Questions regarding tract dissection and parameter settings

[Yusheng Wang]

Hi Frank,

I am currently dissecting the SLF I for one of our healthy participants. I want to get its FA and MD values. The attached image shows my dissection results. As you can see, the left SLF I contains only a very small number of streamlines compared to the right side.

For preprocessing, in Step T2, I applied top-up and eddy correction. In Step T2b, I chose QSDR with 2.0 mm resolution (to register to MNI space), a diffusion sampling length ratio of 1.25, and set “ignore high b” to on.

I have been stuck on this issue for quite a while and have tried different parameter settings, but I am still unsure what would be the most appropriate approach. I would appreciate your advice on the following:

1. In this participant, the left SLF I yielded only 38 streamlines. When I lowered the Min length from the default (30mm) to 20 mm, the number increased to 3,497 streamlines. However, the FA values changed only slightly (0.368 → 0.365). In such a case, can the FA value still be considered representative of the SLF I?
2. When the tract does not have the expected shape, is it necessary to adjust parameters to correct it? Or does this mainly affect the visual presentation without significantly affecting the quantitative metrics?
3. My goal is to perform a group analysis, averaging FA values for the same tract across participants. Currently, in Step T3c (Tracking Parameters), I only changed the Tract to voxel ratio to 5, leaving other settings as default. For cases like this participant, is it acceptable to adjust certain parameters (e.g., Min length) on an individual basis, or should all tracking parameters be kept identical across subjects for group analysis?
4. For Max seeds, I am currently using the default. I noticed that increasing the number of seeds tends to produce more stable FA results, but increases computation time. Previously, I used 100,000 seeds for faster processing, but found that repeated dissections sometimes produced more variation in FA values. I have reviewed the tutorials and videos, but would like to know if you have a recommended seed number for stable results.

I also uploaded the processed data to Dropbox, just in case you might want to look into it. Thank you very much for your help. :)

Best,

Yu-Sheng 

[Frank Yeh]

My apologies for the delayed response because this is a nontrivial situation.

This particular task is quite challenging because the SLF I and
cingulum are located very close to each other. It's likely that a
significant portion of what might appear to be a 'missing' bundle is
actually being identified as part of the cingulum, specifically its
frontal parietal section. The definition of cingulum and SLF 1 is
preliminary determined by the end location, and thus sometime this
situation happens.

There are a couple of potential reasons for this issue:
1. It's possible the tracking algorithm might be unintentionally
overlooking the SLF I, especially if there's a notable turning in the
subject's brain shape.
2. Another possibility is that the subject might naturally have a
smaller SLF I, and thus a closer easy target, cingulum, may be over
mapped here.

My two cents is to also include cingulum in your case and examine
whether this is the case.

Best,
Frank

> --
> You received this message because you are subscribed to the Google Groups "DSI Studio" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/dsi-studio/b4fe1607-cd9a-4991-add8-82016413fa4cn%40googlegroups.com.

[Yusheng Wang]

Thank you so much, Frank!! I really appreciate it :)