Publications [#324794] of Ehsan Samei

Papers Published
  1. Solomon, J; Zhang, Y; Marin, D; Samei, E, SU-G-206-13: Validating Dose Split: A Method to Image the Same Patient at Multiple Doses with a Single CT Acquisition., Medical physics, vol. 43 no. 6 (June, 2016), pp. 3642 [doi] .

    Dose optimization studies in CT have a need for images acquired at multiple dose levels. This presents ethical and logistical challenges for performing such research with human subjects. As a result, most studies rely on phantom data. The purpose of this study was to perform a physics-based validation of a method to obtain images of the same patient at multiple dose levels using a single CT acquisition on a dual-source CT system.The Dose Split (DS) method relies on acquiring raw projection data simultaneously from two separate sources/detectors (denoted Tube A and Tube B) on a dual-source system. By distributing the dose unevenly between Tube A and Tube B (at the same kVp), it is possible to reconstruct images corresponding to any dose in the range of min(A,B) to A+B. CT data of the ACR phantom were acquired on a dual-source system (SOMATOM Flash, Siemens Healthcare) with a traditional single-source (SS) technique at 6 dose levels (25, 50, 75, 100, 150, and 200 mAs). Corresponding data using the DS technique were acquired and compared with the SS data in terms of noise magnitude (pixel STD), contrast, contrast-to-noise ratio, noise power spectrum (average spatial frequency), modulation transfer function (50% frequency) and detectability index (for a non-prewhitening matched filter observer).Between the DS and SS techniques, the differences (SS-DS) in noise magnitude, contrast, and CNR were on average (across doses) 1.2 HU (4.1%), 0.7 HU (11%), and 0.02 (6.9%), respectively. The differences in NPS average frequency, MTF 50% frequency, and detectability index were 0.01 cycles/mm (2.5%), 0.03 cycles/mm (7.5%), and -.03 (-6.8%), respectively.The dose split method can be used to acquire images of the same patient equivalent to many dose levels in a single acquisition. Differences in noise, CNR, NPS, MTF, and detectability were all negligible.