Commit ede2fe12 authored by Paul McCarthy's avatar Paul McCarthy 🚵
Browse files

working on nifti/fslpy practical

parent 53f4fd61
# `fslpy`
`fslpy` is a Python library which is built into FSL, and contains a range of
functionality for working with neuroimaging data in an FSL context.
This practical highlights some of the most useful features provided by
`fslpy`. You may find `fslpy` useful if you are writing Python code to
perform analyses and image processing in conjunction with FSL.
> **Note**: `fslpy` is distinct from `fslpython` - `fslpython` is the Python
> environment that is baked into FSL. `fslpy` is a Python library which is
> installed into the `fslpython` environment.
* [The `Image` class, and other data types](#the-image-class-and-other-data-types)
* [FSL atlases](#fsl-atlases)
* [The `filetree`](#the-filetree)
* [NIfTI coordinate systems](#nifti-coordinate-systems)
* [Image processing](#image-processing)
* [FSL wrapper functions](#fsl-wrapper-functions)
<a class="anchor" id="the-image-class-and-other-data-types"></a>
## The `Image` class, and other data types
The `fsl.data.image` module provides the `Image` class, which sits on top of
`nibabel` and contains some handy functionality if you need to work with
coordinate transformations, or do some FSL-specific processing. The `Image`
class provides features such as:
- Support for NIFTI1, NIFTI2, and ANALYZE image files
- Access to affine transformations between the voxel, FSL and world coordinate
systems
- Ability to load metadata from BIDS sidecar files
Some simple image processing routines are also provided - these are covered
[below](#image-processing).
### Creating images
It's easy to create an `Image` - you can create one from a file name:
```
from fsl.data.image import Image
stddir = op.expandvars('${FSLDIR}/data/standard/')
# load a FSL image - the file
# suffix is optional, just like
# in real FSL-land!
img = Image(op.join(stddir, 'MNI152_T1_1mm'))
```
You can crearte an `Image` from an existing `nibabel` image:
```
# load a nibabel image, and
# convert it into an FSL image
nibimg = nib.load(op.join(stddir, 'MNI152_T1_1mm.nii.gz'))
img = Image(nibimg)
``
Or you can create an `Image` from a `numpy` array:
```
data = np.zeros((100, 100, 100))
img = Image(data, xform=np.eye(4))
```
<a class="anchor" id="fsl-atlases"></a>
## FSL atlases
<a class="anchor" id="the-filetree"></a>
## The `filetree`
<a class="anchor" id="nifti-coordinate-systems"></a>
## NIfTI coordinate systems
<a class="anchor" id="image-processing"></a>
## Image processing
<a class="anchor" id="fsl-wrapper-functions"></a>
## FSL wrapper functions
<a class="anchor" id="nifti-coordinate-systems"></a>
## NIfTI coordinate systems
The `getAffine` method gives you access to affine transformations which can be
used to convert coordinates between the different coordinate systems
associated with an image. Have some MNI coordinates you'd like to convert to
voxels? Easy!
```
mnicoords = np.array([[0, 0, 0],
[0, -18, 18]])
world2vox = img.getAffine('world', 'voxel')
vox2world = img.getAffine('voxel', 'world')
# Apply the world->voxel
# affine to the coordinates
voxcoords = (np.dot(world2vox[:3, :3], mnicoords.T)).T + world2vox[:3, 3]
# The code above is a bit fiddly, so
# instead of figuring it out, you can
# just use the transform() function:
from fsl.transform.affine import transform
voxcoords = transform(mnicoords, world2vox)
# just to double check, let's transform
# those voxel coordinates back into world
# coordinates
backtomni = transform(voxcoords, vox2world)
for m, v, b in zip(mnicoords, voxcoords, backtomni):
print(m, '->', v, '->', b)
```
> The `Image.getAffine` method can give you transformation matrices
> between any of these coordinate systems:
>
> - `'voxel'`: Image data voxel coordinates
> - `'world'`: mm coordinates, defined by the sform/qform of an image
> - `'fsl'`: The FSL coordinate system, used internally by many FSL tools
> (e.g. FLIRT)
Oh, that example was too easy I hear you say? Try this one on for size. Let's
say we have run FEAT on some task fMRI data, and want to get the MNI
coordinates of the voxel with peak activation.
> This is what people used to use `Featquery` for, back in the un-enlightened
> days.
Let's start by identifying the voxel with the biggest t-statistic:
```
featdir = op.join(op.join('05_nifti', 'fmri.feat'))
# The Image.data attribute returns a
# numpy array containing, well, the
# image data.
tstat1 = Image(op.join(featdir, 'stats', 'tstat1')).data
# Recall from the numpy practical that
# argmax gives us a 1D index into a
# flattened view of the array. We can
# use the unravel_index function to
# convert it into a 3D index.
peakvox = np.abs(tstat1).argmax()
peakvox = np.unravel_index(peakvox, tstat1.shape)
print('Peak voxel coordinates for tstat1:', peakvox, tstat1[peakvox])
```
Now that we've got the voxel coordinates in functional space, we need to
transform them into MNI space. FEAT provides a transformation which goes
directly from functional to standard space, in the `reg` directory:
```
func2std = np.loadtxt(op.join(featdir, 'reg', 'example_func2standard.mat'))
```
But ... wait a minute ... this is a FLIRT matrix! We can't just plug voxel
coordinates into a FLIRT matrix and expect to get sensible results, because
FLIRT works in an internal FSL coordinate system, which is not quite
`'voxel'`, and not quite `'world'`. So we need to do a little more work.
Let's start by loading our functional image, and the MNI152 template (the
source and reference images of our FLIRT matrix):
```
func = Image(op.join(featdir, 'reg', 'example_func'))
std = Image(op.expandvars(op.join('$FSLDIR', 'data', 'standard', 'MNI152_T1_2mm')))
```
Now we can use them to get affines which convert between all of the different
coordinate systems - we're going to combine them into a single uber-affine,
which transforms our functional-space voxels into MNI world coordinates via:
1. functional voxels -> FLIRT source space
2. FLIRT source space -> FLIRT reference space
3. FLIRT referece space -> MNI world coordinates
```
vox2fsl = func.getAffine('voxel', 'fsl')
fsl2mni = std .getAffine('fsl', 'world')
```
Combining two affines into one is just a simple dot-product. There is a
`concat()` function which does this for us, for any number of affines:
```
from fsl.transform.affine import concat
# To combine affines together, we
# have to list them in reverse -
# linear algebra is *weird*.
funcvox2mni = concat(fsl2mni, func2std, vox2fsl)
```
So we've now got some voxel coordinates from our functional data, and an affine
to transform into MNI world coordinates. The rest is easy:
```
mnicoords = transform(peakvox, funcvox2mni)
print('Peak activation (MNI coordinates):', mnicoords)
```
> Note that in the above example we are only applying a linear transformation
> into MNI space - in reality you would also want to apply your non-linear
> structural-to-standard transformation too. But this is left as an exercise
> for the reader ;).
<a class="anchor" id="image-processing"></a>
## Image processing
Now, it's all well and good to look at t-statistric values and voxel
coordinates and so on and so forth, but let's spice things up a bit and look
at some images. Let's display our peak activation location in MNI space. To
do this, we're going to resample our functional image into
```
from IPython.display import Image as Screenshot
!fsleyes render -of screenshot.png -std
```
### (Advanced) Transform coordinates with nonlinear warpfields
have to use your own dataset
......@@ -6,7 +6,17 @@
"source": [
"# NIfTI images and python\n",
"\n",
"The [nibabel](http://nipy.org/nibabel/) module is used to read and write NIfTI images and also some other medical imaging formats (e.g., ANALYZE, GIFTI, MINC, MGH). This module is included within the FSL python environment.\n",
"The [`nibabel`](http://nipy.org/nibabel/) module is used to read and write NIfTI\n",
"images and also some other medical imaging formats (e.g., ANALYZE, GIFTI,\n",
"MINC, MGH). `nibabel` is included within the FSL python environment.\n",
"\n",
"\n",
"Building upon `nibabel`, the\n",
"[`fsl.data`](https://users.fmrib.ox.ac.uk/~paulmc/fsleyes/fslpy/latest/fsl.data.html#module-fsl.data)\n",
"package contains a number of FSL-specific classes and functions which you may\n",
"find useful. This is covered in a different practical\n",
"(`advanced_topics/08_fslpy.ipynb`).\n",
"\n",
"\n",
"## Contents\n",
"\n",
......@@ -36,8 +46,10 @@
"import os.path as op\n",
"filename = op.expandvars('${FSLDIR}/data/standard/MNI152_T1_1mm.nii.gz')\n",
"imobj = nib.load(filename, mmap=False)\n",
"\n",
"# display header object\n",
"imhdr = imobj.header\n",
"\n",
"# extract data (as an numpy array)\n",
"imdat = imobj.get_data().astype(float)\n",
"print(imdat.shape)"
......@@ -51,25 +63,31 @@
"\n",
"\n",
"> We use the expandvars() function above to insert the FSLDIR\n",
">environmental variable into our string. This function is\n",
">discussed more fully in the file management practical.\n",
" \n",
"> environmental variable into our string. This function is\n",
"> discussed more fully in the file management practical.\n",
"\n",
"Reading the data off the disk is not done until `get_data()` is called.\n",
"\n",
"> Pitfall:\n",
">\n",
"> The option `mmap=False`is necessary as turns off memory mapping, which otherwise would be invoked for uncompressed NIfTI files but not for compressed files. Since some functionality behaves differently on memory mapped objects, it is advisable to turn this off.\n",
"> The option `mmap=False`is necessary as turns off memory mapping,\n",
"> which otherwise would be invoked for uncompressed NIfTI files but not for\n",
"> compressed files. Since some functionality behaves differently on memory\n",
"> mapped objects, it is advisable to turn this off.\n",
"\n",
"Once the data is read into a numpy array then it is easily manipulated.\n",
"\n",
"> We recommend converting it to float at the start to avoid problems with integer arithmetic and overflow, though this is not compulsory.\n",
"> We recommend converting it to float at the start to avoid problems with\n",
"> integer arithmetic and overflow, though this is not compulsory.\n",
"\n",
"---\n",
"\n",
"<a class=\"anchor\" id=\"header-info\"></a>\n",
"## Header info\n",
"\n",
"There are many methods available on the header object - for example, look at `dir(imhdr)` or `help(imhdr)` or the [nibabel webpage about NIfTI images](http://nipy.org/nibabel/nifti_images.html)\n",
"There are many methods available on the header object - for example, look at\n",
"`dir(imhdr)` or `help(imhdr)` or the [nibabel webpage about NIfTI\n",
"images](http://nipy.org/nibabel/nifti_images.html)\n",
"\n",
"<a class=\"anchor\" id=\"voxel-sizes\"></a>\n",
"### Voxel sizes\n",
......@@ -137,7 +155,10 @@
"<a class=\"anchor\" id=\"writing-images\"></a>\n",
"## Writing images\n",
"\n",
"If you have created a modified image by making or modifying a numpy array then you need to put this into a NIfTI image object in order to save it to a file. The easiest way to do this is to copy all the header info from an existing image like this:"
"If you have created a modified image by making or modifying a numpy array then\n",
"you need to put this into a NIfTI image object in order to save it to a file.\n",
"The easiest way to do this is to copy all the header info from an existing\n",
"image like this:"
]
},
{
......@@ -156,7 +177,8 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"where `newdata` is the numpy array (the above is a random example only) and `imhdr` is the existing image header (as above).\n",
"where `newdata` is the numpy array (the above is a random example only) and\n",
"`imhdr` is the existing image header (as above).\n",
"\n",
"> It is possible to also just pass in an affine matrix rather than a\n",
"> copied header, but we *strongly* recommend against this if you are\n",
......@@ -167,15 +189,20 @@
"> whenever possible, and just use the affine matrix option if you are\n",
"> creating an entirely separate image, like a simulation.\n",
"\n",
"If the voxel size of the image is different, then extra modifications will be required. For this, or for building an image from scratch, see the [nibabel documentation](http://nipy.org/nibabel/nifti_images.html) on NIfTI images.\n",
"If the voxel size of the image is different, then extra modifications will be\n",
"required. Take a look at the `fslpy` practical for more advanced image\n",
"manipulation options (`advanced_topics/08_fslpy.ipynb`).\n",
"\n",
"---\n",
"\n",
"<a class=\"anchor\" id=\"exercise\"></a>\n",
"\n",
"<a class=\"anchor\" id=\"exercises\"></a>\n",
"## Exercise\n",
"\n",
"Write some code to read in a 4D fMRI image (you can find one [here] if\n",
"you don't have one handy), calculate the tSNR and then save the 3D result."
"\n",
"Write some code to read in a 4D fMRI image (you can find one\n",
"[here](http://www.fmrib.ox.ac.uk/~mark/files/av.nii.gz) if you don't have one\n",
"handy), calculate the tSNR and then save the 3D result."
]
},
{
......
%% Cell type:markdown id: tags:
# NIfTI images and python
The [nibabel](http://nipy.org/nibabel/) module is used to read and write NIfTI images and also some other medical imaging formats (e.g., ANALYZE, GIFTI, MINC, MGH). This module is included within the FSL python environment.
The [`nibabel`](http://nipy.org/nibabel/) module is used to read and write NIfTI
images and also some other medical imaging formats (e.g., ANALYZE, GIFTI,
MINC, MGH). `nibabel` is included within the FSL python environment.
Building upon `nibabel`, the
[`fsl.data`](https://users.fmrib.ox.ac.uk/~paulmc/fsleyes/fslpy/latest/fsl.data.html#module-fsl.data)
package contains a number of FSL-specific classes and functions which you may
find useful. This is covered in a different practical
(`advanced_topics/08_fslpy.ipynb`).
## Contents
* [Reading images](#reading-images)
* [Header info](#header-info)
* [Voxel sizes](#voxel-sizes)
* [Coordinate orientations and mappings](#orientation-info)
* [Writing images](#writing-images)
* [Exercise](#exercise)
---
<a class="anchor" id="reading-images"></a>
## Reading images
It is easy to read an image:
%% Cell type:code id: tags:
```
import numpy as np
import nibabel as nib
import os.path as op
filename = op.expandvars('${FSLDIR}/data/standard/MNI152_T1_1mm.nii.gz')
imobj = nib.load(filename, mmap=False)
# display header object
imhdr = imobj.header
# extract data (as an numpy array)
imdat = imobj.get_data().astype(float)
print(imdat.shape)
```
%% Cell type:markdown id: tags:
> Make sure you use the full filename, including the .nii.gz extension.
> We use the expandvars() function above to insert the FSLDIR
>environmental variable into our string. This function is
>discussed more fully in the file management practical.
> environmental variable into our string. This function is
> discussed more fully in the file management practical.
Reading the data off the disk is not done until `get_data()` is called.
> Pitfall:
>
> The option `mmap=False`is necessary as turns off memory mapping, which otherwise would be invoked for uncompressed NIfTI files but not for compressed files. Since some functionality behaves differently on memory mapped objects, it is advisable to turn this off.
> The option `mmap=False`is necessary as turns off memory mapping,
> which otherwise would be invoked for uncompressed NIfTI files but not for
> compressed files. Since some functionality behaves differently on memory
> mapped objects, it is advisable to turn this off.
Once the data is read into a numpy array then it is easily manipulated.
> We recommend converting it to float at the start to avoid problems with integer arithmetic and overflow, though this is not compulsory.
> We recommend converting it to float at the start to avoid problems with
> integer arithmetic and overflow, though this is not compulsory.
---
<a class="anchor" id="header-info"></a>
## Header info
There are many methods available on the header object - for example, look at `dir(imhdr)` or `help(imhdr)` or the [nibabel webpage about NIfTI images](http://nipy.org/nibabel/nifti_images.html)
There are many methods available on the header object - for example, look at
`dir(imhdr)` or `help(imhdr)` or the [nibabel webpage about NIfTI
images](http://nipy.org/nibabel/nifti_images.html)
<a class="anchor" id="voxel-sizes"></a>
### Voxel sizes
Dimensions of the voxels, in mm, can be found from:
%% Cell type:code id: tags:
```
voxsize = imhdr.get_zooms()
print(voxsize)
```
%% Cell type:markdown id: tags:
<a class="anchor" id="orientation-info"></a>
### Coordinate orientations and mappings
Information about the NIfTI qform and sform matrices can be extracted like this:
%% Cell type:code id: tags:
```
sform = imhdr.get_sform()
sformcode = imhdr['sform_code']
qform = imhdr.get_qform()
qformcode = imhdr['qform_code']
print(qformcode)
print(qform)
```
%% Cell type:markdown id: tags:
You can also get both code and matrix together like this:
%% Cell type:code id: tags:
```
affine, code = imhdr.get_qform(coded=True)
print(affine, code)
```
%% Cell type:markdown id: tags:
---
<a class="anchor" id="writing-images"></a>
## Writing images
If you have created a modified image by making or modifying a numpy array then you need to put this into a NIfTI image object in order to save it to a file. The easiest way to do this is to copy all the header info from an existing image like this:
If you have created a modified image by making or modifying a numpy array then
you need to put this into a NIfTI image object in order to save it to a file.
The easiest way to do this is to copy all the header info from an existing
image like this:
%% Cell type:code id: tags:
```
newdata = imdat * imdat
newhdr = imhdr.copy()
newobj = nib.nifti1.Nifti1Image(newdata, None, header=newhdr)
nib.save(newobj, "mynewname.nii.gz")
```
%% Cell type:markdown id: tags:
where `newdata` is the numpy array (the above is a random example only) and `imhdr` is the existing image header (as above).
where `newdata` is the numpy array (the above is a random example only) and
`imhdr` is the existing image header (as above).
> It is possible to also just pass in an affine matrix rather than a
> copied header, but we *strongly* recommend against this if you are
> processing an existing image as otherwise you run the risk of
> swapping the left-right orientation. Those that have used
> `save_avw` in matlab may well have been bitten in this way in the
> past. Therefore, copy a header from one of the input images
> whenever possible, and just use the affine matrix option if you are
> creating an entirely separate image, like a simulation.
If the voxel size of the image is different, then extra modifications will be required. For this, or for building an image from scratch, see the [nibabel documentation](http://nipy.org/nibabel/nifti_images.html) on NIfTI images.
If the voxel size of the image is different, then extra modifications will be
required. Take a look at the `fslpy` practical for more advanced image
manipulation options (`advanced_topics/08_fslpy.ipynb`).
---
<a class="anchor" id="exercise"></a>
<a class="anchor" id="exercises"></a>
## Exercise
Write some code to read in a 4D fMRI image (you can find one [here] if
you don't have one handy), calculate the tSNR and then save the 3D result.
Write some code to read in a 4D fMRI image (you can find one
[here](http://www.fmrib.ox.ac.uk/~mark/files/av.nii.gz) if you don't have one
handy), calculate the tSNR and then save the 3D result.
%% Cell type:code id: tags:
```
# Calculate tSNR
```
......
# NIfTI images and python
The [nibabel](http://nipy.org/nibabel/) module is used to read and write NIfTI images and also some other medical imaging formats (e.g., ANALYZE, GIFTI, MINC, MGH). This module is included within the FSL python environment.
The [`nibabel`](http://nipy.org/nibabel/) module is used to read and write NIfTI
images and also some other medical imaging formats (e.g., ANALYZE, GIFTI,
MINC, MGH). `nibabel` is included within the FSL python environment.
Building upon `nibabel`, the
[`fsl.data`](https://users.fmrib.ox.ac.uk/~paulmc/fsleyes/fslpy/latest/fsl.data.html#module-fsl.data)
package contains a number of FSL-specific classes and functions which you may
find useful. This is covered in a different practical
(`advanced_topics/08_fslpy.ipynb`).
## Contents
......@@ -24,8 +34,10 @@ import nibabel as nib
import os.path as op
filename = op.expandvars('${FSLDIR}/data/standard/MNI152_T1_1mm.nii.gz')
imobj = nib.load(filename, mmap=False)
# display header object
imhdr = imobj.header
# extract data (as an numpy array)
imdat = imobj.get_data().astype(float)
print(imdat.shape)
......@@ -35,25 +47,31 @@ print(imdat.shape)
> We use the expandvars() function above to insert the FSLDIR
>environmental variable into our string. This function is
>discussed more fully in the file management practical.
> environmental variable into our string. This function is
> discussed more fully in the file management practical.
Reading the data off the disk is not done until `get_data()` is called.
> Pitfall:
>
> The option `mmap=False`is necessary as turns off memory mapping, which otherwise would be invoked for uncompressed NIfTI files but not for compressed files. Since some functionality behaves differently on memory mapped objects, it is advisable to turn this off.
> The option `mmap=False`is necessary as turns off memory mapping,
> which otherwise would be invoked for uncompressed NIfTI files but not for
> compressed files. Since some functionality behaves differently on memory
> mapped objects, it is advisable to turn this off.
Once the data is read into a numpy array then it is easily manipulated.
> We recommend converting it to float at the start to avoid problems with integer arithmetic and overflow, though this is not compulsory.
> We recommend converting it to float at the start to avoid problems with
> integer arithmetic and overflow, though this is not compulsory.
---
<a class="anchor" id="header-info"></a>
## Header info
There are many methods available on the header object - for example, look at `dir(imhdr)` or `help(imhdr)` or the [nibabel webpage about NIfTI images](http://nipy.org/nibabel/nifti_images.html)
There are many methods available on the header object - for example, look at
`dir(imhdr)` or `help(imhdr)` or the [nibabel webpage about NIfTI
images](http://nipy.org/nibabel/nifti_images.html)
<a class="anchor" id="voxel-sizes"></a>
### Voxel sizes
......@@ -91,7 +109,10 @@ print(affine, code)
<a class="anchor" id="writing-images"></a>
## Writing images
If you have created a modified image by making or modifying a numpy array then you need to put this into a NIfTI image object in order to save it to a file. The easiest way to do this is to copy all the header info from an existing image like this:
If you have created a modified image by making or modifying a numpy array then
you need to put this into a NIfTI image object in order to save it to a file.
The easiest way to do this is to copy all the header info from an existing
image like this:
```
newdata = imdat * imdat
......@@ -99,7 +120,9 @@ newhdr = imhdr.copy()
newobj = nib.nifti1.Nifti1Image(newdata, None, header=newhdr)
nib.save(newobj, "mynewname.nii.gz")
```
where `newdata` is the numpy array (the above is a random example only) and `imhdr` is the existing image header (as above).
where `newdata` is the numpy array (the above is a random example only) and
`imhdr` is the existing image header (as above).
> It is possible to also just pass in an affine matrix rather than a
> copied header, but we *strongly* recommend against this if you are
......@@ -110,17 +133,21 @@ where `newdata` is the numpy array (the above is a random example only) and `imh
> whenever possible, and just use the affine matrix option if you are
> creating an entirely separate image, like a simulation.
If the voxel size of the image is different, then extra modifications will be required. For this, or for building an image from scratch, see the [nibabel documentation](http://nipy.org/nibabel/nifti_images.html) on NIfTI images.
If the voxel size of the image is different, then extra modifications will be
required. Take a look at the `fslpy` practical for more advanced image
manipulation options (`advanced_topics/08_fslpy.ipynb`).
---
<a class="anchor" id="exercise"></a>
<a class="anchor" id="exercises"></a>
## Exercise
Write some code to read in a 4D fMRI image (you can find one [here](http://www.fmrib.ox.ac.uk/~mark/files/av.nii.gz) if
you don't have one handy), calculate the tSNR and then save the 3D result.
Write some code to read in a 4D fMRI image (you can find one
[here](http://www.fmrib.ox.ac.uk/~mark/files/av.nii.gz) if you don't have one
handy), calculate the tSNR and then save the 3D result.
```
# Calculate tSNR
```
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment