Hi Stam,
Ignoring cross terms can introduce significance errors into the estimate of the diffusion coefficient, doesn’t it?
Thank you for your clarification.
Regards,
Suchada
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Stamatios Sotiropoulos
Sent: 18 May 2011 10:41
To: [log in to unmask]
Subject: Re: [FSL] How to calculate b-matrix
Hi Suchada,
the b-value is obtained by integrating the applied gradients. Depending on features of these gradients (e.g. their shape), the integral might be analytic or not. In the latter case, you need to integrate numerically, which is the case presented in the paper you mentioned.
In general if you already know the b-value, you can obtain the b-matrix via bmatrix = bvalue*gradient*gradient', gradient being a column vector. This is a simplification, as you may ignore cross-terms from imaging gradients, but is what we use in FSL.
Hope this helps,
Stam
On 18 May 2011, at 09:29, Suchada Tantisatirapong wrote:
Hi experts,
I’m trying to calculate b-matrices for DTI images. From
Dirk-Jan Kroon: http://www.mathworks.com/matlabcentral/fileexchange/21130-dti-and-fiber-tracking, he calculated b-matrix
as: bmatrix = bvalue*gradient’*gradient
He referred to the paper ‘The b matrix in Diffusion Tensor Echo-Planar Imaging’,
brr = 3.51 + 136.89G3r + 2228.52G3r^2
bpp = 3.19 +130.63G3p + 2228.52G3p^2
bss = 1.76 – 58.50G3s + 2228.52G3s^2
brp = bpr = 3.17+68.45G3s +65.31G3r+2228.52(G3r)(G3p)
bps = bsp = -1.84-29.25(G3p)+65.31G3s
brs = bsr=-1.93-29.25(G3r)+68.45G3s +2228.52(G3r)(G3s)
which I could not understand that how it is equivalent to his calculation.
The equations above are for 3D DTI data. if I have 2D DTI data, how can I calculate b-matrix?
Thank you very much
Regards,
Suchada