ALFF

Submitted by krishna.dani on Wed, 05/05/2010 - 23:10

Dear RESTers

I am new to REST and to SPM. I would be grateful for any advice.

I have resting state fMRI data (3mins, TR 3s) and I am keen to look at the ALFF in a lesion and on the contralateral hemisphere to detect any differences. For each of my scans the lesion is in a different place, difference size, and may affect either hemisphere.

If I generate the ALFF map and calculate the mean ALFF value in lesion and in a mirror region is this valid or would higher level statistics be more appropriate?

Thanks

Krishna

Submitted by dongzy08 on Wed, 05/05/2010 - 23:50

I think it is not valid to compare ALFF in different places, because different places usually have different ALFF. You could use a mask contained all of the places and compared with them. This is only my suggestion.

Submitted by krishna.dani on Thu, 05/06/2010 - 23:02

Thanks for that.

My aim was not to compare ALFF between lesions; rather it is to determine if the lesion has any effect on ALFF. Looking for the best approach.

Cheers

Krishna

Submitted by ZangYF on Fri, 05/07/2010 - 07:49

It is a good idea to compare ALFF between a pair of homologous regions (voxels), or hemispheric asymmetry. I guess it is particularly important to locate high activity of epileptic focus. However, a common issue for brain lesion studies is that the location varies from case to case. It might be difficult to say if an asymmetric index (e.g., L-R, L/R, 2*(L-R)/(L+R)) is abnormal. You can try to download a large sample of dataset from http://www.nitrc.org/projects/fcon_1000/. You may try comparing your results with that of the large sample size. Although the scanning parameters are different from center to center, the asymmetric index may be considered as a stardardized measure.

Submitted by krishna.dani on Fri, 05/07/2010 - 15:22

This is really helpful; Thanks very much for this advice.

Krishna

Submitted by krishna.dani on Mon, 05/10/2010 - 06:23

This advice is very helpful. Is it reasonble to use the raw signal intensities from the ALFF map (ipsi and contra-lateral regions)  to calculate this asymmetric index?

Thanks, Krishna

Submitted by ZangYF on Mon, 05/10/2010 - 09:40

Hi Krishna,
I think it is OK (if you call it 'raw signal intensity:). Please not that asymmetric index can be calculated in at least two ways, "L-R" and "2*(L-R)/(L+R)". The statistical results will be differernt.

Yufeng

Submitted by krishna.dani on Fri, 06/25/2010 - 05:27

I am currently calculating the ALFF index from raw signal intensities on the ALFF maps.

Are the raw signal intensities in any particular units?

The reason I ask is that, surprisingly, the CSF appears to have amongst the brightest signal on the raw ALFF maps.

Submitted by YAN Chao-Gan on Sat, 06/26/2010 - 00:14

Hi!
This is why the fractional ALFF is proposed.
Please see details in:

J Neurosci Methods. 2008 Jul 15;172(1):137-41. Epub 2008 Apr 22.

An improved approach to detection of amplitude of low-frequency fluctuation (ALFF) for resting-state fMRI: fractional ALFF.

Zou QH, Zhu CZ, Yang Y, Zuo XN, Long XY, Cao QJ, Wang YF, Zang YF.

State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, PR China.

Abstract

Most of the resting-state functional magnetic resonance imaging (fMRI) studies demonstrated the correlations between spatially distinct brain areas from the perspective of functional connectivity or functional integration. The functional connectivity approaches do not directly provide information of the amplitude of brain activity of each brain region within a network. Alternatively, an index named amplitude of low-frequency fluctuation (ALFF) of the resting-state fMRI signal has been suggested to reflect the intensity of regional spontaneous brain activity. However, it has been indicated that the ALFF is also sensitive to the physiological noise. The current study proposed a fractional ALFF (fALFF) approach, i.e., the ratio of power spectrum of low-frequency (0.01-0.08 Hz) to that of the entire frequency range and this approach was tested in two groups of resting-state fMRI data. The results showed that the brain areas within the default mode network including posterior cingulate cortex, precuneus, medial prefrontal cortex and bilateral inferior parietal lobule had significantly higher fALFF than the other brain areas. This pattern was consistent with previous neuroimaging results. The non-specific signal components in the cistern areas in resting-state fMRI were significantly suppressed, indicating that the fALFF approach improved the sensitivity and specificity in detecting spontaneous brain activities. Its mechanism and sensitivity to abnormal brain activity should be evaluated in the future studies.

PMID: 18501969 [PubMed - indexed for MEDLINE]

Submitted by ZangYF on Tue, 06/29/2010 - 09:24

For the EPI data, it is true that the raw signal intensity is higher in CSF than in gray matter. But the ALFF may not be so.

If you are interested in comparing the assymetry, just do not care about the raw signal intensity. I will recommend ALFF or fALFF rather than the raw intensity.

Yufeng

Submitted by krishna.dani on Wed, 06/30/2010 - 05:09

Sorry, when I referred to the raw signal intensity, I meant th signal intensity in the ALFF map.

My question was was that on the ALFF map, the 'units' must be influenced by the scanner arbitrary units for the EPI data. I now understand that is why there is the mALFF.

The problem I have is in my lesioned brain data set, the lesions are all solitary and unilateral, but can be anywhere in either hemisphere. Therefore my understanding is that SPM single sample t-tests etc cannot be performed because all the brains are different. Therefore the approach I adopted was simply to take the ALFF map generated from REST, overlay a region of interest (usually about 50ml volume) and do the same for the mirror region form the other side. Then using there two values I took an asymmetery index. I just wanted to check this approach makes sense and that I can justifiably take the signal intensity from the ALFF map? Does this make sense?

I take the point about fALFF removing some of the noise from the ALFF maos in ventricles etc and this seems to work for my data.

Thanks so much

Krishna

Submitted by ZangYF on Thu, 07/01/2010 - 22:20

Hi Krishna,

Left/Right or (Left-Right) is a good index. If we have had a reference range obtained from a large sample of normal population, we can know whether an individual index is abnormal is that brain area.

The individual or within-group fALFF map do seem more meaningful (Zou et al., 2008). But for between-group comparison, we do not know which of mALFF and fALFF is better.

Yufeng