Related Experiment Video
Updated: Jun 19, 2026
Photorealistic Learned Landscapes for Augmented Reality
Published on: June 27, 2025
Comparison between rebinning methods and geometric re-projection for the estimation of missing oblique data in 3D-PET
Fayçal Ben Bouallègue1, Jean-François Crouzet, Denis Mariano-Goulart
1Department of Mathematics, Montpellier Science University, Montpellier, France. faybenb@hotmail.com
Objective:
Classically, the missing portions in the oblique projection data acquired from cylindrical scanners are synthesized by forward projecting an estimation of the activity computed from non-oblique data. Another quick and powerful way to perform the estimation of the truncated portions consists in applying Fourier rebinning methods. This paper intends to compare the performances of 3D geometric re-projection versus two Fourier rebinning techniques (iterative FOREPROJ and extended FORE) designed for re-projection.
Methods:
Both analytical and noisy 3D projection data corresponding to a phantom constituted of warm and cold ellipsoids over a uniform background are simulated numerically for the SIEMENS Hi-Rez scanner. The object is reconstructed using 3D filtered back-projection. A prior estimation of the missing oblique data is performed using the three algorithms.
Results:
The resolution properties and noise behavior of the three methods are compared and demonstrate a significant signal to noise ratio improvement using rebinning-based algorithms.
Conclusions:
We show that iterative FOREPROJ provides projection data with high accuracy 60% less time than geometric re-projection.
