Dear Adrian,

This would not be a deformation per se (as denoising does not modify the anatomy; the mean functional image across all timepoints will be exactly the same before or after denoising) but I imagine the misalignment that you describe could possibly arise from an incorrect analysis mask. The analysis mask is defined in the Setup.Options tab; typically this is defined automatically to point to a standard brainmask in MNI-space, but it can also be defined differently to match the proper space of your data. In general the denoised data will be masked to only include brain voxels but if the analysis mask is incorrectly defined this could cause a mismatch between the functional and anatomical data. 

That said, your Figure 1 connectivity scatter plots also seem to show a very-strange lack of spatial autocorrelation in your data  before denoising (over short distances voxels would tend to show higher connectivity, but I do not see that in your connectivity-vs-distance plots). This is perhaps unrelated but I would double-check the data before denoising to make sure that everything looks fine there (if in doubt send me a functional run and I will take a closer look)

Best regards

Alfonso

Originally posted by adrianrc:

Dear CONN users,

I am encountering an issue with my preprocessing/denoising pipeline that I cannot explain.

---

Dataset:

• Structural images: T1.nii (256×256×256, 1×1×1 mm)


• Functional images: fmri.nii (96×96×54×400 volumes, 2.5×2.5×2.5 mm)


• N = 100 subjects

---

Pipeline:

• Default CONN preprocessing pipeline


• No custom modifications

---

Preprocessing results:

• Functional (smoothed, MNI space): swaufmri.nii (91×109×91×400, 2×2×2 mm)


• Structural (MNI space): wc0T1.nii (181×217×181, 1×1×1 mm)

The preprocessing step completes successfully for all subjects with no errors.

Importantly:

• swaufmri.nii and wc0T1.nii are perfectly aligned for every subject


• Same origin


• Alignment verified using SPM CheckReg and MRIcroGL

---

Issue:
The problem appears only at the denoising step:

1. In the CONN denoising viewer:
   
   
   
   • Functional and structural images appear strongly misaligned (Fig. 1)
   
   
   • In some cases, the functional images look visibly deformed
   
   
   
   
   
2. After denoising:
   
   
   
   • The resulting dswaufmri.nii images appear distorted (Fig. 2)
   
   
   • This deformation was not present in any earlier preprocessing outputs
   
   
   
   
   
3. No error messages are produced at any stage
   
4. Final connectivity results are clearly invalid
   

FIGURE 1

 

FIGURE 2

---

Importantly:

• All preprocessing outputs (including swaufmri.nii) look correct outside CONN (Fig. 3)


• The issue seems to appear only within CONN at the denoising stage and in the resulting dswaufmri.nii files

FIGURE 3

---

Questions:

• Are there known issues that could produce deformation at this stage without triggering errors?

---

Any suggestions or similar experiences would be greatly appreciated.

Thank you very much for your help.