Quality control for digital mammography in the ACRIN DMIST trial: part I.

The Digital Mammography Imaging Screening Trial, conducted by the American College of Radiology Imaging Network, is a clinical trial designed to compare the accuracy of full-field digital mammography (FFDM) versus screen-film mammography in a screening population. Five FFDM systems from four manufacturers (Fischer, Fuji, General Electric, and Lorad) were employed in the study at 35 clinical sites. A core physics team devised and implemented tests to evaluate these systems. A detailed description of physics and quality control tests is presented, including estimates of: mean glandular dose, modulation transfer function (MTF), 2D noise power spectra, and signal-to-noise ratio (SNR). The mean glandular doses for the standard breast ranged from 0.79 to 2.98 mGy, with 1.62 mGy being the average across all units and machine types. For the five systems evaluated, the MTF dropped to 50% at markedly different percentages (22% to 87%) of the Nyquist limit, indicating that factors other than detector element (del) size have an important effect on spatial resolution. Noise power spectra and SNR were measured; however, we found that it was difficult to standardize and compare these between units. For each machine type, the performance as measured by the tests was very consistent, and no predictive benefit was seen for many of the tests during the 2-year period of the trial. It was found that, after verification of proper operation during acceptance testing, if systems failed they generally did so suddenly rather than through gradual deterioration of performance. Because of the relatively short duration of this study further, investigation of the long-term failure characteristics of these systems is advisable.

Scattered energy deposition under shielding.

Monte Carlo methods are used to generate line spread functions describing dose distributions at a variety of depths within a homogeneous water phantom. The line spread function data are convolved with a step function that represents the edge of a primary radiation field. The dosimetric information beyond the edge of the field is reported in the form of tissue-air ratios for three different beam spectra in the diagnostic energy range.

Internal densitometric gating for digital subtraction angiography.

Motion artifacts create a severe problem in digital subtraction angiography (DSA) studies. Periodic motion can be eliminated by "gating," matching a precontrast mask with a postcontrast image at the same phase position in the cycle. Electrocardiogram (EKG) signals are used in cardiac DSA for this purpose. An alternate method relies on the generation of a density-time curve dependent on the attenuation changes of anatomical motion. A densitometric window placed at an appropriate location records the variation, from which individual images are tagged with phase position information encoded as delay time. Results are similar to the EKG gating method for cardiac DSA when using an appropriate window location. Periodic motion caused by superimposed respiration can be suppressed by tracking diaphragm or other object attenuation changes with the same algorithms. Using these techniques permits the nonarbitrary matching of mask and contrast images without physiologic monitoring. The techniques are straightforward and relatively easy to implement on systems designed for rapid digital imaging.

A fluoroscopy-based computed tomography scanner for small specimen research.

RATIONALE AND OBJECTIVES: A small-laboratory computed tomography (CT) system using a fluoroscopic system and a personal computer was fabricated and tested. The motivation for building this specimen scanner was to provide medical researchers with the capability of using CT as a practical tool in their research, as well as to provide an opportunity for hands-on CT instruction. METHODS AND MATERIALS: The CT system was constructed using mostly off-the-shelf items; however, the CT stage itself was custom fabricated and software development was necessary. In addition, a personal computer and a standard fluoroscopy system were used. RESULTS: The spatial resolution was found to match the 228-microns sampling limitation, yielding approximately 2 line pairs per mm. Iodine contrast sensitivity studies showed that 1% solution of 370 mg/ml iodine solution was easily detected (P = .05). CONCLUSIONS: A small CT scanner for specimen research can be economically constructed, and is capable of good performance. The authors found substantial interest on the part of small animal researchers involved in a wide variety of medical research.

Accuracy and precision of in vitro volumetric measurements by three-dimensional sonography.

RATIONALE AND OBJECTIVES: A radiologist often wishes to measure organ volume or monitor changes in internal lesion volume during treatment. If this can be determined via three-dimensional ultrasound, the relative simplicity of the procedure and the decreased cost and known risks to the patient would make this method an attractive alternative to other modalities. METHODS: Three-dimensional ultrasound scans were made of six phantoms: four nonechogenic spheres, one echogenic sphere, and one echogenic, irregularly shaped phantom. A total of 22 volume scans were produced. Volume estimations were made using data from cross-sectional areas and from linear measurements. In all, 193 volume estimations were made. These results were compared with known volumes and with volume estimates from computed tomography scans. RESULTS: Three-dimensional ultrasound detected size differences of 10% with 95% certainty. CONCLUSIONS: The accuracy and precision of volume estimates via three-dimensional ultrasound is at least as good as those obtained via conventional ultrasound.

Partial splenic embolization guided by blood flow measurements.

The author used blood flow measurements to guide partial splenic embolization procedures in five patients. The measurements were obtained by time density analysis of contrast medium injections in the splenic artery using digital subtraction angiographic techniques and modified software. The blood flow measurements demonstrated blood flow changes occurring as a result of the embolization. The embolization procedures continued until the blood flow in the splenic artery had been reduced to 50% of its original value. The blood flow measurements appear to predict the amount of parenchymal reduction achieved by embolization and show promise as a method of monitoring these procedures.

The beneficial effects of short pulse width acquisition and ECG-gating in digital angiocardiography.

Experimental intravenous left ventriculography and coronary angiography after aortic root injection were performed in 13 animals. Various image acquisition and digital image processing techniques were employed. These included single mask mode subtraction, integrated mask mode subtraction, ECG-gated image acquisition, TV camera read-out in interlaced mode, and pulse progressive read-out with 1- to 5-ms pulse width. High quality images were obtained consistently only after ECG-gating and use of 1- to 5-ms pulse width with pulse progressive read-out of the TV camera. Integration of four to eight background as well as contrast images all from the same phase of the cardiac cycle further improved image quality.

Imaging findings of the developing temporal bone in fetal specimens.

PURPOSE: To trace the development of the normal fetal temporal bone by means of plain radiography, MR, and CT. METHODS: Eighteen formalin-fixed fetal specimens, 13.5 to 24.4 weeks' gestational age, were examined with a mammographic plain film technique, CT, and MR imaging at 1.5 T. Temporal bone development and ossification were assessed. RESULTS: The membranous labyrinth grows with amazing rapidity and attains adult size by the middle of the gestation period. The cochlea, vestibule, and semicircular canals are very prominent and easily recognized on MR images. The otic capsule develops from a cartilage model. Ossification of the otic capsule proceeds rapidly between 18 and 24 weeks from multiple ossification centers that replace the cartilaginous framework. The mastoid, internal auditory canal, vestibular aqueduct, and external auditory canal continue to grow after birth. CONCLUSION: The study of fetal developmental anatomy may lead to a better understanding of congenital disorders of the ear. Faster MR scanning techniques may provide a method for in utero evaluation of the fetal temporal bone.

MR, CT, and plain film imaging of the developing skull base in fetal specimens.

BACKGROUND AND PURPOSE: The developing fetal skull base has previously been studied via dissection and low-resolution CT. Most of the central skull base develops from endochondral ossification through an intermediary chondrocranium. We traced the development of the normal fetal skull base by using plain radiography, MR imaging, and CT. METHODS: Twenty-nine formalin-fixed fetal specimens ranging from 9 to 24 weeks' gestational age were examined with mammographic plain radiography, CT, and MR imaging. Skull base development and ossification were assessed. RESULTS: The postsphenoid cartilages enclose the pituitary and fuse to form the basisphenoid, from which the sella turcica and the posterior body of the sphenoid bone originate. The presphenoid cartilages will form the anterior body of the sphenoid bone. Portions of the presphenoid cartilage give rise to the mesethmoid cartilage, which forms the central portion of the anterior skull base. Ossification begins in the occipital bone (12 weeks) and progresses anteriorly. The postsphenoid (14 weeks) and then the presphenoid portion (17 weeks) of the sphenoid bone ossify. Ossification is seen laterally (16 weeks) in the orbitosphenoid, which contributes to the lesser wing of the sphenoid, and the alisphenoid (15 weeks), which forms the greater wing. CONCLUSION: MR imaging can show early progressive ossification of the cartilaginous skull base and its relation to intracranial structures. The study of fetal developmental anatomy may lead to a better understanding of abnormalities of the skull base.

An analytical edge spread function model for computer fitting and subsequent calculation of the LSF and MTF.

The previous work of Yin, Giger, and Doi [Med. Phys. 17, 962-966 (1990)] demonstrated that using a computerized fit of an analytic line spread function to experimentally measured data is very useful for determining the presampling modulation transfer function of an imaging system. In this report, the work of Yin et al. is extended to include an analytic expression for the edge spread function (ESF). By fitting experimentally determined edge spread function data to the analytical expression, the normalized line spread function (LSF) and the normalized modulation transfer function (MTF) can be easily calculated from four ESF fit coefficients. The extension from the line spread function to the edge spread function should be valuable in cases where slit measurements are impractical, for example, in very high resolution imaging systems where the required slit dimensions become impractically small, or in measurements of the transfer properties of scattered radiation or off-focus radiation, where large area exposures are necessary.

A comparison of mono- and poly-energetic x-ray beam performance for radiographic and fluoroscopic imaging.

The purpose of this investigation was to evaluate the potential of monoenergetic x-ray sources to improve image quality and reduce patient dose compared to conventional tungsten anode x-ray spectra. This was purely a computer simulation study. X-ray spectra were generated using the Birch and Marshal spectral model, patient x-ray transmission was calculated using Monte Carlo techniques, and a numerical method was developed for determining antiscatter grid performance. A 120 mg/cm2 Gd2O2S intensifying screen was simulated for radiography and a 144 mg/cm2 CsI image intensifier was simulated for fluoroscopy. The source of subject contrast that was simulated included tissue, calcium, and iodine targets which varied in mass thicknesses from 10 to 1000 mg/cm2. The figure of merit of the [contrast to noise ratio] 2/[integral dose] was used as a relative measure of dose utilization. Depending on the object thickness, monoenergetic x-ray sources with a screen-film detector exhibited a 1.4 to 2.4 improvement over tungsten anode spectra for iodine contrast, a 1.5 to 2.0 improvement for calcium imaging, and about a 1.4 to 1.6 improvement for tissue contrast. The thicker patients (30 cm) benefitted more than thinner (10 cm) ones. For the image intensifier as a detector, a 1.4-2.3 improvement factor was found for monoenergetic sources and an iodine signal object. For the practical range of radiographic imaging scenarios using present-day detector technologies, monoenergetic sources may provide an improvement in dose utilization that is comparable to the improvements that can be expected with scanning slit devices over conventional antiscatter grids.

An accurate method for computer-generating tungsten anode x-ray spectra from 30 to 140 kV.

A tungsten anode spectral model using interpolating polynomials (TASMIP) was used to compute x-ray spectra at 1 keV intervals over the range from 30 kV to 140 kV. The TASMIP is not semi-empirical and uses no physical assumptions regarding x-ray production, but rather interpolates measured constant potential x-ray spectra published by Fewell et al. [Handbook of Computed Tomography X-ray Spectra (U.S. Government Printing Office, Washington, D.C., 1981)]. X-ray output measurements (mR/mAs measured at 1 m) were made on a calibrated constant potential generator in our laboratory from 50 kV to 124 kV, and with 0-5 mm added aluminum filtration. The Fewell spectra were slightly modified (numerically hardened) and normalized based on the attenuation and output characteristics of a constant potential generator and metal-insert x-ray tube in our laboratory. Then, using the modified Fewell spectra of different kVs, the photon fluence phi at each 1 keV energy bin (E) over energies from 10 keV to 140 keV was characterized using polynomial functions of the form phi (E) = a0[E] + a1[E] kV + a2[E] kV2 + ... + a(n)[E] kVn. A total of 131 polynomial functions were used to calculate accurate x-ray spectra, each function requiring between two and four terms. The resulting TASMIP algorithm produced x-ray spectra that match both the quality and quantity characteristics of the x-ray system in our laboratory. For photon fluences above 10% of the peak fluence in the spectrum, the average percent difference (and standard deviation) between the modified Fewell spectra and the TASMIP photon fluence was -1.43% (3.8%) for the 50 kV spectrum, -0.89% (1.37%) for the 70 kV spectrum, and for the 80, 90, 100, 110, 120, 130 and 140 kV spectra, the mean differences between spectra were all less than 0.20% and the standard deviations were less than approximately 1.1%. The model was also extended to include the effects of generator-induced kV ripple. Finally, the x-ray photon fluence in the units of photons/mm2 per mR was calculated as a function of HVL, kV, and ripple factor, for various (water-equivalent) patient thicknesses (0, 10, 20, and 30 cm). These values may be useful for computing the detective quantum efficiency, DQE(f), of x-ray detector systems. The TASMIP algorithm and ancillary data are made available on line at http:/(/)www.aip.org/epaps/epaps.html.

X-ray scatter removal by deconvolution.

The distribution of scattered x rays detected in a two-dimensional projection radiograph at diagnostic x-ray energies is measured as a function of field size and object thickness at a fixed x-ray potential and air gap. An image intensifier-TV based imaging system is used for image acquisition, manipulation, and analysis. A scatter point spread function (PSF) with an assumed linear, spatially invariant response is modeled as a modified Gaussian distribution, and is characterized by two parameters describing the width of the distribution and the fraction of scattered events detected. The PSF parameters are determined from analysis of images obtained with radio-opaque lead disks centrally placed on the source side of a homogeneous phantom. Analytical methods are used to convert the PSF into the frequency domain. Numerical inversion provides an inverse filter that operates on frequency transformed, scatter degraded images. Resultant inverse transformed images demonstrate the nonarbitrary removal of scatter, increased radiographic contrast, and improved quantitative accuracy. The use of the deconvolution method appears to be clinically applicable to a variety of digital projection images.

Monte Carlo simulation of the scattered radiation distribution in diagnostic radiology.

Monte Carlo techniques were employed to evaluate the point spread function (PSF) of scattered radiation in diagnostic radiology. The Monte Carlo procedure is described and shown to compare well with Monte Carlo scatter analysis of other authors. The intensity and distribution of the PSF are described independently. The effects of object thickness, air gap, and beam spectra are examined. An analytic derivation of the scatter PSF is presented in a companion article, and the Monte Carlo results discussed herein are used for comparison.

An analytical model of the scattered radiation distribution in diagnostic radiology.

A simple scatter model is used to analytically derive the point spread function (PSF) for scattered radiation in diagnostic radiology. The resulting equation is a function of four physical parameters; object thickness, object-to-detector distance (air gap), and the linear attenuation coefficients for both primary and scatter radiation. Though the model is based upon single scattering, it is shown that by reducing the scatter attenuation coefficient the analytic model compares well to the multiple scattering PSF determined using Monte Carlo analysis.

Sinusoidal modulation analysis for optical system MTF measurements.

The modulation transfer function (MTF) is a commonly used metric for defining the spatial resolution characteristics of imaging systems. While the MTF is defined in terms of how an imaging system demodulates the amplitude of a sinusoidal input, this approach has not been in general use to measure MTFs in the medical imaging community because producing sinusoidal x-ray patterns is technically difficult. However, for optical systems such as charge coupled devices (CCD), which are rapidly becoming a part of many medical digital imaging systems, the direct measurement of modulation at discrete spatial frequencies using a sinusoidal test pattern is practical. A commercially available optical test pattern containing spatial frequencies ranging from 0.375 cycles/mm to 80 cycles/mm was sued to determine the MRF of a CCD-based optical system. These results were compared with the angulated slit method of Fujita [H. Fujita, D. Tsia, T. Itoh, K. Doi, J. Morishita, K. Ueda, and A. Ohtsuka, "A simple method for determining the modulation transfer function in digital radiography," IEEE Trans. Medical Imaging 11, 34-39 (1992)]. The use of a semiautomated profiled iterated reconstruction technique (PIRT) is introduced, where the shift factor between successive pixel rows (due to angulation) is optimized iteratively by least-squares error analysis rather than by hand measurement of the slit angle. PIRT was used to find the slit angle for the Fujita technique and to find the sine-pattern angle for the sine-pattern technique. Computer simulation of PIRT for the case of the slit image (a line spread function) demonstrated that it produced a more accurate angle determination than "hand" measurement, and there is a significant difference between the errors in the two techniques (Wilcoxon Signed Rank Test, p < 0.001). The sine-pattern method and the Fujita slit method produced comparable MTF curves for the CCD camera evaluated.

Removal of image intensifier veiling glare by mathematical deconvolution techniques.

X-ray images acquired with an image intensifier detector system suffer from veiling glare, a low-frequency degradation described by a point spread function (PSF). The PSF has two experimentally determined parameters unique to a given image intensifier. This information is utilized to deconvolve the degradation from digitally acquired images. Results demonstrate a significant increase in contrast ratio of high-contrast objects after deconvolution and image restoration.

Characterization of the veiling glare PSF in x-ray image intensified fluoroscopy.

A theoretical derivation for the point spread function (PSF) which describes the veiling glare in x-ray image intensifiers (II) is presented. The PSF is dependent on two parameters which can be determined experimentally for a given II. An experimental investigation of the linearity of veiling glare phenomenon is undertaken. The experimental results indicate that veiling glare could be described as a linear process to a high degree of accuracy.

A Monte Carlo study of x-ray fluorescence in x-ray detectors.

Advances in digital x-ray detector systems have led to a renewed interest in the performance of x-ray phosphors and other detector materials. Indirect flat panel x-ray detector and charged coupled device (CCD) systems require a more technologically challenging geometry, whereby the x-ray beam is incident on the front side of the scintillator, and the light produced must diffuse to the back surface of the screen to reach the photoreceptor. Direct detector systems based on selenium have also enjoyed a growing interest, both commercially and academically. Monte Carlo simulation techniques were used to study the x-ray scattering (Rayleigh and Compton) and the more prevalent x-ray fluorescence properties of seven different x-ray detector materials, Gd2O2S, CsI, Se, BaFBr, YTaO4, CaWO4, and ThO2. The redistribution of x-ray energy, back towards the x-ray source, in a forward direction through the detector, and lateral reabsorption in the detector was computed under monoenergetic conditions (1 keV to 130 keV by 1 keV intervals) with five detector thicknesses, 30, 60, 90, 120, and 150 mg/cm2 (Se was studied from 30 to 1000 mg/cm2). The radial distribution (related to the point spread function) of reabsorbed x-ray energy was also determined. Representative results are as follows: At 55 keV, more (31.3%) of the incident x-ray energy escaped from a 90 mg/cm2Gd2O2S detector than was absorbed (27.9%). Approximately 1% of the total absorbed energy was reabsorbed greater than 0.5 mm from the primary interaction, for 90 mg/cm2 CsI exposed at 100 kVp. The ratio of reabsorbed secondary (fluorescence + scatter) radiation to the primary radiation absorbed in the detectors (90 mg/cm2) (S/P) was determined as 10%, 16%, 2%, 12%, 3%, 3%, and 0.3% for a 100 kVp tungsten anode x-ray spectrum, for the Gd2O2S, CsI, Se, BaFBr, YTaO4, CaWO4, and ThO2 detectors, respectively. The results indicate significant x-ray fluorescent escape and reabsorption in common x-ray detectors. These findings suggest that x-ray fluorescent radiation redistribution should be considered in the design of digital x-ray imaging systems.

Characterization of the point spread function and modulation transfer function of scattered radiation using a digital imaging system.

A digital radiographic system was used to measure the distribution of scattered x radiation from uniform slabs of Lucite at various thicknesses. Using collimation and air gap techniques, [primary + scatter] images and primary images were digitally acquired, and subtracted to obtain scatter images. The scatter distributions measured using small circular apertures were computer fit to an analytical function, representing the circular aperture function convolved with a modified Gaussian point spread function (PSF). On the basis of goodness of fit criterion, the proposed Gaussian function is a very good model for the scatter PSF. The measured scatter PSF's are reported for various Lucite thicknesses. Using the PSF's, the modulation transfer functions are calculated, and this spatial frequency information may have value in analytical scatter removal techniques, grid design, and air gap optimization.