Dose reduction efforts in diagnostic CT have brought the tradeoff of dose versus image quality to the forefront. The need for meaningful characterization of image noise beyond that offered by pixel standard deviation is becoming increasingly important. This work aims to study the implementation of the noise power spectrum (NPS) and noise equivalent quanta (NEQ) on modern, multislice diagnostic CT scanners. The details of NPS and NEQ measurement are outlined and special attention is paid to issues unique to multislice CT. Aliasing, filter design and effects of acquisition geometry are investigated. While it was found that both metrics can be implemented in modern CT, it was discovered that NEQ cannot be aptly applied with certain non-traditional reconstruction filters or in helical mode. NPS and NEQ under a variety of conditions are examined. Extensions of NPS and NEQ to uses in protocol standardization are also discussed.

Monte Carlo procedures using the SIERRA code (validated in a companion article) were used to investigate the scatter properties in mammography. The scatter to primary ratio (SPR) was used for quantifying scatter levels as a function of beam spectrum, position in the field, air gap, breast thickness, tissue composition, and the area of the field of view (FOV). The geometry of slot scan mammography was also simulated, and SPR values were calculated as a function of slot width. The influence of large air gaps (to 30 cm) was also studied in the context of magnification mammography. X-ray energy and tissue composition from 100% adipose to 100% glandular demonstrated little effect on the SPR. Air gaps over a range from 0 to 30 mm showed only slight effects. The SPR increased with increased breast thickness and with larger fields of view. Measurements from 82 mammograms provided estimates of the range of compressed breast thickness (median: 5.2 cm, 95% range: 2.4 cm to 7.9 cm) and projected breast area onto the film (left craniocaudal view, median: 146 cm2, 95% range: 58 cm2 to 298 cm2). SPR values for semicircular breast shapes, Mo/Mo spectra, and a 15 mm air gap were parametrized as a function of breast thickness and (semicircular) breast diameter. With the coefficients a = - 2.35452817439093, b = 22.3960980055927, and c = 8.85064260299289, the equation SPR= [a + b x (diameter in cm)--(-1.5) + c x (thickness in cm) --(-0.5)]-- -1 produces SPR data from 2 to 8 cm and from 3 to 30 cm breast diameters with an average error of about 1%.

The purpose of this investigation was to compare and validate the performance of the SIERRA Monte Carlo simulation routines for the analysis of the scatter to primary ratio (SPR) in the mammography setting. Two Monte Carlo simulation methods were addressed, the direct method was a straightforward and geometrically accurate simulation procedure, and the convolution method uses idealized geometry (monoenergetic, normally incident delta function input to the scattering medium) to produce scatter point spread functions (PSFs). The PSFs were weighted by the x-ray spectrum of interest and convolved with the field of view to estimate SPR values. The SPR results of both Monte Carlo procedures were extensively compared to five published sources, including Monte-Carlo-derived and physically measured SPR assessments. The direct method demonstrated an overall agreement with the literature of 3.7% accuracy (N=5), and the convolution method demonstrated an average of 7.1% accuracy (N=14). The comparisons were made over a range of parameters which included field of view, phantom thickness, x-ray energy, and phantom composition. Limitations of the beam stop method were also discussed. The results suggest that the SIERRA Monte Carlo routines produce accurate SPR calculations and may be useful for a more comprehensive study of scatter in mammography.

Monte Carlo analysis in the radiological sciences has been used for several decades, however with the ever-increasing power of desktop computers, the utility of Monte Carlo simulation is increasing. A Monte Carlo code called the Simple Investigative Environment for Radiological Research Applications (SIERRA) is described mathematically, and is then compared against an array of published and unpublished results determined by other means. A series of 32 comparisons between data sets, 22 from independent Monte Carlo simulations and 10 from physically measured data, were assessed. The compared parameters included depth dose curves, lateral energy scattering profiles, scatter to primary ratios, normalized glandular doses, angular scattering distributions, and computed tomography dose index (CTDI) values. Three of the 32 comparison data sets were excluded as they were identified as outliers. Of the remaining 29 data sets compared, the mean differences ranged from -14.8% to +17.2%, and the average of the mean differences was 0.12% (sigma = 1.64%), and the median difference was 1.57%. Fifty percent of the comparisons showed mean differences of approximately 5% or less, and 93% of the comparisons showed mean differences of 12% or less. We conclude that for research applications in diagnostic radiology, the SIERRA Monte Carlo code demonstrates accuracy and precision to well within acceptable levels.

An experimental measurement technique that directly measures the magnitude and spatial distribution of scatter in relation to primary radiation is presented in this work. The technique involves the acquisition of magnified edge spread function (ESF) images with and without scattering material present. The ESFs are normalized and subtracted to yield scatter-to-primary ratios (SPRs), along with the spatial distributions of scatter and primary radiation. Mammography is used as the modality to demonstrate the ESF method, which is applicable to all radiographic environments. Sets of three images were acquired with a modified clinical mammography system employing a flat panel detector for 2, 4, 6, and 8 cm thick breast tissue equivalent material phantoms composed of 0%, 43%, and 100% glandular tissue at four different kV settings. Beam stop measurements of scatter were used to validate the ESF methodology. There was good agreement of the mean SPRs between the beam stop and ESF methods. There was good precision in the ESF-determined SPRs with a coefficient of variation on the order of 5%. SPRs ranged from 0.2 to 2.0 and were effectively independent of energy for clinically realistic kVps. The measured SPRs for 2, 4, and 6 cm 0% glandular phantoms imaged at 28 kV were 0.21+/-0.01, 0.39+/-0.01, and 0.57+/-0.02, respectively. The measured SPRs for 2, 4, and 6 cm 43% glandular phantoms imaged at 28 kV were 0.20+/-0.01, 0.35+/-0.02, and 0.53+/-0.02, respectively. The measured SPRs for 2, 4, and 6 cm 100% glandular phantoms imaged at 28 kV were 0.22+/-0.02, 0.42+/-0.03, and 0.88+/-0.08, respectively.

No abstract

The assessment of the radiation dose to internal organs or to an embryo or fetus is required on occasion for risk assessment or for comparing imaging studies. Limited resources hinder the ability to accurately assess the radiation dose received to locations outside the tissue volume actually scanned during computed tomography (CT). The purpose of this study was to assess peripheral doses and provide tabular data for dose evaluation. Validated Monte Carlo simulation techniques were used to compute the dose distribution along the length of water-equivalent cylindrical phantoms, 16 and 32 cm in diameter. For further validation, comparisons between physically measured and Monte Carlo-derived air kerma profiles were performed and showed excellent (1% to 2%) agreement. Polyenergetic x-ray spectra at 80, 100, 120, and 140 kVp with beam shaping filters were studied. Using 10(8) simulated photons input to the cylinders perpendicular to their long axis, line spread functions (LSF) of the dose distribution were determined at three depths in the cylinders (center, mid-depth, and surface). The LSF data were then used with appropriate mathematics to compute dose distributions along the long axis of the cylinder. The dose distributions resulting from helical (pitch = 1.0) scans and axial scans were approximately equivalent. Beyond about 3 cm from the edge of the CT scanned tissue volume, the fall-off of radiation dose was exponential. A series of tables normalized at 100 milliampere seconds (mAs) were produced which allow the straight-forward assessment of dose within and peripheral to the CT scanned volume. The tables should be useful for medical physicists and radiologists in the estimation of dose to sites beyond the edge of the CT scanned volume.

No abstract