How to verify functional connectivity difference in Parkinson's is not artifactual?

[Negar Memarian]

Dear group,

We are comparing average rs-fMRI connectivity (averaged across subjects) between Parkinson's patients and controls and noticed that the patients always show globally lower values than controls. This is unlike the published literature, which show a mix of lower and higher connectivity differences between Parkinson's and controls. This has made us wonder if what our results are showing might be potentially artifactual? We've calculated our pconn.nii matrices from the denoised HCP data (rest_AP_PA_Atlas_MSMAll_Test_hp2000_clean.dtseries), but wonder if differences in head motion or brain volume between groups could have affected the functional connectivity measures?

Any thoughts on this would be much appreciated.

Kind regards,
Negar

[Glasser, Matthew]

Are you using HCP-YA data as your controls?  Need to know much more about the Parkinson’s data (how was it acquired, preprocessed, analyzed?).

Matt.

--
You received this message because you are subscribed to the Google Groups "HCP-Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hcp-users+...@humanconnectome.org.
To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/CAC3GDx1s-2RqB-pm%2BJ%3DYELvzUgZFXzGiK5dbkcdyjusd-GVC6w%40mail.gmail.com.

 

---

The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.

[Negar Memarian]

Hi Matt,

Thank you for your reply. We acquired both control (HC) and Parkinson's (PD) data ourselves at UCLA Brain Mapping Center. We have a total of 57 PD and 28 HC subjects. The scanner model and data export protocol changed in the middle of our data collection and consequently our dcm2nii conversion software changed. I have listed the data acquisition parameters of both parts of our data (before and after scanner model change) in the table below. We then ran the HCP minimal preprocessing pipeline on all our data and parcellated the resulting denoised time series (rest_AP_PA_Atlas_MSMAll_Test_hp2000_clean.dtseries.nii) into a 13x13 functional network pconn matrices (based on the parcellation in Ragothaman et al., 2022 -  https://doi.org/10.1016/j.gaitpost.2022.07.003). The final goal of our study was to investigate if the network level association of rs-fMRI connectivity and task-related kinematic measures is different between patients and controls (MRI acquisition and kinematic task were performed on separate days). 

In light of this information, I would appreciate it if you kindly help me with my original question: how to verify if our PD data showing globally lower mean connectivity values compared to HC is a real biological effect or artifactual?

Parameter	Dataset 1(included both patients and controls)	Dataset 2 (included both patients and controls)
"Manufacturer"	"Siemens"	"Siemens"
"ManufacturersModelName"	"Prisma_fit"	"MAGNETOM Prisma Fit"
"MagneticFieldStrength"	3.0	3
"ProtocolName"	"T1w_MPR_vNav_4e"	"T1w_MPR_vNav_4e"
"InversionTime"	1	1
"FlipAngle"	52(functional) ; 8(structural)	52(functional);  8(structural)
"RepetitionTime"	0.8(functional); 2.5(structural)	0.8(functional); 2.5(structural)
"EchoTime"	0.037(functional); 0.00181(structural)	0.037(functional); 0.005386017797(structural)
"PhaseEncodingDirection"	"j-" for rest_acq-AP and "j" for rest_acq-PA	"j-" for rest_acq-AP and "j" for rest_acq-PA
"ParallelReductionFactorInPlane"	-	2
"TotalReadoutTime"	0.0597409	0.0597409
"MultibandAccelerationFactor"	8	8
"SliceTiming"	[0, 0.4325, 0.0875, 0.52, 0.1725, 0.6075, 0.26, 0.6925, 0.3475, 0, 0.4325, 0.0875, 0.52, 0.1725, 0.6075, 0.26, 0.6925, 0.3475, 0, 0.4325, 0.0875, 0.52, 0.1725, 0.6075, 0.26, 0.6925, 0.3475, 0, 0.4325, 0.0875, 0.52, 0.1725, 0.6075, 0.26, 0.6925, 0.3475, 0, 0.4325, 0.0875, 0.52, 0.1725, 0.6075, 0.26, 0.6925, 0.3475, 0, 0.4325, 0.0875, 0.52, 0.1725, 0.6075, 0.26, 0.6925, 0.3475, 0, 0.4325, 0.0875, 0.52, 0.1725, 0.6075, 0.26, 0.6925, 0.3475, 0, 0.4325, 0.0875, 0.52, 0.1725, 0.6075, 0.26, 0.6925, 0.3475]	[0.0000              0.4325 0.0850                0.5200 0.1725 0.6050                0.2600 0.6925 0.3450                0.0000 0.4325 0.0850                0.5200 0.1725 0.6050                0.2600 0.6925 0.3450                0.0000 0.4325 0.0850                0.5200 0.1725 0.6050                0.2600 0.6925 0.3450                0.0000 0.4325 0.0850                0.5200 0.1725 0.6050                0.2600 0.6925 0.3450                0.0000 0.4325 0.0850                0.5200 0.1725 0.6050                0.2600 0.6925 0.3450                0.0000 0.4325 0.0850                0.5200 0.1725 0.6050                0.2600 0.6925 0.3450                0.0000 0.4325 0.0850                0.5200 0.1725 0.6050                0.2600 0.6925 0.3450                0.0000 0.4325 0.0850                0.5200 0.1725 0.6050                0.2600 0.6925                0.3450]
"SliceThickness"	2(functional); 0.8(structural)	2(functional);  0.8(structural)
"BandwidthPerPixelPhaseEncode"	16.578	16.578

Many thanks,

Negar

To view this discussion visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/LV4PR02MB1105169835D0715D401BCBBF9D9E62%40LV4PR02MB11051.namprd02.prod.outlook.com.

[Glasser, Matthew]

I see.  For cleaning the data, did you use MR+FIX?  Usually, we use hp0 with that instead of hp2000.  If there is a difference in respiration between study groups, it may be helpful to use temporal ICA cleanup as well.  You may wish to use the initialize mode with a smaller group of less than 100.

[Negar Memarian]

Hi Matt,

Many thanks for your reply.  Yes, we used MR+IX. 

As for hp2000, we applied it after ICA-FIX to remove the absolute flat-line baseline drifts, trusting that the ICA-FIX identified and removed structured noise (like motion and respiration) across all frequencies.

Kind regards,

Negar

[Glasser, Matthew]

hp2000 is an incredibly slow way of doing hp0.