Biophysical models

All available biophysical models are variants of the “Standard model” (see for example) and estimate tissue specific biophysical parameters. In ACID currently there are three biophysical models available:

1. NODDI-DTI:

This model neglects the cerebrospinal fluid compartment and estimates the neurite density and neurite dispersion based on the DTI parameters FA and MD. The model inputs are:

• FA map: the DTI based FA map
• MD map: the DTI based MD map
• Eigenvalues (3maps): the three DTI tensor eigenvalue maps
• Region of interest: a region of interest confining the analysis
• b-values: the b-values used in the DTI acquisition
• In vivo or ex vivo: a switch for setting the tissue state that is being imaged

2. WMTI-Watson:

This model is more general than the NODDI-DTI model but needs the 5 DKI based axisymmetric DKI tensor metrics axial and radial diffusion and kurtosis and mean of the kurtosis tensor to work. The model inputs are:

• Maps (AD,RD, MW/MK, AW/AK,RW/RK) (order is mandatory): the axisymmetric DKI tensor metrics estimated with a previous DKI fit
• Type of input maps: a switch for telling the algorithm whether you loaded the apparent kurtosis parameters (AK, RK and MK) or the kurtosis tensor based parameters (AW, RW and MW)
• Region of interest: a region of interest confining the analysis
• Biophysical branch: a switch for branch selection (since the WMTI-Watson model has more than one solution (“branches”))
• Number of workers (parallel programming): an option to set the number of workers for parallel execution of the code speeding up parameter estimation

3) WMTI (found in the external tools section):

Please cite [Fieremans, Els, Jens H. Jensen, and Joseph A. Helpern. White matter characterization with diffusional kurtosis imaging, Neuroimage 58.1 (2011): 177-188.] when using this parameter since it is an external implementation of the WMTI model that is similar to the WMTI-Watson model but neglects axon dispersion and assumes perfectly parallel axons. The model inputs are:

• Diffsuion Tensor: the diffusion tensor estimated from a previous DKI fit
• Kurtosis Tensor: the kurtosis tensor estimated from a previous DKI fit
• Mask (binary): a region of interest confining the analysis

Note that for diffusion and kurtosis tensor estimation “acid_def.diffusion.dummy_write_tensors” needs to be set to 1 in the defaults for ACID to write out both the diffusion and kurtosis tensor in the DKI fit.