From 61be100ca88ae8af01a34f14baa1abeae8a1c8d3 Mon Sep 17 00:00:00 2001 From: "Documenter.jl" Date: Mon, 6 Nov 2023 11:00:47 +0000 Subject: [PATCH] build based on 56b18e3 --- dev/.documenter-siteinfo.json | 2 +- dev/basicReconstruction.html | 2 +- dev/datasetStore.html | 2 +- dev/index.html | 2 +- dev/matrixCompression.html | 2 +- dev/multiContrast.html | 2 +- dev/multiPatch.html | 2 +- dev/overview.html | 2 +- dev/parameters.html | 2 +- dev/recoResults.html | 2 +- 10 files changed, 10 insertions(+), 10 deletions(-) diff --git a/dev/.documenter-siteinfo.json b/dev/.documenter-siteinfo.json index aa25803..5fb7762 100644 --- a/dev/.documenter-siteinfo.json +++ b/dev/.documenter-siteinfo.json @@ -1 +1 @@ -{"documenter":{"julia_version":"1.9.3","generation_timestamp":"2023-11-06T10:23:59","documenter_version":"1.1.2"}} \ No newline at end of file +{"documenter":{"julia_version":"1.9.3","generation_timestamp":"2023-11-06T11:00:43","documenter_version":"1.1.2"}} \ No newline at end of file diff --git a/dev/basicReconstruction.html b/dev/basicReconstruction.html index bb0b00b..162df4a 100644 --- a/dev/basicReconstruction.html +++ b/dev/basicReconstruction.html @@ -16,4 +16,4 @@ spectralCleaning=true, fgFrames=1:10, bgCorrectionInternal=false, noiseFreqThresh=0.0, kargs...)

This function is responsible for loading the measurement data and potential background data that is subtracted from the measurements. For any frame to be reconstructed, the low level reconstruction routine is called.

Low Level Reconstruction

Finally, we have arrived at the low level reconstruction routine that has the signature

function reconstruction(S, u::Array; sparseTrafo = nothing,
                         lambd=0, progress=nothing, solver = "kaczmarz",
-                        weights=nothing, kargs...)

One can see that it requires the system matrix S and the measurements u to be already loaded.

We note that S is typeless for a reason here. For a regular reconstruction one will basically feed in an Array{ComplexF32,2} in here, although more precisely it will be a Transposed version of that type if the Kaczmarz algorithm is being used for efficiency reasons.

However, in case that matrix compression is applied S will be of type SparseMatrixCSC. And for Multi-Patch Reconstruction S will be of type MultiPatchOperator. Hence, the solvers are implemented in a very generic way and require only certain functions to be implemented. The low level reconstruction method calls one of the solvers from RegularizedLeastSquares.jl.

+ weights=nothing, kargs...)

One can see that it requires the system matrix S and the measurements u to be already loaded.

We note that S is typeless for a reason here. For a regular reconstruction one will basically feed in an Array{ComplexF32,2} in here, although more precisely it will be a Transposed version of that type if the Kaczmarz algorithm is being used for efficiency reasons.

However, in case that matrix compression is applied S will be of type SparseMatrixCSC. And for Multi-Patch Reconstruction S will be of type MultiPatchOperator. Hence, the solvers are implemented in a very generic way and require only certain functions to be implemented. The low level reconstruction method calls one of the solvers from RegularizedLeastSquares.jl.

diff --git a/dev/datasetStore.html b/dev/datasetStore.html index 4211c39..817e9d3 100644 --- a/dev/datasetStore.html +++ b/dev/datasetStore.html @@ -1,2 +1,2 @@ -- · MPI Reconstruction
GitHub

Layers

The reconstruction function has several layers starting from a high level over several middle layer to low layer functions. The most high level method has the following signature `$julia reconstruction(d::MDFDatasetStore, study::Study, exp::Experiment, recoParams::Dict{String,Any})$

+- · MPI Reconstruction
GitHub

Layers

The reconstruction function has several layers starting from a high level over several middle layer to low layer functions. The most high level method has the following signature `$julia reconstruction(d::MDFDatasetStore, study::Study, exp::Experiment, recoParams::Dict{String,Any})$

diff --git a/dev/index.html b/dev/index.html index b4593fa..4014338 100644 --- a/dev/index.html +++ b/dev/index.html @@ -1,2 +1,2 @@ -Home · MPI Reconstruction
GitHub

MPIReco.jl

Julia package for the reconstruction of magnetic particle imaging (MPI) data

Introduction

This project provides functions for the reconstruction of MPI data. The project is implemented in the programming language Julia and contains algorithms for

Key features are

  • Frequency filtering for memory efficient reconstruction. Only frequencies used during reconstructions are loaded into memory.
  • Different solvers provided by the package RegularizedLeastSquares.jl
  • High-level until low-level reconstruction providing maximum flexibility for the user
  • Spectral leakage correction (implemented in MPIFiles.jl)

Installation

Start julia and open the package mode by entering ]. Then enter

add MPIReco

This will install the packages MPIReco.jl and all its dependencies. In particular this will install the core dependencies MPIFiles and RegularizedLeastSquares.

License / Terms of Usage

The source code of this project is licensed under the MIT license. This implies that you are free to use, share, and adapt it. However, please give appropriate credit by citing the project.

Contact

If you have problems using the software, find mistakes, or have general questions please use the issue tracker to contact us.

Contributors

+Home · MPI Reconstruction
GitHub

MPIReco.jl

Julia package for the reconstruction of magnetic particle imaging (MPI) data

Introduction

This project provides functions for the reconstruction of MPI data. The project is implemented in the programming language Julia and contains algorithms for

Key features are

  • Frequency filtering for memory efficient reconstruction. Only frequencies used during reconstructions are loaded into memory.
  • Different solvers provided by the package RegularizedLeastSquares.jl
  • High-level until low-level reconstruction providing maximum flexibility for the user
  • Spectral leakage correction (implemented in MPIFiles.jl)

Installation

Start julia and open the package mode by entering ]. Then enter

add MPIReco

This will install the packages MPIReco.jl and all its dependencies. In particular this will install the core dependencies MPIFiles and RegularizedLeastSquares.

License / Terms of Usage

The source code of this project is licensed under the MIT license. This implies that you are free to use, share, and adapt it. However, please give appropriate credit by citing the project.

Contact

If you have problems using the software, find mistakes, or have general questions please use the issue tracker to contact us.

Contributors

diff --git a/dev/matrixCompression.html b/dev/matrixCompression.html index be47862..52489c0 100644 --- a/dev/matrixCompression.html +++ b/dev/matrixCompression.html @@ -1,2 +1,2 @@ -Compression · MPI Reconstruction
GitHub

Matrix-Compression Techniques

The reconstruction can be accelerated by applying matrix compression. To this end, the system matrix S is transformed into a different domain by applying a basis transformation on the rows of the system matrix. In MPIReco.jl, matrix compression can be enabled by specifying sparseTrafo which can be "DCT-IV" or "FFT".

The transformations can be restricted to the drive-field field-of-view by setting useDFFoV = true. The compression factor that controls how many coefficients are dropped after application of the transformation is controlled by the parameter redFactor. For instance a reduction factor of redFactor = 0.01 will drop 99 % of the data.

+Compression · MPI Reconstruction
GitHub

Matrix-Compression Techniques

The reconstruction can be accelerated by applying matrix compression. To this end, the system matrix S is transformed into a different domain by applying a basis transformation on the rows of the system matrix. In MPIReco.jl, matrix compression can be enabled by specifying sparseTrafo which can be "DCT-IV" or "FFT".

The transformations can be restricted to the drive-field field-of-view by setting useDFFoV = true. The compression factor that controls how many coefficients are dropped after application of the transformation is controlled by the parameter redFactor. For instance a reduction factor of redFactor = 0.01 will drop 99 % of the data.

diff --git a/dev/multiContrast.html b/dev/multiContrast.html index 4f0c327..6372d32 100644 --- a/dev/multiContrast.html +++ b/dev/multiContrast.html @@ -2,4 +2,4 @@ Multi-Contrast · MPI Reconstruction
GitHub

Multi-Contrast Reconstruction

Until now we have discussed single-contrast reconstruction in which case the reconstructed image c has a singleton first dimension. To perform multi-contrast reconstruction one has to specify multiple system matrices

bSFa = MPIFile(filenameA)
 bSFb = MPIFile(filenameB)

and can then invoke

c = reconstruction([bSFa, bSFb], b;
                     SNRThresh=5, frames=1, minFreq=80e3,
-                    recChannels=1:2, iterations=1)

Now one can access the first and second channel by c[1,:,:,:] and c[2,:,:,:].

+ recChannels=1:2, iterations=1)

Now one can access the first and second channel by c[1,:,:,:] and c[2,:,:,:].

diff --git a/dev/multiPatch.html b/dev/multiPatch.html index a82eb52..773de8a 100644 --- a/dev/multiPatch.html +++ b/dev/multiPatch.html @@ -18,4 +18,4 @@ c4 = reconstruction(bSFs, b; SNRThresh=5, frames=1, minFreq=80e3, recChannels=1:2,iterations=1, spectralLeakageCorrection=false, mapping=mapping, systemMatrices = S, SFGridCenter=SFGridCenter, - FFPos=FFPos, FFPosSF=FFPos) + FFPos=FFPos, FFPosSF=FFPos) diff --git a/dev/overview.html b/dev/overview.html index 9cee5dc..c3197c4 100644 --- a/dev/overview.html +++ b/dev/overview.html @@ -12,4 +12,4 @@ recChannels=1:2, iterations=1, spectralLeakageCorrection=true) -

Lets go through that script step by step. First, we create handles for the system matrix and the measurement data. Both are of the type MPIFile which is an abstract type that can for instance be an MDFFile or a BrukerFile.

Using the handles to the MPI datasets we can call the reconstruction function that has various variants depending on the types that are passed to it. Here, we exploit the multiple dispatch mechanism of julia. In addition to the file handles we also apply several reconstruction parameters using keyword arguments. In this case, we set the SNR threshold to 5 implying that only matrix rows with an SNR above 5 are used during reconstruction. The parameter frame decides which frame of the measured data should be reconstructed.

The object c is of type ImageMeta and contains not only the reconstructed data but also several metadata such as the reconstruction parameters being used. More details on the return type are discussed in the Reconstruction Results

Data Storage

One can store the reconstruction result into an MDF file by calling

saveRecoData("filename.mdf", c)

In order to load the data one calls

c = loadRecoData("filename.mdf", c)

We will next take a closer look at different forms of the reconstruction routine.

+

Lets go through that script step by step. First, we create handles for the system matrix and the measurement data. Both are of the type MPIFile which is an abstract type that can for instance be an MDFFile or a BrukerFile.

Using the handles to the MPI datasets we can call the reconstruction function that has various variants depending on the types that are passed to it. Here, we exploit the multiple dispatch mechanism of julia. In addition to the file handles we also apply several reconstruction parameters using keyword arguments. In this case, we set the SNR threshold to 5 implying that only matrix rows with an SNR above 5 are used during reconstruction. The parameter frame decides which frame of the measured data should be reconstructed.

The object c is of type ImageMeta and contains not only the reconstructed data but also several metadata such as the reconstruction parameters being used. More details on the return type are discussed in the Reconstruction Results

Data Storage

One can store the reconstruction result into an MDF file by calling

saveRecoData("filename.mdf", c)

In order to load the data one calls

c = loadRecoData("filename.mdf", c)

We will next take a closer look at different forms of the reconstruction routine.

diff --git a/dev/parameters.html b/dev/parameters.html index 9f04a11..2e927bb 100644 --- a/dev/parameters.html +++ b/dev/parameters.html @@ -1,2 +1,2 @@ -Parameters · MPI Reconstruction
GitHub
+Parameters · MPI Reconstruction
GitHub
diff --git a/dev/recoResults.html b/dev/recoResults.html index 654b302..0ae30b8 100644 --- a/dev/recoResults.html +++ b/dev/recoResults.html @@ -1,2 +1,2 @@ -Results · MPI Reconstruction
GitHub

Reconstruction Results

The object c is of type ImageMeta and contains not only the reconstructed data but also several metadata such as the reconstruction parameters being used. c has in total 5 dimensions. The first dimension encodes multi-spectral channels. Dimensions 2-4 encode the three spatial dimensions. The last dimension contains the number of frames being stored in c.

+Results · MPI Reconstruction
GitHub

Reconstruction Results

The object c is of type ImageMeta and contains not only the reconstructed data but also several metadata such as the reconstruction parameters being used. c has in total 5 dimensions. The first dimension encodes multi-spectral channels. Dimensions 2-4 encode the three spatial dimensions. The last dimension contains the number of frames being stored in c.