I think being normal is the best. Too high and too low are both harmful. 

Thank you Xiang Zhen Khong.  Would you mind explaining further. We all understand that conventionally a reduction in the FA value  would mean some amount of disruption in the brain fibre/ tract. But the question here is how do we explain the increased FA value with reduced performance in the neuropsychological tests?

I think the following laterature may help, since there is some discussion about the FA-poor performance correlation. 

http://www.pnas.org/content/102/34/12212.full

Did you compare MD as well ?

It might give more indication of what exactly is different when considering other elements.

Thanks Xiang for the article. Appreciated.

Aside to Taylor Winter,

Yes, in this particular study, we found that for certain tracts, the FA values increased, and the MD decreased at admission and these mTBI patients also had corresponding poor domain specific cognitive performance at admission.  As previously mentioned , the literatures are rather equivocal about it. (Kou et al, 2010, Henry et al 2011, Bazarian et al 2007). Besides cytotoxic edema causing the FA value to be rather increased,  what else could be the possible explanation.

While FA is a commonly used parameter, it does not capture all aspects of diffusion.  One possible explanation for a (perhaps counterintuitive) increase in FA can apply in regions of crossing fiber tracts.  Crossing fiber tracts can result in a voxel with restricted diffusion that is relatively isotropic.  That mean a lower FA value than if there was only one of the fiber tracts there.  So if one of the fiber tracts later provides less of a restrictive barrier, then the FA value for the voxel could be increased.

It depends a lot on the area you are finding FA increase. I would think it is an area with crossing fibers. So here is the story why: when two fibers cross the same voxel, FA goes down. One fiber may degenerate (or demyelinate) and the orientation in the voxels remains more unidirectional. What happens is that you see FA increase, but is in fact a loss of structural orientation in a subpopulation of fibers within the area.

Now you may ask: what proportion of voxels in the brain have crossing fibers? The most recent paper says 90%. The one before that says 65%. The area with more obvsious crossing fibers is the corona radiata (see figure 5 in Mori et al atlas 2010, 2nd edition). 

Are you the first to find such paradoxic correlation? No.

There is a paper from Hoeft 2006 (titled something like "More is not always better"). Then there are other papers on Alzheimer and multiple sclerosis, both showing FA drop in areas with unique fibers, but "magical" increases in the corona radiata.

So, I know how fantastic is if FA would correlate with behavioral performance, but this is not at all straightforward, depends which tract is it, where is it passing, what is it correlating with. FA is an outdated and unspecific measure that does not tell much except that something down there is not well oriented (or is too well oriented). That something may be a lot of things, myelin, axons, thicker and thinner axons, glial cells, multiple fibers, etc. It tells very little about white matter, reason why some authorities argue against ever using FA as a synonim with white matter integrity.

Factoring in your reduced MD supports Dorians explanation above. One of two crossing fibres could degenerate causing increased FA but the MD has decreased (although is by no means the definite cause).

In this particular ongoing study of mine, I have the medial lemniscus , superior occipital frontal fasiculus, and splenium of corpus callosum  showing increased FA, reduced MD and impaired domain specific cognition at admission. It makes make a lot of sense now , especially when you factor in the limitation of the DTI in crossing fibres and all the above have crossing fibres at certain point. 

Don't know at which anatomical level you see the increase, but the splenium does not have much crossing fibers. Is actually one of the purest single orientation regions, if measured straight at the midline.

There is a measure called "mode of anisotropy", which is used as a marker of crossing fibers. Mode is not as established but may be useful to understand effects caused by crossing fibers when used to complement FA findings.

Thank you so much for your suggestions Dorian Pustina. I will definitely take your suggestion into account. Truly appreciate your kind gesture in lending your expertise in this area.