Hi Derek,
With apologies for waffling a bit, I think diffusion MRI is limited by:
1. Acquisition time. DTI and particularly HARDI are time-consuming to acquire, and it can be very tough to convince clinicians to include them on clinical protocols as a matter of course.
2. Processing complexity and the lack of consensus. Diffusion analysis methods are developed much faster than they are used, partly because a bunch of geeky computer science types like me have jumped on the bandwagon, but largely because even a straightforward analysis pipeline presents too many decisions to be made (acquisition sequence, preprocessing, diffusion model, tractography algorithm, etc. etc.) and knobs to twiddle for most people. And we techies can't decide on best practices.
3. Lack of specific outcome measures. There's plenty of life in FA and MD yet, no doubt, but they lack specificity and do not have a simple interpretation. (Task) fMRI may only have one outcome measure, but the infinite spectrum of possible tasks gives it substantial breadth of scope. Diffusion, by contrast, can only give you a limited amount of information about the brain (or muscle, etc.). And you have to work hard to get it (viz. points 1 and 2).
Cheers,
Jon
On 23 Oct 2011, at 11:49, Derek Jones wrote:
> Hi Folks
> I'm mailing for your assistance. On Tuesday/ Wednesday, I'm giving a talk at the Wellcome Trust for a meeting called 'Next Generation Imaging Technologies'. Part of the aim of the meeting is to see where things are / ought to be going next. I've been asked to speak about diffusion MRI. Now, if i were to say "To be honest, we have all the technology that we need for the next 10 years" - that
>
> (a) wouldn't be true
> (b) would not help shape funding decisions / strategies for the Trust.
>
> So - thinking about the implications of the Trust hearing an consensus from the UK imaging community, I am mailing to ask you for your knee-jerk reactions.....
>
> What do you think is currently hindering progress with diffusion MRI in understanding the brain.
>
> I don't want to prime you too much - but I invite you to provide your 'top 3' (and I'd be really grateful if you could do this a.s.a.p.).
>
> Here i'm thinking about various factors - such as field strength, gradient performance, modelling limitations, sensitivity, specificity, antipodally-symmetrical nature of diffusion, acquisition times - but - I'd like you to let me know your 'Top 3'
>
> I think it would look pretty good that the diffusion community had got together and voted (albeit extremely last-minute.com - for which I apologise).
>
> hoping that lots of you can see the value of contributing and thanks in advance to those that do!
>
> Derek
>
>
> ______________________________
> Derek K Jones
> CUBRIC
> School of Psychology
> Cardiff University
> Park Place
> Cardiff
> CF10 3AT
> Tel: +44 (0)29 2087 9412
> Fax: +44 (0)29 2087 0339
> Web: http://www.cf.ac.uk/psych/cubric/index.html
>
|