a new noise value based on the initial recorded amplitude image data containing the noise value;
if the target error value is smaller than a preset error threshold value, acquiring a target second noise level metric value corresponding to the target error value; the target error value is one of the obtained error values;
inputting the amplitude image data acquired at each echo time point as input data into a preset noise reduction convolutional neural network corresponding to the target second noise level metric value to obtain a noise value of the amplitude image data of each echo time point output by the noise reduction convolutional neural network; the noise reduction convolutional neural network is obtained by training a convolutional neural network based on residual learning based on noise distribution corresponding to different second noise level measurement values;
and the noise reduction unit is used for reducing noise of the amplitude image data acquired at the corresponding echo time point by adopting the noise value corresponding to each echo time point to obtain noiseless amplitude image data of each echo time point.
9. An electronic device, characterized in that the electronic device comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus;
a memory for storing a computer program;
a processor for implementing the method steps of any one of claims 1 to 7 when executing a program stored on a memory.
10. A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, which computer program, when being executed by a processor, carries out the method steps of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210579138.6A CN114972565A (en) | 2022-05-25 | 2022-05-25 | Image noise reduction method and device, electronic equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210579138.6A CN114972565A (en) | 2022-05-25 | 2022-05-25 | Image noise reduction method and device, electronic equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114972565A true CN114972565A (en) | 2022-08-30 |
Family
ID=82956138
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210579138.6A Pending CN114972565A (en) | 2022-05-25 | 2022-05-25 | Image noise reduction method and device, electronic equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114972565A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117093827A (en) * | 2023-10-16 | 2023-11-21 | 欣灵电气股份有限公司 | Intelligent fire control water supply data processing system based on Internet of things |
-
2022
- 2022-05-25 CN CN202210579138.6A patent/CN114972565A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117093827A (en) * | 2023-10-16 | 2023-11-21 | 欣灵电气股份有限公司 | Intelligent fire control water supply data processing system based on Internet of things |
| CN117093827B (en) * | 2023-10-16 | 2024-01-30 | 欣灵电气股份有限公司 | Intelligent fire control water supply data processing system based on Internet of things |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Mason et al. | Comparison of objective image quality metrics to expert radiologists’ scoring of diagnostic quality of MR images | |
| Yang et al. | Local statistics and non-local mean filter for speckle noise reduction in medical ultrasound image | |
| Aja-Fernandez et al. | Spatially variant noise estimation in MRI: A homomorphic approach | |
| Ganzetti et al. | Intensity inhomogeneity correction of structural MR images: a data-driven approach to define input algorithm parameters | |
| EP2660618B1 (en) | Biomedical image reconstruction method | |
| Rajan et al. | A new non-local maximum likelihood estimation method for Rician noise reduction in magnetic resonance images using the Kolmogorov–Smirnov test | |
| JP5773398B2 (en) | System and method for estimating an amount of interest in an arterial / tissue / venous dynamic system | |
| US10698065B2 (en) | System, method and computer accessible medium for noise estimation, noise removal and Gibbs ringing removal | |
| Aja-Fernández et al. | Effective noise estimation and filtering from correlated multiple-coil MR data | |
| Tabelow et al. | Local estimation of the noise level in MRI using structural adaptation | |
| Martin-Fernandez et al. | Sequential anisotropic Wiener filtering applied to 3D MRI data | |
| Reischauer et al. | Image denoising substantially improves accuracy and precision of intravoxel incoherent motion parameter estimates | |
| CN114972565A (en) | Image noise reduction method and device, electronic equipment and medium | |
| Fang et al. | Convolutional neural network for accelerating the computation of the extended Tofts model in dynamic contrast‐enhanced magnetic resonance imaging | |
| Feng et al. | Improved pixel‐by‐pixel MRI R2* relaxometry by nonlocal means | |
| Chahid et al. | A New ROI-Based performance evaluation method for image denoising using the Squared Eigenfunctions of the Schrödinger Operator | |
| CN108523890A (en) | Local field method of estimation based on magnetic resonance dipole field spatial distribution | |
| CN114529473A (en) | Image reconstruction method, image reconstruction device, electronic apparatus, storage medium, and computer program | |
| US7508211B2 (en) | Regularized species separation | |
| CN112581385B (en) | Diffusion kurtosis imaging tensor estimation method, medium and device based on multiple prior constraints | |
| Ferrari | Off-line determination of the optimal number of iterations of the robust anisotropic diffusion filter applied to denoising of brain MR images | |
| CN114821049A (en) | Method, device and equipment for segmenting image and storage medium | |
| CN115375583A (en) | PET parameter image enhancement method, device, equipment and storage medium | |
| Kwon et al. | Multi‐contrast MR image denoising for parallel imaging using multilayer perceptron | |
| Brzostowski et al. | Combining variational mode decomposition with regularisation techniques to denoise MRI data |
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 |