def test_mvpa_voxel_selection():
data = prng.rand(5, 5, 5, 8).astype(np.float32)
# all MPI processes read the mask; the mask file is small
mask = np.ones([5, 5, 5], dtype=np.bool)
mask[0, 0, :] = False
labels = [0, 1, 0, 1, 0, 1, 0, 1]
# 2 subjects, 4 epochs per subject
sl = Searchlight(sl_rad=1)
mvs = MVPAVoxelSelector(data, mask, labels, 2, sl)
# for cross validation, use SVM with precomputed kernel
clf = svm.SVC(kernel='rbf', C=10)
result_volume, results = mvs.run(clf)
if MPI.COMM_WORLD.Get_rank() == 0:
output = []
for tuple in results:
if tuple[1] > 0:
output.append(int(8*tuple[1]))
expected_output = [6, 6, 5, 5, 5, 5, 5, 5, 5, 4, 4, 4, 4, 4,
4, 4, 4, 3, 3, 3, 3, 3, 2, 2, 2, 1]
assert np.allclose(output, expected_output, atol=1), \
'voxel selection via SVM does not provide correct results'
def test_mvpa_voxel_selection():
raw_data = []
# all MPI processes read the mask; the mask file is small
mask = np.ones([5, 5, 5], dtype=np.bool)
mask[0, 0, :] = False
epoch_info = None
if MPI.COMM_WORLD.Get_rank()==0:
data1 = prng.rand(5, 5, 5, 100).astype(np.float32)
data2 = prng.rand(5, 5, 5, 100).astype(np.float32)
raw_data = [data1, data2]
epoch_info = []
epoch_info.append((0, 0, 10, 20))
epoch_info.append((1, 0, 30, 40))
epoch_info.append((0, 0, 50, 60))
epoch_info.append((1, 0, 70, 80))
epoch_info.append((0, 1, 10, 20))
epoch_info.append((1, 1, 30, 40))
epoch_info.append((0, 1, 50, 60))
epoch_info.append((1, 1, 70, 80))
# 2 subjects, 4 epochs per subject
sl = Searchlight(sl_rad=1)
mvs = MVPAVoxelSelector(raw_data, mask, epoch_info, 2, sl)
# for cross validation, use SVM with precomputed kernel
clf = svm.SVC(kernel='rbf', C=10)
result_volume, results = mvs.run(clf)
if MPI.COMM_WORLD.Get_rank() == 0:
output = []
for tuple in results:
if tuple[1] > 0:
output.append(int(8*tuple[1]))
expected_output = [6, 6, 6, 6, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4,
3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 2, 2]
assert np.allclose(output, expected_output, atol=1), \
'voxel selection via SVM does not provide correct results'
# There are three random options:
# RandomType.NORANDOM is the default
# RandomType.REPRODUCIBLE permutes the voxels in the same way every run
# RandomType.UNREPRODUCIBLE permutes the voxels differently across runs
# example
#from brainiak.fcma.preprocessing import RandomType
#data, labels = prepare_searchlight_mvpa_data(images, conditions,
# random=RandomType.UNREPRODUCIBLE)
# the following line is an example to leaving a subject out
#epoch_info = [x for x in epoch_info if x[1] != 0]
num_subjs = int(sys.argv[5])
# create a Searchlight object
sl = Searchlight(sl_rad=1)
mvs = MVPAVoxelSelector(data, mask, labels, num_subjs, sl)
clf = svm.SVC(kernel='linear', shrinking=False, C=1)
# only rank 0 has meaningful return values
score_volume, results = mvs.run(clf)
# this output is just for result checking
if MPI.COMM_WORLD.Get_rank()==0:
score_volume = np.nan_to_num(score_volume.astype(np.float))
io.save_as_nifti_file(score_volume, mask_image.affine,
'result_score.nii.gz')
seq_volume = np.zeros(mask.shape, dtype=np.int)
seq = np.zeros(len(results), dtype=np.int)
with open('result_list.txt', 'w') as fp:
for idx, tuple in enumerate(results):
fp.write(str(tuple[0]) + ' ' + str(tuple[1]) + '\n')
seq[tuple[0]] = idx
seq_volume[mask] = seq
# There are three random options:
# RandomType.NORANDOM is the default
# RandomType.REPRODUCIBLE permutes the voxels in the same way every run
# RandomType.UNREPRODUCIBLE permutes the voxels differently across runs
# example
#from brainiak.fcma.preprocessing import RandomType
#data, labels = prepare_searchlight_mvpa_data(images, conditions,
# random=RandomType.UNREPRODUCIBLE)
# the following line is an example to leaving a subject out
#epoch_info = [x for x in epoch_info if x[1] != 0]
num_subjs = int(sys.argv[5])
# create a Searchlight object
sl = Searchlight(sl_rad=1)
mvs = MVPAVoxelSelector(data, mask, labels, num_subjs, sl)
clf = svm.SVC(kernel='linear', shrinking=False, C=1, gamma='auto')
# only rank 0 has meaningful return values
score_volume, results = mvs.run(clf)
# this output is just for result checking
if MPI.COMM_WORLD.Get_rank()==0:
score_volume = np.nan_to_num(score_volume.astype(np.float))
io.save_as_nifti_file(score_volume, mask_image.affine,
'result_score.nii.gz')
seq_volume = np.zeros(mask.shape, dtype=np.int)
seq = np.zeros(len(results), dtype=np.int)
with open('result_list.txt', 'w') as fp:
for idx, tuple in enumerate(results):
fp.write(str(tuple[0]) + ' ' + str(tuple[1]) + '\n')
seq[tuple[0]] = idx
seq_volume[mask] = seq
# all MPI processes read the mask; the mask file is small
mask_img = nib.load(mask_file)
mask = mask_img.get_data()
epoch_info = None
if MPI.COMM_WORLD.Get_rank()==0:
files = [f for f in sorted(os.listdir(data_dir))
if os.path.isfile(os.path.join(data_dir, f))
and f.endswith(extension)]
for f in files:
img = nib.load(os.path.join(data_dir, f))
data = img.get_data()
raw_data.append(data)
logger.info(
'file %s is loaded, with data shape %s' % (f, data.shape)
)
epoch_list = np.load(epoch_file)
epoch_info = generate_epochs_info(epoch_list)
num_subjs = int(sys.argv[5])
# create a Searchlight object
sl = Searchlight(sl_rad=2)
mvs = MVPAVoxelSelector(raw_data, mask, epoch_info, num_subjs, sl)
# for cross validation, use SVM with precomputed kernel
# no shrinking, set C=10
clf = svm.SVC(kernel='rbf', shrinking=False, C=10)
# only rank 0 has meaningful return values
result_volume, results = mvs.run(clf)
# this output is just for result checking
if MPI.COMM_WORLD.Get_rank()==0:
result_volume = np.nan_to_num(result_volume.astype(np.float))
write_nifti_file(result_volume, mask_img.affine, 'result.nii.gz')
raw_data = []
# all MPI processes read the mask; the mask file is small
mask_img = nib.load(mask_file)
mask = mask_img.get_data().astype(np.bool)
epoch_info = None
if MPI.COMM_WORLD.Get_rank() == 0:
raw_data = io.read_activity_data(data_dir, extension)
epoch_list = np.load(epoch_file)
epoch_info = io.generate_epochs_info(epoch_list)
# the following line is an example to leaving a subject out
#epoch_info = [x for x in epoch_info if x[1] != 0]
num_subjs = int(sys.argv[5])
# create a Searchlight object
sl = Searchlight(sl_rad=1)
mvs = MVPAVoxelSelector(raw_data, mask, epoch_info, num_subjs, sl)
clf = svm.SVC(kernel='linear', shrinking=False, C=1)
# only rank 0 has meaningful return values
score_volume, results = mvs.run(clf)
# this output is just for result checking
if MPI.COMM_WORLD.Get_rank() == 0:
score_volume = np.nan_to_num(score_volume.astype(np.float))
io.write_nifti_file(score_volume, mask_img.affine,
'result_score.nii.gz')
seq_volume = np.zeros(mask.shape, dtype=np.int)
seq = np.zeros(len(results), dtype=np.int)
with open('result_list.txt', 'w') as fp:
for idx, tuple in enumerate(results):
fp.write(str(tuple[0]) + ' ' + str(tuple[1]) + '\n')
seq[tuple[0]] = idx
seq_volume[mask] = seq