Key words: mammography, phantom, dosimetry, thermoluminescent dosimetry, tissue-equivalence, research
OST staff continued to investigate the clinical potential of the optimized mammography system that has been developed by CDRH in collaboration with the University of Southern California and the Clinical Center at NIH. During FY 97, the optimized system was used in a study to evaluate its performance using both screen-film and stimulable-phosphor image receptors when operated at dose levels typical of current clinical practice. The project was carried out in collaboration with colleagues at Georgetown University. The results of the study showed that the optimized system can provide significantly improved imaging performance, compared to current systems, at comparable dose levels. The study was reported at the SPIE meeting Medical Imaging 1997 and published in the proceedings of that meeting.
Additional progress was made in developing a protocol for the clinical evaluation of the optimized system in its original, low-dose configuration. The planned protocol involves scoring the visibility of normal anatomic structures and assessing the volume of tissue imaged as a measure of the patient positioning capability of the system. A preliminary, retrospective trial has been planned to determine the reader variability in performing these tasks. Assistance in the planning of the clinical evaluation is being provided by a statistician from the Office of Surveillance and Biometrics.
Three research projects were started in FY 97 to support the MQSA program. The first project, which has been completed, was an examination of the feasibility of visually assessing image noise independent of other image attributes such as sharpness and radiographic contrast. A two-alternative, forced choice (2AFC) experimental paradigm was used to perform pair-wise comparisons of noise samples from five different mammographic film-screen combinations. The results showed that observers can reliably distinguish differences in noise power of less than a factor of two. Observers can also reliably distinguish between noise samples with similar low-frequency content but different intermediate and high-frequency content. The study thus supports the feasibility of assessing noise by providing a selection of noise samples of known characteristics, i.e., a noise step tablet, that observers can use to grade the noise content of images from a system under evaluation.
The purpose of the second project is to examine the phantom image scoring process used in the ACR Mammography Accreditation Program (MAP) and in the MQSA facility inspection program. Several aspects of the process are being considered. First, it is important to characterize the imaging tasks used in commercially available imaging phantoms and determine their sensitivity to changes in mammography system conditions. OST will also evaluate the ability of the phantoms to mimic a range of breast tissue characteristics.
Additionally, OST scientists want to evaluate new approaches to phantom image scoring for both new, digital, and traditional, film-screen mammography systems. One such approach is machine reading of phantom images. OST plans to propose changes in imaging phantom design, construction, and image-scoring protocols to increase the image phantom's sensitivity to changes in mammography system conditions and its ability to mimic a range of breast tissue characteristics, if the results support doing so.
The third project is intended to provide experimental verification of the exposure-to-dose conversion factors used in the ACR MAP and MQSA inspection procedures. The plan is to use both conventional thermoluminescent dosimeters (TLDs) and a new technology, laser-heated thermoluminescent dosimetry (LHTLD), to measure depth dose curves in a variety of tissue-equivalent phantoms. The phantoms are designed to match the mathematical phantoms used in the Monte Carlo calculations of Wu, Barnes, and Tucker that are the basis for the conversion factors to be verified. The project will also serve to evaluate the utility of LHTLD technology, which is being supplied by Keithley Instruments under a CRADA, for quality assurance applications in mammography.
Both the second and third projects have required the acquisition of equipment and materials. This has taken up a significant portion of the FY 97 mammography effort. It has also resulted in significant enhancement of the capabilities of the CDRH mammography research facility. The list of items acquired includes the following: a prototype of the NIST-designed crystal spectrometer that permits precise, noninvasive kVp determinations; a laser film digitizer (Lumisys 85, 50µ (pixel size) and scanning software (the software allows viewing of digitized images on an existing high-resolution display system (1728 X 2304, two monitors) as well as import and export of images in the DICOM format developed jointly by the ACR and NEMA); an Agfa ARCUS II scanner, a high-resolution (1200 dpi) document scanner useful for evaluating images of line-pair phantoms; software for machine scoring of digitized phantom images from Radiological Imaging Technology; phantom image evaluation software from Dr. Dev Chakraborty at the University of Pennsylvania; and breast-tissue-equivalent phantoms in a range of compositions and thicknesses.