Premium
SU‐E‐J‐167: Optimal Number of Respiratory Phases in 4D PET for Radiotherapy Planning: Motion‐Simulated Phantom Study
Author(s) -
Budzevich M,
Dilling T,
Zhang G,
Kuykendall C,
Moros E
Publication year - 2012
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4735006
Subject(s) - imaging phantom , bin , iterative reconstruction , medical imaging , physics , nuclear medicine , data set , spheres , mathematics , optics , computer science , computer vision , artificial intelligence , algorithm , medicine , astronomy
Purpose: Four‐dimensional PET/CT is increasingly used in radiotherapy treatment planning. One issue that is still under investigation is the optimal number of bins or phases into which the respiratory cycle needs to be divided. We performed 4D PET moving phantom study and compared results between 6 and 10 bins. Methods: A Jaszczak Phantom™ containing six hollow spheres (0.95 – 3.18 cm inner diameters) was used. Three sinusoidal motion patterns were accomplished with peak‐to‐peak amplitudes of 1, 1.5 and 2.0 cm with a respiratory period of 5 s. The background and the six spheres were filled with 18F‐FDG solution to achieve three SBR: 1:3.65, 1:7.95, and 1:10.22. Data were collected in 3‐D mode for 10 min. Images were reconstructed using OSEM reconstruction: 32 subsets with 2 iterations, variable FWHM Gaussian post‐filter 5–8 mm, and using image matrix sizes of 128×128, 192×192 and 256×256. The spheres were auto‐segmented using pre‐calculated optimal thresholds for a 1:1 volumetric correlation between actual‐ and PET‐delineated spheres from static phantom studies. Results: According to the static data the following set of parameters are optimal for static PET target (sphere) delineation: OSEM reconstruction with 32 subsets and 2 iterations, FWHM of 5 mm, and image size 256×256. The 4D data studies (with pre‐calculated optimal thresholds) have shown that a 6‐bin set shows less volume distortions then the 10‐bin set. Conclusions: The authors found that 6‐bin reconstruction is more reliable for delineation of a target in motion than 10‐bin reconstruction. Further investigation with optimal thresholds obtained from 4D data, not static, is required no conflict of interest for all authors.