PPMI Resting State and T1 Standardization projects Darren Gitelman - - PowerPoint PPT Presentation

PPMI Resting State and T1 Standardization projects

Darren Gitelman (NU) Xue Wang (NU) Todd Parrish (NU) Petros Petrosyan (LONI) Ivo Dinov (LONI) Alen Zamanyan (LONI) Karen Crawford (LONI) Tanya Simuni (NU) John Seibyl (MNI) Arthur Toga (LONI)

Resting State Project

• Acquisition started in October 2012

– PD: 86 (17 subjects studied twice) – Control: 10 (1 subject studied twice) – SWEDD: 7

• All scans undergo stringent quality control

including measures of movement and noise.

– QC website built when LONI was at UCLA. – Rebuilt when LONI moved to USC – Currently undergoing redesign to add features.

Resting state project

PD (n=86) Mean ± StDev Control (n=10) SWEDD (n=7) Age 61.6 ± 10.25 65.1 ± 5.71 66.7 ± 8.06 Gender 25/84 F 1/10 F 3/7 F Scan QC: Good 39 6 3 Scan QC: Questionable 39 2 1 Scan QC: Bad 6 2 3

Resting State: Phantom QA pipeline

Resting State: Human QA Pipeline

Resting State Project: QC Main Page

Resting State Project: QC Rating

Resting State: Good Phantom QC

Resting State: Good Human QC

Resting State: Good Human QC

fd < 0.5 DVARS < 50

Resting state: Bad Human QC

Resting State Project: scans by site

0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 007 028 032 088 289 290 291 120

Percent total scans Site

Percent of total scans by site (n=121)

Resting State MRI: QC by Site

2 4 6 8 10 12 14 16 18 7 28 32 88 289 290 291 120

scan quality (number / site) Site

Scan quality by site (n=121) Good Questionable Bad

Resting State Project: Data Quality

• Take home Message

– Reduced data quality nearly always related to subject movement – Make subjects as comfortable as possible. – Cushion their head to minimize movement. – Instruct them often not to move, even before each scan (especially resting state and DTI)

Resting State Project: Upcoming changes

• Further improvements to QC process

– Faster feedback to sites

• Increased recruitment of controls

– Currently only 11 controls. – Amendment 8 will allow recruitment at years 2-4 for one scan.

Preliminary Group Level Analysis of Resting- State Functional MRI

(PD = 75, HC = 9) Matrices of t-values and corresponding p-values of group differences in pairwise correlations between brain regions are shown. P-values are FDR corrected

Within-group Functional Connectogram: Healthy Controls

Blue = positive t-value; red = negative t-value; line weight indicates t’s relative magnitude

Within-group Functional Connectogram: Parkinson’s Patients

Blue = positive t-value; red = negative t-value; line weight indicates t’s relative magnitude

T1 standardization

• T1 MRI scans not standardized across scan

sites

• T1 MRI scans not undergoing quality review
• Many scans were marred by artifact or the

variations in acquisition made them unusable for research.

• In order to maximize the research utility of all

the T1 scans it was proposed to utilize the published ADNI sequences at all scan sites.

T1 MRI: Bad data

Site No. Scanner Manufacturer Model Field strength (T) ADNI Protocol

001 GE Discovery MR750w 3 GE 750 006 Philips Ingenia Omega 3 Philips_3T 007 Siemens TrioTim 3 Siemens VB17_3T 012 Siemens TrioTim 3 Siemens VB17_3T 018 GE Optima MR450w 1.5 GE_16_15T 019 GE Signa HDxt 3 GE_15_3T 023 GE Signa HDxt 1.5 GE_16_15T 028 Siemens TrioTim 3 Siemens VB17_3T 032 Siemens TrioTim 3 Siemens VB17_3T 034 Siemens Espree 1.5 Siemens VB17_15T 040 GE Discovery MR750w 3 GE 750 057 Philips Achieva 3 Philips_3T 088 Siemens TrioTim 3 Siemens VB17_3T 089 GE Signa HDxt 3 GE_15_15T 096 Siemens TrioTim 3 Siemens VB17_3T 120 Siemens TrioTim 3 Siemens VB17_3T 154 Siemens Symphony 1.5 Siemens_V30_15T 289 Siemens TrioTim 3 Siemens VB17_3T 290 Siemens TrioTim 3 Siemens VB17_3T 291 Siemens Verio 3 Siemens VB17_3T 295 Siemens Verio 3 Siemens VB17_3T 304 Siemens Verio 3 Siemens VB17_3T 327 Philips Intera 1.5 Philips_15T 335 Philips Intera 1.5 Philips_15T 307 Siemens Trio 3 Siemens VB17_3T 002 GE Discovery MR750w 3 GE 750 334 GE Signa Excite 3 GE_16_3T 086 GE Signa HDxt 1.5 GE_16_15T

T1 standardization

• Use ADNI Protocols
• Detailed site instructions on acquisition

– Proper subject positioning – Screening for artifacts

• Centralized QC (at MNI)

– Scan parameters – Errors in subject positioning – Wrap Around – Motion artifacts – Signal loss – Inhomogeneity artifacts

• Centralized processing (at LONI)

– Correction of bias artifact (N4)