Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

cumsum(trj_l) for speed with long trajectories #46

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

cumsum(trj_l) for speed with long trajectories #46

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 8 additions & 8 deletions src/BackProjection.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,17 +30,17 @@ function calculateBackProjection(data::AbstractVector{<:AbstractArray{cT}}, trj:
Ncoil = size(data[1], 2)
xbp = Array{cT}(undef, img_shape..., Ncoef, Ncoil)

trj_l = [size(trj[it], 2) for it in eachindex(trj)]
data_temp = Vector{cT}(undef, sum(trj_l))
trj_idx = cumsum([size(trj[it],2) for it in eachindex(trj)])
data_temp = Vector{cT}(undef, trj_idx[end])

img_idx = CartesianIndices(img_shape)
verbose && println("calculating backprojection...")
flush(stdout)
for icoef ∈ axes(U, 2)
t = @elapsed for icoil ∈ axes(data[1], 2)
@simd for it in eachindex(data)
idx1 = sum(trj_l[1:it-1]) + 1
idx2 = sum(trj_l[1:it])
idx1 = (it == 1) ? 1 : trj_idx[it-1] + 1
idx2 = trj_idx[it]
@views data_temp[idx1:idx2] .= data[it][:, icoil] .* conj(U[it, icoef])
end
applyDensityCompensation!(data_temp, trj_v; density_compensation)
Expand All @@ -64,16 +64,16 @@ function calculateBackProjection(data::AbstractVector{<:AbstractMatrix{cT}}, trj
xbp = zeros(cT, img_shape..., Ncoef)
xtmp = Array{cT}(undef, img_shape)

trj_l = [size(trj[it], 2) for it in eachindex(trj)]
data_temp = Vector{cT}(undef, sum(trj_l))
trj_idx = cumsum([size(trj[it],2) for it in eachindex(trj)])
data_temp = Vector{cT}(undef, trj_idx[end])
img_idx = CartesianIndices(img_shape)
verbose && println("calculating backprojection...")
flush(stdout)
for icoef ∈ axes(U, 2)
t = @elapsed for icoil ∈ eachindex(cmaps)
@simd for it in eachindex(data)
idx1 = sum(trj_l[1:it-1]) + 1
idx2 = sum(trj_l[1:it])
idx1 = (it == 1) ? 1 : trj_idx[it-1] + 1
idx2 = trj_idx[it]
@views data_temp[idx1:idx2] .= data[it][:, icoil] .* conj(U[it, icoef])
end
applyDensityCompensation!(data_temp, trj_v; density_compensation)
Expand Down
16 changes: 8 additions & 8 deletions src/NFFTNormalOpBasisFunc.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -105,21 +105,21 @@ function calculateToeplitzKernelBasis(img_shape_os, trj::AbstractVector{<:Abstra
λ2 = similar(λ)
Λ = Array{Complex{T}}(undef, Ncoeff, Ncoeff, length(kmask_indcs))

trj_l = [size(trj[it],2) for it in eachindex(trj)]
S = Vector{Complex{T}}(undef, sum(trj_l))
trj_idx = cumsum([size(trj[it],2) for it in eachindex(trj)])
S = Vector{Complex{T}}(undef, trj_idx[end])

fftplan = plan_fft(λ; flags = FFTW.MEASURE, num_threads=Threads.nthreads())
nfftplan = plan_nfft(reduce(hcat, trj), img_shape_os; precompute = TENSOR, blocking = true, fftflags = FFTW.MEASURE, m=5, σ=2)

# Evaluating only the upper triangular matrix assumes that the PSF from the rightmost voxel to the leftmost voxel is the adjoint of the PSF in the opposite direction.
# For the outmost voxel, this is not correct, but the resulting images are virtually identical in our test cases.
# To avoid this error, remove the `if ic2 >= ic1` and the `Λ[ic2,ic1,it] = conj.(λ[kmask_indcs[it]])` statements at the cost of computation time.
# Evaluating only the upper triangular matrix assumes that the PSF from the rightmost voxel to the leftmost voxel is the adjoint of the PSF in the opposite direction.
# For the outmost voxel, this is not correct, but the resulting images are virtually identical in our test cases.
# To avoid this error, remove the `if ic2 >= ic1` and the `Λ[ic2,ic1,it] = conj.(λ[kmask_indcs[it]])` statements at the cost of computation time.
for ic2 ∈ axes(Λ, 2), ic1 ∈ axes(Λ, 1)
if ic2 >= ic1 # eval. only upper triangular matrix
t = @elapsed begin
@simd for it ∈ axes(U,1)
idx1 = sum(trj_l[1:it-1]) + 1
idx2 = sum(trj_l[1:it])
idx1 = (it == 1) ? 1 : trj_idx[it-1] + 1
idx2 = trj_idx[it]
@inbounds S[idx1:idx2] .= conj(U[it,ic1]) * U[it,ic2]
end

Expand All @@ -128,7 +128,7 @@ function calculateToeplitzKernelBasis(img_shape_os, trj::AbstractVector{<:Abstra
mul!(λ, fftplan, λ2)

Threads.@threads for it ∈ eachindex(kmask_indcs)
@inbounds Λ[ic2,ic1,it] = conj.(λ[kmask_indcs[it]])
@inbounds Λ[ic2,ic1,it] = conj.(λ[kmask_indcs[it]])
@inbounds Λ[ic1,ic2,it] = λ[kmask_indcs[it]]
end
end
Expand Down
Loading