Page 376 - Biomedical Engineering and Design Handbook Volume 2, Applications

354 DIAGNOSTIC EQUIPMENT DESIGN

storage phosphor system, and a scanned CCD-based system that is no longer being manufactured.
Several other manufacturers are in the process of gathering data for submission to the FDA for approval.
Challenges Facing Digital Mammography. The clinical presence of FFDM is increasing at a
high rate. According to the American College of Radiology, as of May 2007, 20 percent of accred-
ited facilities have at least one digital mammography system versus only 16 percent as of January
2007. The current rapid growth is attributed both to the overall trend in radiology toward digital
imaging, and to the results of the Digital Mammographic Imaging Screening Trial (DMIST).
DMIST enrolled 49,500 asymptomatic women in the United States and Canada, each of whom had
both a digital and screen-film mammographic study at the time of enrollment and again a year later.
The study concluded that, compared to film-based mammography, FFDM was significantly better
for screening women who fell into any of the following categories: (1) under 50 years of age,
(2) of any age with heterogeneously or extremely dense breasts, or (3) pre- or perimenopausal
women of any age. Although adoption of FFDM into clinical practice is well under way, there are
still major challenges that must be addressed. Perhaps the most immediate obstacles have to do with
image display. The very large size of the digital images (up to 30 megapixels, with 12 to 16 bits per
pixel) far exceeds the matrix size (up to 2000 × 2500 or 5 megapixels) and grayscale depth (typi-
cally 8 bits) of current commercially available high-resolution displays. Furthermore, a typical
screening study involves simultaneous viewing of four current views (two per breast) alongside the
corresponding four views from the previous screening study. Partially for this reason, many early
practitioners of FFDM used laser film printers to produce hard copies of the digitally obtained
mammograms, so that they could be viewed on a conventional light box, alongside the previous
(analog) study.
Another area of ongoing research is the development of image-processing techniques for both laser
film and grayscale monitor viewing. In the former case (hard copy display), the characteristics of the
laser film automatically apply a nonlinear dynamic range compression to the digital pixel values. In the
case of workstation viewing (soft copy display), it is desirable to limit the amount of time that the radi-
ologist must spend adjusting the image display (i.e., adjusting the range of pixel values mapped into
the available grayscale levels of the display), so some means of reducing the dynamic range in the
mammogram without compromising image quality is needed. One approach is thickness compensation
(Bick et al., 1996; Byng et al., 1997). This is a software technique that compensates for the decrease in
attenuating thickness near the breast periphery by locating region between the area of uniform thick-
ness (where the breast is in contact with the flat compression paddle) and the skin line. Pixel values
corresponding to that region are scaled downward to make their values more similar to those in the
region of uniform thickness.

12.3.2 Diagnostic Ultrasound
Ultrasound (US) is used as a diagnostic (as opposed to screening) breast imaging modality, in
part because it lacks the sensitivity of x-ray mammography for microcalcifications, and its speci-
ficity for small masses varies considerably among operators. However, US is recommended by
the American College of Radiology as the initial imaging technique for evaluation of masses in
women under 30 and in lactating and pregnant women—populations for whom mammography is
less sensitive and for whom radiation dose should be minimized. In the majority of breast imag-
ing cases, the primary diagnostic use of ultrasound is the differentiation of solid masses from
fluid-filled cysts (Fig. 12.3). Regarding solid mass, there has been a general lack of confidence,
in the ability of US to characterize solid masses as benign versus malignant without obtaining
histopathology for many cases. Although at least one study has reported that benign and malig-
nant solid breast masses could be differentiated based on US alone (Stavros et al., 1995), sub-
sequent studies have not confirmed this hypothesis, and it is now generally believed that at the
present time there are no ultrasound features that, by themselves, are sufficient evidence to
forgo biopsy. Recently, the American College of Radiology Imaging Network published its trial
evaluating breast ultrasound as a breast cancer screening tool. The study enrolled over 2,800
patients with radiodense breast tissue, of which 2,637 were fully evaluable by either gold stan-
dard pathology or follow up imaging. In this study, the investigators found the diagnostic yield