Hi Anderson,

Thanks very much for your answer. The image is of NON-corrected p-values. Unfortunately, no clusters exceed reasonable thresholds for corrected TFCE p-values.

I assume it is still not appropriate, isn't it?

Cesar

----------------------------------------------------------------------
Cesar Caballero
MRI engineer
www.bcbl.eu

Twitter: @caballerogaudes
www.researchgate.net/profile/Cesar_Caballero2

Legal disclaimer/Aviso legal/Lege-oharra: www.bcbl.eu/legal-disclaimer

---

From: "Anderson M. Winkler" <[log in to unmask]');" target="_blank">[log in to unmask]>
To: "FSL" <[log in to unmask]');" target="_blank">[log in to unmask]>
Sent: Monday, November 16, 2015 7:53:56 AM
Subject: Re: [FSL] minimum cluster size after TFCE?

Hi Cesar,

So you have an image of corrected TFCE p-values and there are some tiny regions surviving that are difficult to explain. It isn't appropriate to just remove them based on size -- they are significant after all. So, no, this shouldn't be done.

All the best,

Anderson

On 15 November 2015 at 23:06, Cesar Caballero <[log in to unmask]');" target="_blank">[log in to unmask]> wrote:

Dear all,

We are using TFCE statistics with randomise to define a set of ROIs for a subsequent analysis of resting state functional connectivity. Since one of the aims of TFCE is to avoid the specification of a minimum cluster size statistic, would it be correct to apply a minimum cluster size (e.g. 10 voxels) on the TFCE non-corrected maps?

Best,
Cesar






----------------------------------------------------------------------
Cesar Caballero
MRI engineer
www.bcbl.eu

Twitter: @caballerogaudes
www.researchgate.net/profile/Cesar_Caballero2

Legal disclaimer/Aviso legal/Lege-oharra: www.bcbl.eu/legal-disclaimer