On 11/23/10, Bhavani shankara <[log in to unmask]> wrote:
> Dear Sir,
>
> Thank you very much for all the information, it is very useful
>
> On 11/23/10, Mark Jenkinson <[log in to unmask]> wrote:
>> Hi,
>>
>> Unfortunately we don't have any other correction method implemented
>> currently, although we are working on them. I cannot really comment on
>> the biological implications of any of this, and caution you not to
>> over-interpret
>> anything as a non-significant result does *not* imply that there was
>> nothing
>> happening (or that it was intrinsically weaker) - it could be that, or it
>> could just
>> be that the specific noise you had manifest as differences in sensitivity
>> that
>> were like this, without any connection to biology. This is especially
>> true
>> if
>> one result is just over and another just under the statistical threshold.
>> There
>> is normally no interpretation that can be made from such a result. If
>> you
>> are
>> interested in laterality differences then you need to formulate a
>> statistical
>> test that directly looks for this and does not compare two different
>> statistical
>> results, as that is itself not a statistically valid approach.
>>
>> All the best,
>> Mark
>>
>>
>> On 22 Nov 2010, at 15:53, Bhavani shankara wrote:
>>
>>> Thanks for the reply,
>>>
>>> is their any other method through which i can try other correction
>>> method
>>> (like FWE) to check the result. But on FDR uncorrected maps both
>>> hippocampus are showing difference between the groups.
>>>
>>> Is it may be due to biological reason of differential involvement in the
>>> structure of hippocampus between dominant and non-dominant hemispheres
>>> ??
>>>
>>> On Mon, Nov 22, 2010 at 9:16 PM, Mark Jenkinson <[log in to unmask]>
>>> wrote:
>>> Hi,
>>>
>>> This is perfectly reasonable for a couple of reasons:
>>>
>>> 1 - the volume change is not very spatially specific (or consistent)
>>> and hence you can have a strong overall volume change
>>> without a local vertex change (that is above threshold).
>>>
>>> 2 - the FDR thresholding is not the most sensitive technique for
>>> surface inference (we are working on coding up better ones)
>>> and so it is possible that it is just not quite sensitive enough
>>> although you might see interesting results in the uncorrected
>>> analyses. Unfortunately, the FDR results are very dependent
>>> on the small p-values which can be significantly influenced by
>>> slight outliers. It is worth looking at the uncorrected results
>>> to
>>> see if there is some consistent change which is just failing to
>>> meet the FDR threshold.
>>>
>>>
>>> I hope this helps.
>>> All the best,
>>> Mark
>>>
>>>
>>>
>>>
>>>
>>>
>>> On 22 Nov 2010, at 04:58, Bhavani shankara wrote:
>>>
>>> > Dear FSL experts,
>>> >
>>> > I am using FSL-FIRST, on extracting the Volumes and statistics were
>>> > done
>>> > on R software and found the subcortical structures and also both left
>>> > and right hippocampus were significantly differing between the groups
>>> > (AD and controls (all are right handed) ) (p<0.01).
>>> >
>>> > However on vertex analysis only on left side hippocampus is not
>>> > differing between the groups standing out at FDR correction (on using
>>> > surface_fdr input.vtk). However other structures including the
>>> > sub-cortical deep gray matter regions were differing at FDR threshold
>>> > output (p value) below 0.05.
>>> >
>>> > Is it possible ?? is it due to technical/analysis issue or due to
>>> > biological reason??
>>> >
>>> > Thanks in advance for your response
>>> >
>>> > --
>>> > Dr. BHAVANI SHANKARA BAGEPALLY
>>> > MBBS, (PhD in Clinical Neuroscience) NIMHANS,
>>> > Bangalore,
>>> > INDIA-560029
>>> > email: [log in to unmask]
>>> >
>>>
>>>
>>>
>>> --
>>> Dr. BHAVANI SHANKARA BAGEPALLY
>>> MBBS, (PhD in Clinical Neuroscience) NIMHANS,
>>> Bangalore,
>>> INDIA-560029
>>> email: [log in to unmask]
>>>
>>
>
>
> --
> Dr. BHAVANI SHANKARA BAGEPALLY
> MBBS, (PhD in Clinical Neuroscience) NIMHANS,
> Bangalore,
> INDIA-560029
> email: [log in to unmask]
>
--
Dr. BHAVANI SHANKARA BAGEPALLY
MBBS, (PhD in Clinical Neuroscience) NIMHANS,
Bangalore,
INDIA-560029
email: [log in to unmask]
|