Coordinate Systems

There are a wide variety of coordinate systems used in Neuroimaging. Neuroimaging.jl provides a coordinate data type, and several subtypes associated with specific systems. Conversion between coordinate systems is available. For a general overview of coordinate systems in neuroimaging see:

• https://mne.tools/dev/autotutorials/forward/20source_alignment.html
• https://www.fieldtriptoolbox.org/faq/coordsys/

This package currently supports SPM, BrainVision, and Talairach coordinates and conversion. Additionally, an Unknown coordinate system can be used if you don't know how your locations are mapped.

Create a coordinate

We can create a coordinate point by calling the appropriate type. Internally Neuroimaging.jl uses SI units, so meters for these locations. But the results are displayed in the more convenient centimeter notation.

By default everything is SI units, so if you pass in a distance without specifying the unit it will be in meters.

using Neuroimaging, Unitful

SPM(0.0737, -0.0260, 0.0070)

SPM(0.0737 m, -0.026 m, 0.007 m)

However, it is clearer and less prone to error if you specify the unit at construction, and let the internals handle the conversion.

location_1 = SPM(73.7u"mm", -26u"mm", 7u"mm")

SPM(0.0737 m, -0.026 m, 0.007 m)

Note that this position is taken from table IV from:

• Lancaster, Jack L., et al. "Bias between MNI and Talairach coordinates analyzed using the ICBM‐152 brain template." Human brain mapping 28.11 (2007): 1194-1205.

Conversion between coordinates

To convert between different coordinate systems simply call the convert function with the first arguments as the desired coordinate system. The show function displays the passed type nicely with domain appropriate units.

show(convert(Talairach, location_1))

[ Info: Converted 1 coordinates to talairach space from SPM MNI
Coordinate: Talairach - (67.03628 mm, -26.95176 mm, 9.16384 mm)

And we can see that the resulting value is similar to what is provided in the Lancaster 2007 article. Although not exact, there is some loss in the transformations.

Sensors

Sensors contain location information stored as coordinate types. So if we load an EEG measurement...

using Neuroimaging, DataDeps
data_path = joinpath(
    datadep"ExampleSSR",
    "Neuroimaging.jl-example-data-master",
    "neuroimaingSSR.bdf",
)

s = read_SSR(data_path)

SSR measurement of 5.0 mins with 7 channels sampled at 8192.0 Hz
  Modulation frequency: NaN Hz

we can then query the sensors by calling...

sensors(s)

7-element Vector{Sensor}:
 Electrode("F5", Talairach(-0.06560941273964613 m, 0.039533392370485365 m, 0.023293714059226867 m), Dict{Any, Any}())
 Electrode("TP7", Talairach(-0.08554294425259022 m, -0.04529458746142005 m, -0.003140954703225086 m), Dict{Any, Any}())
 Electrode("O1", Talairach(-0.027794587461420057 m, -0.10304294425259022 m, -0.003140954703225086 m), Dict{Any, Any}())
 Electrode("F6", Talairach(0.06560941273964613 m, 0.039533392370485365 m, 0.023293714059226867 m), Dict{Any, Any}())
 Electrode("Cz", Talairach(0.0 m, -0.0175 m, 0.09 m), Dict{Any, Any}())
 Electrode("TP8", Talairach(0.08554294425259022 m, -0.04529458746142005 m, -0.003140954703225086 m), Dict{Any, Any}())
 Electrode("O2", Talairach(0.027794587461420057 m, -0.10304294425259022 m, -0.003140954703225086 m), Dict{Any, Any}())

And we can see that there are 7 electrodes with standard 10-20 names. However, they do not have positions encoded by default.

We can query the coordinate positions for the electrodes. For example, to obtain the x locations for all sensors in the EEG measurement use...

x(sensors(s))

7-element Vector{Unitful.Quantity{Float64, 𝐋, Unitful.FreeUnits{(m,), 𝐋, nothing}}}:
  -0.06560941273964613 m
  -0.08554294425259022 m
 -0.027794587461420057 m
   0.06560941273964613 m
                   0.0 m
   0.08554294425259022 m
  0.027794587461420057 m

or to get all the labels use...

labels(sensors(s))

7-element Vector{String}:
 "F5"
 "TP7"
 "O1"
 "F6"
 "Cz"
 "TP8"
 "O2"