MIRTio.jl Documentation

MIRTio I/O routines for Michigan Image Reconstruction Toolbox (MIRT)

esc = h5_get_ESC(filename::String; T::DataType = ComplexF32)

Return Array of ESC (emulated single coil) data from file.

HDF5.jl reads the data differently than Python's h5py. This is significant because the fastMRI paper says that the datasets have certain dimensionality, but the dimensions are permuted in Julia compared to Python, hence the calls to permutedims in the following functions.

rss = h5_get_RSS(filename::String; T::DataType = ComplexF32)

Return Array of RSS (root sum of squares) data from file.

attr = h5_get_attributes(filename::String)

Get attributes attr from file. Returns a Dict

hdr = h5_get_ismrmrd(filename::String)

Get ISMRM header data from file. Returns a ?

keys = h5_get_keys(filename::String)

Get keys from file using keys(). Returns an Array{String}

kspace = h5_get_kspace(filename::String; T::DataType = ComplexF32))

Return Array of kspace data from file.

hd = header_example()

Return example of header description

out

• hd header definition array (N × 3)

ht = header_init(hd::Matrix{Any})

Return named tuple initialized with arbitrary values for testing

in

• hd header definition array (N × 3), e.g., from header_example()

out

• ht::NamedTuple

ht = header_read(fid::IOStream, hd::Matrix{Any} ; seek0::Bool)

Read header from scan file

in

• fid::IOStream from open(file, "r")
• hd header definition array (N × 3), e.g., from header_example()

option

• seek0::Bool seek to 0 location in file first? default true

out

• ht::NamedTuple with header values, accessed by ht.key

bytes = header_size(hd::Matrix{Any})

return total size of header in bytes

in

• hd header definition array (N × 3), e.g., from header_example()

out = bytes::Int

ht = header_string(ht::NamedTuple ; strip::Bool)

Convert each Array{Cuchar} to String in NamedTuple

in

• ht::NamedTuple from header_init() or header_read()

option

• strip::Bool strip tail white space characters from strings? default false

out

• ht::NamedTuple

header_write(fid::IOStream, ht::NamedTuple)

Write header to fid

in

• fid::IOStream from open(file, "w")
• ht::NamedTuple from header_init() or header_read()

(dat, rdb_hdr) = loadpfile(fid::IOStream ; ...)

Load data for one or more echoes from GE MRI scan Pfile.

in

• fid::IOStream

option

• coils::AbstractVector{Int}

only get data for these coils; default: all coils

• echoes::AbstractVector{Int}

only get data for these echoes; default: all echoes

• slices::AbstractVector{Int}

only get data for these slices; default: 2:nslices (NB!) because first slice (dabslice=0 slot) may contain corrupt data.

• views::AbstractVector{Int}

only get data for these views; default: all views

• quiet::Bool non-verbosity, default false

out

• dat::Array{Complex{Int16}} [ndat, ncoil, nslice, necho, nview]
• rdb_hdr::NamedTuple header information

To save memory the output type is complex-valued Int16.

(dat, rdb_hdr) = loadpfile(file, echo::Integer ; ...)

Load a single echo from file.

(dat, rdb_hdr) = loadpfile(file::String ; ...)

Load from file.

hd = rdb_hdr_26_002_def()

Return array [Symbol Int DataType] ? × 3 used in header_read()

ht = read_rdb_hdr(fid::IOStream)

Read GE raw (RDB) header for MRI scans

Returns NamedTuple ht with header values accessible by ht.key

ht = read_rdb_hdr(file::String)

Read from file