Diffusion models and metrics

There are currently four diffusion models that are supported by the ElikoPy pipeline. These models are listed below with their accessible parameters.

  • Diffusion Tensor Imaging (DTI)

  • Neurite Orientation Dispersion and Density Imaging (NODDI)

    • lambda_iso_diff - isotropic diffusivity for the CSF model

    • lambda_par_diff - axial diffusivity of the intra-neurite space

  • DIstribution of 3D Anisotropic MicrOstructural eNvironment in Diffusion compartment imaging (DIAMOND)

  • Microstructure Fingerprinting (MF)

    • dictionary_path - Path to the dictionary of fingerprints (mandatory)

    • CSD_bvalue - If the DIAMOND outputs are not available, the fascicles directions are estimated using a CSD with the images at the b-values specified in this argument.

study.dti()
study.noddi(lambda_iso_diff=3.e-9, lambda_par_diff=1.7e-9)
study.diamond()
study.fingerprinting(dictionary_path="my_dictionary", CSD_bvalue=None)

The metrics outputted by the functions are listed below.

Diffusion Tensor Imaging (DTI)

  • Fractional anisotropy (FA).

  • Axial diffusivity (AD).

  • Radial diffusivity (RD).

  • Mean diffusivity (MD).

  • Colored FA i.e. RGB map, a color is attributed to each voxel depending on the direction of the first eigenvalue and the intensity of the color depends on the FA value (fargb).

  • Residual (residual).

  • Eigenvectors of the diffusion tensor (evecs).

  • Eigenvalues of the diffusion tensor (evals).

  • Diffusion Tensor (dtensor).

Neurite Orientation Dispersion and Density Imaging (NODDI)

  • Thresholded intra-cellular volume fraction $nu_{ic}$ (icvf).

  • Fiber orientation dispersion index (odi).

  • Mean of the watson distribution of the Intra-cellular model (mu).

  • Fiber bundles volume fraction (fbundle).

  • Extra-cellular volume fraction (fextra).

  • Intra-cellular volume fraction (fintra).

  • Free water volume fraction (fiso).

  • Mean squared error (mse).

  • R squared (R2).

DIstribution of 3D Anisotropic MicrOstructural eNvironment in Diffusion compartment imaging (DIAMOND)

  • Tensor orientations of the fiber population 0 (t0).

  • Tensor orientations of the fiber population 1 (t1).

  • Residuals (residuals).

  • Intermediary step of the t0 output (mtm_t0).

  • Intermediary step of the t1 output (mtm_t1).

  • Intermediary step of the fractions output (mtm_fractions).

  • Volume with a null b-value (b0).

  • DTI estimate (dti).

  • Automose model selection map. Gray and white matter correspond to positive values and CSF to negative values (aicu).

  • Fraction of voxel attributed to each compartment (fractions).

  • Shape parameters of the mv-$Gamma$ distribution, Homogeneity index (kappa).

  • Log of kappa (logkappa).

  • The heterogeneity indexes defined as $H E I=frac{2}{pi} arctan left(frac{1}{kappa}right)$ (hei).

  • Number of fascicles by voxel (mosemap).

Microstructure Fingerprinting (MF)

  • Extra-axonal diffusivity of fascicle 0 (DIFF_ex_f0).

  • Extra-axonal diffusivity of fascicle 1 (DIFF_ex_f1).

  • Total extra-axonal diffusivity (DIFF_ex_tot).

  • Volume fraction of cerebrospinal fluid (frac_csf).

  • Volume fractions of fascicle 0 (frac_f0).

  • Volume fractions of fascicle 1 (frac_f1).

  • Fiber volume fractions of the fascicle 0 (fvf_f0).

  • Fiber volume fractions of the fascicle 1 (fvf_f1).

  • Total fiber volume fraction of all fascicles (fvf_tot).

  • Mask (M0).

  • Mean squared error (MSE).

  • Peak map fascicle 0 (peak_f0).

  • Peak map fascicle 1 (peak_f1).

  • R squared (coefficient of determination, square of Pearson) (R2).