Related Experiment Video
Updated: Feb 5, 2026

Meta-analysis of Voxel-Based Neuroimaging Studies using Seed-based d Mapping with Permutation of Subject Images SDM-PSI
Published on: November 27, 2019
Dynamic image denoising for voxel-wise quantification with Statistical Parametric Mapping in molecular neuroimaging
Stergios Tsartsalis1,2,3, Benjamin B Tournier1, Christophe E Graf4
1Division of Adult Psychiatry, Geneva University Hospitals, Geneva, Switzerland.
Denoising dynamic brain SPECT and PET images with Factor Analysis (FA) improves the detection of binding differences at the voxel level using Statistical Parametric Mapping (SPM). This noise reduction enhances quantification accuracy and reduces variability in binding potential estimations.
Area of Science:
- Neuroimaging
- Radiochemistry
- Biostatistics
Background:
- Positron Emission Tomography (PET) and Single-Photon Emission Computed Tomography (SPECT) imaging exhibit high noise levels in voxel kinetics.
- Detecting subtle differences in binding at the voxel level using Statistical Parametric Mapping (SPM) is challenging due to noise.
Purpose of the Study:
- To evaluate the impact of denoising on voxel-wise binding difference detection using SPM.
- To assess the efficacy of Factor Analysis (FA) in reducing noise in dynamic brain SPECT and PET images.
Main Methods:
- Simulated groups of images with 10% and 20% binding differences were denoised using FA.
- Binding potential (BPND) images were generated using simplified reference tissue model (SRTM) and Logan non-invasive graphical analysis (LNIGA).
- Denoised and raw images were analyzed using SPM for group differences; FA was also applied to clinical [123I]iomazenil (IMZ) and [11C]flumazenil (FMZ) images.
Main Results:
- FA-denoising improved the bias-precision profile for SRTM and LNIGA quantification.
- More simulated binding differences were detected in denoised images compared to raw images.
- Voxel-wise binding estimations on denoised clinical SPECT and PET images showed significantly reduced variability.
Conclusions:
- Noise removal from dynamic brain SPECT and PET images can optimize voxel-wise BPND estimations.
- Denoising facilitates the detection of biological differences using SPM.
- FA-based denoising is a promising technique for enhancing quantitative analysis in neuroimaging.
Related Concept Videos
Statistical Methods to Analyze Parametric Data: ANOVA
One-way ANOVA is applied when a single independent variable or factor is scrutinized. It compares...
Statistical Methods to Analyze Parametric Data: Student t-Test and Goodness-of-Fit Test
The Student's t-test is a statistical test that examines if there is a statistically significant difference between the means of two groups. This test is instrumental when dealing with...
Statistical Inference Techniques in Hypothesis Testing: Parametric Versus Nonparametric Data
Parametric statistics, as the name suggests, assumes that data follow a specific distribution, often a normal distribution. This assumption enables robust hypothesis testing and estimation. Parametric methods, like the Student's t-test or Goodness-of-fit test, are frequently employed in biostatistics due to their robustness. For instance,...
Statistical Significance
Probability in Statistics
An example of a simple event is a coin toss. The result of a coin toss is either a head or a tail. Here, head and tail are two simple events. These two simple events make up the sample space. Further, the probability of an event occurring falls within the range of 0 to 1. The probability of an...
Introduction to Statistics
In statistics, the collection of individuals or objects under study is called population. The idea of sampling is to select a portion of the larger population...

