Flowchart for multivariate calculation of peak detectability as a function of the tissue cancellation parameter , dose allocation , and kVp. The inserted graph illustrates the detectability index as a function of the tissue cancellation parameter and dose allocation, where , , and are determined for a single energy pair.
Illustration of (a) the double-shot DE imaging FPD employed in this work and (b) a hypothetical single-shot DE imaging system. The former involves a single CsI scintillator, whereas the latter consists of a CsI:Tl front-detector, a Cu interdetector filter, and a CsI:Tl back detector. For the single-shot case, absorption in the interdetector layer results in a low- and high-energy image acquired on the front and back detectors, respectively. The plotted spectra illustrate the difference in low- and high-energy separation between the double- and single-shot DE systems.
(a) Anatomical NPS measurements in the anthropomorphic chest phantom at plotted as a function of the tissue cancellation parameter and spatial frequency. (b) DE GNEQ at plotted as a function of the tissue cancellation parameter and spatial frequency. The superimposed black line depicts the maximum DE GNEQ at each spatial frequency. An optimal value for the tissue cancellation parameter is suggested in the range , with the optimum depending on the imaging task.
(a) Nodule contrast relative to lung as a function of the difference in kVp in DE images. The insert graph shows a 1D Gaussian approximating a nodule and background function approximated by a constant. (b) Task function for nodule detection in DE images at various energy pairs. The magnitude of the task function scales with the nodule-to-lung contrast plotted in (a).
Dual-energy images acquired across a range of low- and high-kVp. The detectability index was calculated for each image for a nodule detection task taking into account anatomical noise, contrast, and exposure at the detector. Correlation between nodule conspicuity and detectability index provides qualitative verification that detectability index provides a reasonable surrogate for image quality.
Detectability index computed at as a function of the tissue cancellation parameter and dose allocation between the low- and high-energy image. The optimal parameters, and , are identified at peak detectability, .
Optimal peak detectability index, , optimal tissue cancellation parameter , and optimal dose allocation computed as a function of low- and high-kVp and shown for three levels of DE entrance surface dose: “high-dose” , “radiographic” , and “low-dose” . The optimal kVp pairs were found at, , and, , respectively.
(a) Optimal tissue cancellation (dashed line) and optimal dose allocation (solid line) calculated as a function of total entrance surface dose. At very low doses and converge to zero, corresponding to acquiring a single projection. (b) Comparison of detectability index for radiography and DE imaging. The solid line depicts the peak detectability index computed for a DE imaging system at optimal kVp, dose allocation, and [see (a)] as a function of the total entrance surface dose . The dashed line depicts the detectability index computed for a radiographic system at . The performance of the DE imaging system for a detection task increases significantly with respect to the radiographic system at higher ESD, where anatomical noise limits conventional radiographic performance.
Single-shot vs double-shot DE imaging. The solid line depicts the computed for a double-shot DE system. The dashed line depicts the computed for a single-shot DE system (illustrated in Fig. 2). Both systems were modeled with an ESD of . The double-shot DE imaging system provides a factor of 2 increase in both and nodule contrast.
Diagram of the simple pillbox geometry used to compute contrast between nodule and lung in DE images.
Article metrics loading...
Full text loading...