Related Experiment Video
Updated: Jul 9, 2026
11:09
High-Resolution Cardiac Positron Emission Tomography/Computed Tomography for Small Animals
Published on: December 16, 2022
Deconvolution-based partial volume correction in Raclopride-PET and Monte Carlo comparison to MR-based method
Jussi Tohka1, Anthonin Reilhac
1Institute of Signal Processing, Tampere University of Technology, Tampere, Finland. jussi.tohka@tut.fi
Neuroimage
|December 14, 2007
Summary
Deconvolution methods offer an accurate alternative to partial volume (PV) correction in brain PET imaging, showing similar results to structural imaging methods but with less susceptibility to errors. The reblurred Van Cittert algorithm performed best among tested deconvolution techniques.
Area of Science:
- Nuclear Medicine
- Neuroimaging
- Medical Physics
Background:
- Quantitative analysis in brain PET imaging is crucial for understanding neurodegenerative diseases.
- Partial Volume (PV) effects introduce significant errors in PET quantification.
- Existing PV correction methods, like the GTM method, rely on structural imaging and are sensitive to segmentation and registration errors.
Purpose of the Study:
- To evaluate and compare the performance of three iterative deconvolution algorithms against structural imaging-based PV correction in brain PET.
- To assess the accuracy of regional binding potential (BP) values and time-activity curves (TACs) using simulated (11)C-Raclopride striatal imaging.
- To investigate the resolution/noise tradeoff in parametric BP images derived from deconvolution.
Main Methods:
- Monte Carlo-simulated brain PET images were used for evaluation.
- Three deconvolution algorithms were tested: Richardson-Lucy, reblurred Van Cittert, and reblurred Van Cittert with total variation regularization.
- Performance was compared to a structural imaging-based PV correction (GTM method) using simulated (11)C-Raclopride data.
Main Results:
- Deconvolution methods, particularly reblurred Van Cittert, achieved accuracy comparable to the GTM method when using ideal, slightly eroded volumes of interest.
- Deconvolution-based PV correction demonstrated greater robustness against segmentation and registration errors compared to the GTM method.
- Parametric BP images from deconvolution showed improved resolution with minimal noise increase.
Conclusions:
- Iterative deconvolution presents a viable alternative to structural imaging-based PV correction in brain PET.
- Deconvolution offers similar quantitative accuracy while being less susceptible to common image processing errors.
- Deconvolution facilitates the computation of PV-corrected parametric images from deconvolved dynamic PET data.
