Comparing sizes of identical .nii.gz data saved as int16, float32, float64

compression_check.py

import os
import nibabel as nb
import numpy as np

# Create an image with a "plausible" range of fMRI values
img = nb.Nifti1Image(np.random.uniform(500, 2000, (256,256,256,1)), np.eye(4))
# Verify numpy's giving us float64
assert img.get_data_dtype() == np.float64

# Quantize to 16 bits and reload
img.set_data_dtype(np.int16)
img.to_filename('int16.nii.gz')
del img
img = nb.load('int16.nii.gz')

# Save identical data with different representation
img.set_data_dtype(np.float32)
img.to_filename('float32.nii.gz')
img.set_data_dtype(np.float64)
img.to_filename('float64.nii.gz')

# Check that we did save the same data...
assert np.allclose(img.get_fdata(), nb.load('float32.nii.gz').get_fdata())
assert np.allclose(img.get_fdata(), nb.load('float64.nii.gz').get_fdata())

# Check sizes
def print_size(fname):
    stat = os.stat(fname)
    print(f"{fname}: {stat.st_size // 1048576}MiB")

print_size("int16.nii.gz")
print_size("float32.nii.gz")
print_size("float64.nii.gz")

output.txt

int16.nii.gz: 32MiB
float32.nii.gz: 53MiB
float64.nii.gz: 99MiB