Simple script to downsample niftis.

resample_nifti.py

#!/usr/bin/env python
import numpy as np
import nibabel as nib
import sys

def main(input, output, dx, dy, dz):
	img = nib.load(input)
	data = img.get_data()
	downsampling_factor = np.array([int(dx), int(dy), int(dz)])
	print('Input image is size {} {} {}'.format(data.shape[0], data.shape[1], data.shape[2]))

	new_data = np.zeros([int(np.ceil(data.shape[0]/float(downsampling_factor[0]))), int(np.ceil(data.shape[1]/float(downsampling_factor[1]))), int(np.ceil(data.shape[2]/float(downsampling_factor[2])))])
	print('Downsampling to size {} {} {}'.format(new_data.shape[0], new_data.shape[1], new_data.shape[2]))

	for ix in range(new_data.shape[0]):
		min_x = ix*downsampling_factor[0]
		max_x = (ix+1)*downsampling_factor[0]
		if ix == new_data.shape[0]-1:
			max_x = min_x + (data.shape[0] - downsampling_factor[0]*ix)
		for iy in range(new_data.shape[1]):
			min_y = iy*downsampling_factor[1]
			max_y = (iy+1)*downsampling_factor[1]
			if iy == new_data.shape[1]-1:
				max_y = min_y + (data.shape[1] - downsampling_factor[1]*iy)
			for iz in range(new_data.shape[2]):
				min_z = iz*downsampling_factor[2]
				max_z = (iz+1)*downsampling_factor[2]
				if iz == new_data.shape[2]-1:
					max_z = min_z + (data.shape[2] - downsampling_factor[2]*iz)
				new_data[ix, iy, iz] = data[min_x:max_x, min_y:max_y, min_z:max_z].mean()


	new_image = nib.nifti1.Nifti1Image(new_data, img.affine)
	nib.save(new_image, output)


if __name__ == "__main__":
	main(*sys.argv[1:])