Related Experiment Videos
Validation of statistical parametric mapping (SPM) in assessing cerebral lesions: A simulation study
E A Stamatakis1, M F Glabus, D J Wyper
1Department of Psychology, University of Stirling, Stirling, Scotland.
Neuroimage
|September 24, 1999
Summary
This study validates the use of Statistical Parametric Mapping (SPM) for analyzing SPECT lesion studies. SPM effectively detects hypoperfusion abnormalities, with optimal results achieved for decreases of 50% or more.
Area of Science:
- Neuroimaging
- Nuclear Medicine
Background:
- Single-photon emission computed tomography (SPECT) is crucial for assessing brain perfusion.
- Validating analysis methods like Statistical Parametric Mapping (SPM) is essential for accurate SPECT lesion interpretation.
Purpose of the Study:
- To validate SPECT lesion analysis using SPM by simulating abnormalities.
- To determine the sensitivity of SPM to hypoperfusion depth and lesion size.
- To identify optimal normalization, thresholding, and statistical probability settings for SPM analysis.
Main Methods:
- Two simulations were conducted: one varying hypoperfusion depth and another varying lesion size.
- Mean local intensity was altered to simulate hypoperfusion, and voxel counts simulated lesion size.
- Proportional scaling was evaluated as a normalization method, alongside ANCOVA for specific cases.
Main Results:
- Proportional scaling emerged as the most suitable normalization method for SPM.
- ANCOVA proved useful for large abnormalities when external normalization was unavailable.
- SPM demonstrated best performance with hypoperfusion decreases of approximately 50% or greater.
Conclusions:
- SPM is a validated tool for analyzing SPECT lesion studies, capable of detecting significant hypoperfusion.
- Optimal settings for SPM analysis include proportional scaling normalization, a grey matter threshold below 0.5, and a statistical probability peak threshold of p(u) = 0.01.