Skip to content

Spin-off toolbox from SPM for fitting and comparing haemodynamic models (HDMs)

Notifications You must be signed in to change notification settings

pzeidman/HDM-toolbox

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

44 Commits
 
 
 
 
 
 
 
 

Repository files navigation

HDM-toolbox

Spinoff toolbox from SPM for developing, fitting and comparing single region haemodynamic models (HDMs).

NB 1: This is intended for situations where only a single brain region is being modelled. If data from multiple regions are available, then DCM is recommended (which has both hidden neural and haemodynamic states).

NB 2: This work is unpublished.

Instructions for running the demo

  1. Download the toolbox, unzip it and add the 'toolbox' directory to your Matlab path.
  2. In Matlab, go into the attention_example directory and run spm_hdm_demo.m

Function details

The code is arranged into a set of functions, mirroring the DCM framework:

Function Description
spm_hdm_specify Create an HDM from a subject's SPM.mat and a representative timeseries (VOI)
spm_hdm_estimate Fits an HDM to the data
spm_hdm_review GUI for inspecting an HDM

Default model

A default model is included, as introduced in Friston et. al 2000 and Stephan et al. 2007. It is specified in the following Matlab functions:

Function Description
spm_fx_hdm2 Neurovascular / haemodynamic model
spm_gx_hdm2 Observation model
spm_hdm_priors_hdm2 Priors on log scaling parameters as well as the default values they scale

Priors

The priors on parameters that can be estimated from the data are as follows:

Section Parameter Default Variance Notes
Neurovascular Efficacy 0Hz 1 Efficacy of stimuli
Neurovascular Decay 0.64Hz 1/32 Half-life of vasoactive signal: 1/decay x log(2)
Neurovascular Feedback 0.41Hz 0 Period of vasoactive signal: 1/(2 x pi x sqrt(1/feedback))
Vascular Transit 1.02Hz 1/32 Vascular transit rate
Vascular Alpha 0.33 0 Grubb's exponent
Vascular E0 0.40 0 Resting oxygen extraction fraction
BOLD V0 0.04 0 Resting venous blood volume fraction
BOLD Epsilon 0.72 (1.5T), 0.44 (3T), 0 (7T) 0 Ratio of intra-to extra-vascular signal contributions

Fixed parameters are as follows:

Section Parameter Value Notes
BOLD r0 25 (1.5T), 110 (3T), 325 (7T) Slope of intravascular relaxation rate against O2
BOLD nu0 28.265 * B0 Frequency offset at the outer surface of magnetized vessels

Notes:

  • The haemodynamic priors have been changed to be log scaling parameters that multiply default values. These default values come from Figure 7 of Friston et. al 2000. The resulting prior density over parameters can be viewed with the function attention_example/view_hdm_priors.m .
  • Parameters of the haemodynamic / observation models have been changed to be suitable for data of different field strengths (parameters epsilon, r0, nu0), based on recommendations from Heinzle et al. 2016 and Havlicek et al. 2015.

About

Spin-off toolbox from SPM for fitting and comparing haemodynamic models (HDMs)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages