Table 1 The summary of powder-averaging techniques, different vector sets and different experiments conducted in this study.

From: Computing the orientational-average of diffusion-weighted MRI signals: a comparison of different techniques

 

Estimation method

References

1- Shell-by-shell estimation

1- Arithmetic averaging

6

2- Quadratures on the sphere (Lebedev)

31

3- Representation of signal in spherical harmonic

3

4- Representation of signal in Cartesian tensors

41

5- Knutsson

42

2- Estimation from the entire 3D data

1- MAP-MRI

44

2- MAP with Laplacian regularization

51

Point sets

Sampling scheme

References

Shelled

488 (61\(\times\)8) samples (Knutsson)

21

344 (43\(\times\)8) samples (Lebedev)

31

152 (19\(\times\)8) samples (Lebedev)

31

344 (43\(\times\)8) random samples

 

Non-shelled

488 samples (Knutsson)

21

344 samples (Knutsson)

21

152 samples (Knutsson)

21

344 random samples

 

Experiments

Sampling scheme

 

Effect of noise and number of samples

488 (61\(\times\)8) samples21 with Gaussian noise

Figure 1a

152 (19\(\times\)8)21,31 samples with Gaussian noise

Figure 1b

344 (43\(\times\)8)21,31 samples with Gaussian noise

Figure 2a

344 (43\(\times\)8)21,31 samples with Rician noise

Figure 2b

Effect of dispersion

344 (43\(\times\)8) samples21,31 with Gaussian noise

Figure 3

Crossing configuration

344 (43\(\times\)8) samples21,31 with Gaussian noise

Figure 4

Random sampling

344 (43\(\times\)8) samples with Gaussian noise

Figure 5a

Bias in gradient strength

344 (43\(\times\)8) samples21,31 with Gaussian noise

Figure 5b

Statistical analysis

344 (43\(\times\)8) samples21,31 with Gaussian noise

Figure 6

In vivo measurements

488 (61\(\times\)8) samples21

Figure 7