# Denoise images using Local PCA via empirical thresholds

PCA-based denoising algorithms are effective denoising methods because they explore the redundancy of the multi-dimensional information of diffusion-weighted datasets. In this example, we will show how to perform the PCA-based denoising using the algorithm proposed by Manjon et al. [Manjon2013].

This algorithm involves the following steps:

• First, we estimate the local noise variance at each voxel.

• Then, we apply PCA in local patches around each voxel over the gradient directions.

• Finally, we threshold the eigenvalues based on the local estimate of sigma and then do a PCA reconstruction

To perform PCA denoising without a prior noise standard deviation estimate please see the following example instead: denoise_mppca

import numpy as np
import nibabel as nib
import matplotlib.pyplot as plt
from time import time
from dipy.denoise.localpca import localpca
from dipy.denoise.pca_noise_estimate import pca_noise_estimate


Load one of the datasets. These data were acquired with 63 gradients and 1 non-diffusion (b=0) image.

img, gtab = read_isbi2013_2shell()

data = img.get_data()
affine = img.affine

print("Input Volume", data.shape)


## Estimate the noise standard deviation

We use the pca_noise_estimate method to estimate the value of sigma to be used in local PCA algorithm proposed by Manjon et al. [Manjon2013]. It takes both data and the gradient table object as input and returns an estimate of local noise standard deviation as a 3D array. We return a smoothed version, where a Gaussian filter with radius 3 voxels has been applied to the estimate of the noise before returning it.

We correct for the bias due to Rician noise, based on an equation developed by Koay and Basser [Koay2006].

t = time()
sigma = pca_noise_estimate(data, gtab, correct_bias=True, smooth=3)
print("Sigma estimation time", time() - t)


## Perform the localPCA using the function localpca.

The localpca algorithm takes into account the multi-dimensional information of the diffusion MR data. It performs PCA on local 4D patch and then removes the noise components by thresholding the lowest eigenvalues. The eigenvalue threshold will be computed from the local variance estimate performed by the pca_noise_estimate function, if this is inputted in the main localpca function. The relationship between the noise variance estimate and the eigenvalue threshold can be adjusted using the input parameter tau_factor. According to Manjon et al. [Manjon2013], this parameter is set to 2.3.

t = time()

denoised_arr = localpca(data, sigma, tau_factor=2.3, patch_radius=2)

print("Time taken for local PCA (slow)", -t + time())


The localpca function returns the denoised data which is plotted below (middle panel) together with the original version of the data (left panel) and the algorithm residual (right panel) .

sli = data.shape // 2
gra = data.shape // 2
orig = data[:, :, sli, gra]
den = denoised_arr[:, :, sli, gra]
rms_diff = np.sqrt((orig - den) ** 2)

fig, ax = plt.subplots(1, 3)
ax.imshow(orig, cmap='gray', origin='lower', interpolation='none')
ax.set_title('Original')
ax.set_axis_off()
ax.imshow(den, cmap='gray', origin='lower', interpolation='none')
ax.set_title('Denoised Output')
ax.set_axis_off()
ax.imshow(rms_diff, cmap='gray', origin='lower', interpolation='none')
ax.set_title('Residual')
ax.set_axis_off()
plt.savefig('denoised_localpca.png', bbox_inches='tight')

print("The result saved in denoised_localpca.png") Below we show how the denoised data can be saved.

nib.save(nib.Nifti1Image(denoised_arr,
affine), 'denoised_localpca.nii.gz')

print("Entire denoised data saved in denoised_localpca.nii.gz")

Manjon2013(1,2,3)

Manjon JV, Coupe P, Concha L, Buades A, Collins DL “Diffusion Weighted Image Denoising Using Overcomplete Local PCA” (2013). PLoS ONE 8(9): e73021. doi:10.1371/journal.pone.0073021.

Koay2006

Koay CG, Basser PJ (2006). “Analytically exact correction scheme for signal extraction from noisy magnitude MR signals”. JMR 179: 317-322.

Example source code

You can download the full source code of this example. This same script is also included in the dipy source distribution under the doc/examples/ directory.