Generate modeled activation maps

For coordinate-based data, individual studies’ statistical maps are mimicked by generating “modeled activation” (MA) maps. These MA maps are used in the CBMA algorithms, although the specific method used to generate the MA maps differs by algorithm.

Start with the necessary imports

import os

import numpy as np
import matplotlib.pyplot as plt
from nilearn.plotting import plot_stat_map

import nimare
from nimare.tests.utils import get_test_data_path

Load Dataset

dset_file = os.path.join(get_test_data_path(), 'nidm_pain_dset.json')
dset = nimare.dataset.Dataset(dset_file)

Each kernel can taken certain parameters that control behavior

For example, the MKDA kernel accepts an r argument to control the radius of the kernel.

kernel = nimare.meta.cbma.kernel.MKDAKernel(r=2)
mkda_r08 = kernel.transform(dset)
kernel = nimare.meta.cbma.kernel.MKDAKernel(r=6)
mkda_r09 = kernel.transform(dset)
kernel = nimare.meta.cbma.kernel.MKDAKernel(r=10)
mkda_r10 = kernel.transform(dset)
kernel = nimare.meta.cbma.kernel.MKDAKernel(r=14)
mkda_r11 = kernel.transform(dset)

fig, axes = plt.subplots(nrows=2, ncols=2, figsize=(20, 10))
plot_stat_map(mkda_r08[2], cut_coords=[-2, -10, -4],
              title='r=2mm', vmax=2, axes=axes[0, 0],
              draw_cross=False)
plot_stat_map(mkda_r09[2], cut_coords=[-2, -10, -4],
              title='r=6mm', vmax=2, axes=axes[0, 1],
              draw_cross=False)
plot_stat_map(mkda_r10[2], cut_coords=[-2, -10, -4],
              title='r=10mm', vmax=2, axes=axes[1, 0],
              draw_cross=False)
plot_stat_map(mkda_r11[2], cut_coords=[-2, -10, -4],
              title='r=14mm', vmax=2, axes=axes[1, 1],
              draw_cross=False)
fig.show()

There are several kernels available

MKDA kernels convolve coordinates with a sphere and take the union across voxels. KDA kernels convolve coordinates with a sphere as well, but take the sum across voxels. ALE kernels convolve with a 3D Gaussian, for which the FWHM is determined by the sample size of each study.

kernel = nimare.meta.cbma.kernel.MKDAKernel(r=10)
mkda_res = kernel.transform(dset)
kernel = nimare.meta.cbma.kernel.KDAKernel(r=10)
kda_res = kernel.transform(dset)
kernel = nimare.meta.cbma.kernel.ALEKernel(sample_size=20)
ale_res = kernel.transform(dset)
max_conv = np.max(kda_res[2].get_fdata())
plot_stat_map(ale_res[2], cut_coords=[-2, -10, -4], title='ALE')
plot_stat_map(mkda_res[2], cut_coords=[-2, -10, -4], title='MKDA', vmax=max_conv)
plot_stat_map(kda_res[2], cut_coords=[-2, -10, -4], title='KDA', vmax=max_conv)

• 
• 
• 

Out:

<nilearn.plotting.displays.OrthoSlicer object at 0x7f0f791a6320>