Premium
SU‐E‐T‐95: Imaging Protocol Investigations with a Fan‐Beam Optical CT Scanner for 3D Dosimetry
Author(s) -
Campbell W,
Jirasek A,
Wells D
Publication year - 2011
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 - 10.1118/1.3612046
Subject(s) - scanner , dosimeter , noise (video) , dosimetry , optics , projection (relational algebra) , blank , nuclear medicine , image quality , physics , cone beam computed tomography , image noise , materials science , mathematics , computer science , computed tomography , computer vision , medicine , algorithm , image (mathematics) , radiology , composite material
Purpose: To investigate data acquisition protocol choices and their effects on image noise for a prototype fan‐beam optical CT scanner intended for evaluating 3D dosimeters. Methods: The prototype scans dosimeters submerged in a refractive index‐matching bath. For all tests, uniform samples of scattering solution (synthetic polymer in water) were housed in cylindrical containers (1‐litre, 95 mm diameter) and scanned in a water bath. for reference data (I_o), options evaluated were: i) a projection through the bath, ii) an average projection through a “blank” (water‐filled container), and iii) an entire sinogram through a “blank”. Scan parameters considered were: i) temporal delay between I_o and I scans, ii) number of projections per rotation, iii) number of samples averaged per projection, and iv) vertical displacement between I_o and I slices. Images were reconstructed using filtered backprojection. Relative noise (st.dev./mean) in a consistent region of interest was used to quantify image quality. Results: Using a reference scan through a “blank” as opposed to one through only a water bath eliminated major ring artefacts. Furthermore, using a full I_o sinogram through a “blank” instead of an averaged profile considerably reduced noise (22.4% to 3.1%). Minimal increases in noise were seen when delaying scans (3% to 5%, one hour later). Increases in noise were seen with fewer than 360 projections per rotation; with > 360 projections, no significant reductions in noise were seen (up to 1800). No benefit was found by acquiring multiple samples per projection. Significant noise increase was seen when I_o and I slices were displaced by even 1 mm (2.6% to 15.5%). Conclusions: Noise reduction is optimized when comparing scan sinograms ray‐by‐ray to full sinograms of a “blank”. Main sources of image noise were attributed to variations in the surface quality of the containers used. Funding from the Canadian Institutes of Health Research and the University of Victoria supported portions of this work.