def test_ALE_ma_map_reuse(testdata_cbma, tmp_path_factory, caplog):
"""Test that MA maps are re-used when appropriate."""
from nimare.meta import kernel
tmpdir = tmp_path_factory.mktemp("test_ALE_ma_map_reuse")
testdata_cbma.update_path(tmpdir)
# ALEKernel cannot extract sample_size from a Dataset,
# so we need to set it for this kernel and for the later meta-analyses.
kern = kernel.ALEKernel(sample_size=20)
dset = kern.transform(testdata_cbma, return_type="dataset")
# The associated column should be in the new Dataset's images DataFrame
cols = dset.images.columns.tolist()
assert any(["ALEKernel" in col for col in cols])
# The Dataset without the images will generate them from scratch.
# If drop_invalid is False, then there should be an Exception, since two studies in the test
# dataset are missing coordinates.
meta = ale.ALE(kernel__sample_size=20)
with pytest.raises(Exception):
meta.fit(testdata_cbma, drop_invalid=False)
with caplog.at_level(logging.DEBUG, logger="nimare.meta.cbma.base"):
meta.fit(testdata_cbma)
assert "Loading pre-generated MA maps" not in caplog.text
# The Dataset with the images will re-use them, as evidenced by the logger message.
with caplog.at_level(logging.DEBUG, logger="nimare.meta.cbma.base"):
meta.fit(dset)
assert "Loading pre-generated MA maps" in caplog.text
def test_ALEKernel_smoke(testdata_cbma):
"""Smoke test for nimare.meta.kernel.ALEKernel."""
# Manually override dataset coordinates file sample sizes
# This column would be extracted from metadata and added to coordinates
# automatically by the Estimator
coordinates = testdata_cbma.coordinates.copy()
coordinates["sample_size"] = 20
kern = kernel.ALEKernel()
ma_maps = kern.transform(coordinates, testdata_cbma.masker, return_type="image")
assert len(ma_maps) == len(testdata_cbma.ids) - 2
ma_maps = kern.transform(coordinates, testdata_cbma.masker, return_type="array")
assert ma_maps.shape[0] == len(testdata_cbma.ids) - 2
# Test set_params
kern.set_params(fwhm=10, sample_size=None)
kern2 = kernel.ALEKernel(fwhm=10)
ma_maps1 = kern.transform(coordinates, testdata_cbma.masker, return_type="array")
ma_maps2 = kern2.transform(coordinates, testdata_cbma.masker, return_type="array")
assert ma_maps1.shape[0] == ma_maps2.shape[0] == len(testdata_cbma.ids) - 2
assert np.array_equal(ma_maps1, ma_maps2)
def test_ALEKernel_fwhm(testdata_cbma):
"""Peaks of ALE kernel maps should match the foci fed in (assuming focus isn't masked out).
Test with explicit FWHM.
"""
coordinates = testdata_cbma.coordinates.copy()
id_ = "pain_03.nidm-1"
kern = kernel.ALEKernel(fwhm=10)
ma_maps = kern.transform(coordinates, masker=testdata_cbma.masker, return_type="image")
ijk = coordinates.loc[coordinates["id"] == id_, ["i", "j", "k"]]
ijk = np.squeeze(ijk.values.astype(int))
kern_data = ma_maps[0].get_fdata()
max_idx = np.array(np.where(kern_data == np.max(kern_data))).T
max_ijk = np.squeeze(max_idx)
assert np.array_equal(ijk, max_ijk)
def test_ALEKernel_sample_size(testdata_cbma):
"""Peaks of ALE kernel maps should match the foci fed in (assuming focus isn't masked out).
Test with explicit sample size.
"""
coordinates = testdata_cbma.coordinates.copy()
id_ = "pain_03.nidm-1"
kern = kernel.ALEKernel(sample_size=20)
ma_maps = kern.transform(coordinates, masker=testdata_cbma.masker, return_type="image")
xyz = coordinates.loc[coordinates["id"] == id_, ["x", "y", "z"]]
ijk = mm2vox(xyz, testdata_cbma.masker.mask_img.affine)
ijk = np.squeeze(ijk.astype(int))
kern_data = ma_maps[0].get_fdata()
max_idx = np.array(np.where(kern_data == np.max(kern_data))).T
max_ijk = np.squeeze(max_idx)
assert np.array_equal(ijk, max_ijk)
def test_ALEKernel_1mm(testdata_cbma):
"""Peaks of ALE kernel maps should match the foci fed in (assuming focus isn't masked out).
Test on 1mm template.
"""
# Manually override dataset coordinates file sample sizes
# This column would be extracted from metadata and added to coordinates
# automatically by the Estimator
coordinates = testdata_cbma.coordinates.copy()
coordinates["sample_size"] = 20
id_ = "pain_03.nidm-1"
kern = kernel.ALEKernel()
ma_maps = kern.transform(coordinates, testdata_cbma.masker, return_type="image")
ijk = coordinates.loc[coordinates["id"] == id_, ["i", "j", "k"]]
ijk = ijk.values.astype(int)
kern_data = ma_maps[0].get_fdata()
max_idx = np.where(kern_data == np.max(kern_data))
max_ijk = np.array(max_idx).T
assert np.array_equal(ijk, max_ijk)
def test_ALEKernel_inputdataset_returndataset(testdata_cbma, tmp_path_factory):
"""Check that all return types produce equivalent results (minus the masking element)."""
tmpdir = tmp_path_factory.mktemp("test_ALEKernel_inputdataset_returndataset")
testdata_cbma.update_path(tmpdir)
kern = kernel.ALEKernel(sample_size=20, memory_limit="1gb")
ma_maps = kern.transform(testdata_cbma, return_type="image")
ma_arr = kern.transform(testdata_cbma, return_type="array")
dset = kern.transform(testdata_cbma, return_type="dataset")
ma_maps_from_dset = kern.transform(dset, return_type="image")
ma_arr_from_dset = kern.transform(dset, return_type="array")
ma_maps_arr = testdata_cbma.masker.transform(ma_maps)
ma_maps_from_dset_arr = dset.masker.transform(ma_maps_from_dset)
dset_from_dset = kern.transform(dset, return_type="dataset")
ids = dset.coordinates["id"].unique()
ma_maps_dset = testdata_cbma.masker.transform(dset.get_images(ids=ids, imtype=kern.image_type))
assert isinstance(dset_from_dset, Dataset)
assert np.array_equal(ma_arr, ma_maps_arr)
assert np.array_equal(ma_arr, ma_maps_dset)
assert np.array_equal(ma_arr, ma_maps_from_dset_arr)
assert np.array_equal(ma_arr, ma_arr_from_dset)
def test_ALESubtraction_ma_map_reuse(testdata_cbma, tmp_path_factory, caplog):
"""Test that MA maps are re-used when appropriate."""
from nimare.meta import kernel
tmpdir = tmp_path_factory.mktemp("test_ALESubtraction_ma_map_reuse")
testdata_cbma.update_path(tmpdir)
# ALEKernel cannot extract sample_size from a Dataset,
# so we need to set it for this kernel and for the later meta-analyses.
kern = kernel.ALEKernel(sample_size=20)
dset = kern.transform(testdata_cbma, return_type="dataset")
# The Dataset without the images will generate them from scratch.
sub_meta = ale.ALESubtraction(n_iters=10, kernel__sample_size=20)
with caplog.at_level(logging.DEBUG, logger="nimare.meta.cbma.base"):
sub_meta.fit(testdata_cbma, testdata_cbma)
assert "Loading pre-generated MA maps" not in caplog.text
# The Dataset with the images will re-use them,
# as evidenced by the logger message.
with caplog.at_level(logging.DEBUG, logger="nimare.meta.cbma.base"):
sub_meta.fit(dset, dset)
assert "Loading pre-generated MA maps" in caplog.text
# There are three standard kernels that are currently available: {py:class}`~nimare.meta.kernel.MKDAKernel`, {py:class}`~nimare.meta.kernel.KDAKernel`, and {py:class}`~nimare.meta.kernel.ALEKernel`.
# Each class may be configured with certain parameters when a new object is initialized.
# For example, `MKDAKernel` accepts an `r` parameter, which determines the radius of the spheres that will be created around each peak coordinate.
# `ALEKernel` automatically uses the sample size associated with each experiment in the `Dataset` to determine the appropriate full-width-at-half-maximum of its Gaussian distribution, as described in {cite:t}`EICKHOFF20122349`; however, users may provide a constant `sample_size` or `fwhm` parameter when sample size information is not available within the `Dataset` metadata.
#
# Here we show how these three kernels can be applied to the same `Dataset`.
# In[2]:
from nimare.meta import kernel
mkda_kernel = kernel.MKDAKernel(r=10)
mkda_ma_maps = mkda_kernel.transform(sleuth_dset1)
kda_kernel = kernel.KDAKernel(r=10)
kda_ma_maps = kda_kernel.transform(sleuth_dset1)
ale_kernel = kernel.ALEKernel(sample_size=20)
ale_ma_maps = ale_kernel.transform(sleuth_dset1)
# In[3]:
# Here we delete the recent variables for the sake of reducing memory usage
del mkda_kernel, kda_kernel, ale_kernel
# In[4]:
# Generate figure
study_idx = 10 # a study with overlapping kernels
max_value = np.max(kda_ma_maps[study_idx].get_fdata()) + 1
ma_maps = {
"MKDA Kernel": mkda_ma_maps[study_idx],
