Issues in voxel-based analysis of PV-corrected data - PowerPoint PPT Presentation

Loading...

PPT – Issues in voxel-based analysis of PV-corrected data PowerPoint presentation | free to download - id: 6dda03-ZjY1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Issues in voxel-based analysis of PV-corrected data

Description:

Issues in voxel-based analysis of PV-corrected data Mario Quarantelli Biostructure and Bioimaging Institute CNR Naples - Italy HBM2004 - PVEOut Satellite Meeting – PowerPoint PPT presentation

Number of Views:6
Avg rating:3.0/5.0
Slides: 11
Provided by: csv62
Learn more at: http://pveout.ibb.cnr.it
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Issues in voxel-based analysis of PV-corrected data


1
Issues in voxel-based analysisof PV-corrected
data
Mario Quarantelli Biostructure and Bioimaging
Institute CNR Naples - Italy HBM2004 -
PVEOut Satellite Meeting Budapest, 12 June 2004
2
  • Previous studies with voxel-based PV-C images
    have used voxel-based analysis to detect
    differences between patients and controls
  • Ibanez V, et al. Neurology 1998 501585
  • Bokde AL, et al. 200158480
  • Sakamoto S, et al. J Neuroimaging 2003
    13113-123
  • Matsuda H, et al. Eur J Nucl Med Mol Imaging
    2004291502

3
Smoothing
  • Voxel-based analysis, in its most diffuse
    implementations, involves smoothing of the images
  • Inherent to the warping step, performed to adapt
    the study to a standard normalized space (e.g.
    normalization step in SPM)
  • Added to reduce noise and to guarantee that
    residuals have a normal distribution, to apply
    statistical tests

4
Smoothing
  • Typically, the analysis of uncorrected PET / SPET
    data will highlight regions in which the overall
    tracer content is decreased, as a consequence of
    either GM loss and/or reduced tracer uptake
  • In its simplest conception, the aim of PV-C is to
    detect areas with a tracer concentration lower
    than what we would expect based on the
    corresponding degree of atrophy
  • The local re-distribution of the activity, as a
    consequence of smoothing, may re-introduce a
    difference in areas with atrophy
  • (theoretically, if maximum warping is allowed
    during the normalization process, such that the
    GM of the patients is deformed to perfectly match
    the GM of the template, this would solve the
    problem)

5
Testing
  • Three sets of simulated PET
  • 10 NV segmented MR studies brought to PET voxel
    size (2.03x2.03x2.43mm), with GM255, WM64
  • Same 10 studies with -20 activity in Left
    Parietal
  • Same 10 studies with -23.7 GM in Left Parietal
    (using an erosion algorithm)
  • All studies PV-C using mMG in PVELab

6
SPM Analysis
  • 4 comparisons
  • Uncorrected NV vs. Hypo
  • Uncorrected NV vs. Atrophic
  • PV-corrected NV vs. Hypo
  • PV-corrected NV vs. Atrophic
  • Unpaired T-Test
  • No Grand-mean scaling
  • Threshold masking fixed (20)
  • Global calculation Mean Voxel Value
  • W/O Non-sphericity correction
  • Uncorrected Plt0.001
  • No restriction to voxel size

7
Original
-20
-31 GM
8
(No Transcript)
9
(No Transcript)
10
Preliminary results
  • In simulated atrophy, smoothing partly
    re-introduces PVE
  • Reduction in cluster size when applying PVC to
    simulated hypometabolism will result mainly from
    zeroing of non-brain voxels in PVC images
    (increased noise in PVC images may also play a
    role)
  • Alternative analysis techniques should be
    exploited
  • Group X Modality Interaction (Richardson MP,et
    al. Brain (1997), 120, 19611973)
  • Voxel-based analysis of uncorrected data ?
    definition of VOI ? confirmation of
    hypometabolism with VOI-based PVC
About PowerShow.com