Premium
Evaluation of clip localization for different kilovoltage imaging modalities as applied to partial breast irradiation setup
Author(s) -
Buehler Andreas,
Ng SookKien,
Lyatskaya Yulia,
Stsepankou Dzmitry,
Hesser Jurgen,
Zygmanski Piotr
Publication year - 2009
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 - 10.1118/1.3075904
Subject(s) - imaging phantom , image quality , radiography , nuclear medicine , torso , computed radiography , iterative reconstruction , cone beam computed tomography , medical imaging , medicine , dosimetry , computer vision , computer science , radiology , computed tomography , image (mathematics) , anatomy
Surgical clip localization and image quality were evaluated for different types of kilovoltage cone beam imaging modalities as applied to partial breast irradiation (PBI) setup. These modalities included (i) clinically available radiographs and cone beam CT (CB‐CT) and (ii) various alternative modalities based on partial/sparse/truncated CB‐CT. An anthropomorphic torso‐breast phantom with surgical clips was used for the imaging studies. The torso phantom had artificial lungs, and the attached breast phantom was a mammographic phantom with realistic shape and tissue inhomogeneities. Three types of clips of variable size were used in two orthogonal orientations to assess their in‐/cross‐plane characteristics for image‐guided setup of the torso‐breast phantom in supine position. All studies were performed with the Varian on‐board imaging (OBI™, Varian) system. CT reconstructions were calculated with the standard Feldkamp–Davis–Kress algorithm. First, the radiographs were studied for a wide range of viewing angles to characterize image quality for various types of body anatomy in the foreground/background of the clips. Next, image reconstruction quality was evaluated for partial/sparse/truncated CB‐CT. Since these modalities led to reconstructions with strong artifacts due to insufficient input data, a knowledge‐based CT reconstruction method was also tested. In this method, the input data to the reconstruction algorithm were modified by combining complementary data sets selected from the treatment and reference projections. Different partial/sparse/truncated CB‐CT scan types were studied depending on the total arc angle, angular increment between the consequent views (CT projections), orientation of the arc center with respect to the imaged breast and chest wall, and imaging field size. The central angles of the viewing arcs were either tangential or orthogonal to the chest wall. Several offset positions of the phantom with respect to the reference position were studied. The acquired and reconstructed image data sets were analyzed using home‐built software focusing on the ability to localize clips in 3D. Streaking and leakage reconstruction artifacts and spatial distortions of breast surface were analyzed as well. Advantages and disadvantages of each kilovoltage CB imaging modality as applied to partial breast setup evaluation based on clips are presented. Because clips were found to be difficult to recognize in radiographs, 3D reconstructions were preferred. Even though it was possible to localize clips with about ±1 mm accuracy based on reconstructions for short arcs of 40° and incremental angle up to about 5°, identification of clips in such reconstructions is difficult. Reconstructions obtained for arcs of as low as 80° and incremental angle of as high as 3° were suggested for easier clip identification. For more severely undersampled data, iterative CB‐CT reconstruction is recommended to decrease the artifacts.