Page 380 - Biomedical Engineering and Design Handbook Volume 2, Applications

358 DIAGNOSTIC EQUIPMENT DESIGN

tomosynthesis images because of the elimination of the superimposed dense breast tissue present in the
planar images. Dobbins and Godfrey have written an excellent review of tomographic imaging and
tomosynthesis, including descriptions of the several image reconstruction and deblurring strategies
(Dobbins and Godfrey, 2003). Among the many clinical applications of digital tomosynthesis are breast
imaging (Mainprize et al., 2006; Niklason et al., 1997; Sechopoulos et al., 2007; Wu et al., 2003), chest
radiography (Godfrey et al., 2006; McAdams et al., 2006; Pineda et al., 2006), dental imaging (Nair et
al., 2003; Webber et al., 1997; Webber et al., 2002; Ziegler et al., 2003), and patient positioning and
brachytherapy seed localization in radiation therapy (Godfrey et al., 2007; Reddy and Mendelson,
2005). At the present time a number of mammography equipment manufacturers are actively develop-
ing digital tomosynthesis systems for breast imaging. Using an early commercial tomosynthesis unit,
researchers at Dartmouth have recently published the results of a study designed to measure the impact
of tomosynthesis on recall rates (the fraction of screening mammography patients who are subse-
quently referred for further workup due to an abnormal mammogram). They reported a reduction of 40
percent in the recall rate among 98 women, when digital screening mammography was supplemented
with tomosynthesis (Poplack et al., 2007). Contrast-enhanced tomosynthesis, in which an iodine-based
contrast agent is injected intravenously and patients are imaged before and after administration of con-
trast, is also being investigated as an adjunct to digital mammography (Chen et al., 2007).
A major technical challenge facing breast tomosynthesis is identification of the best acquisition strat-
egy (number of breast compressions, and for each compression the number of views, angular range of
views, angular distribution of views, and distribution of x-ray dose among views). The asymmetrical
shape of the compressed breast results in preferential radiographic properties for views along directions
at small angles relative to the direction of compression, for which the thickness of breast that must be
penetrated is less. However, clustering the viewing directions near the compression direction results in
incomplete sampling of frequency space and relatively poorer spatial resolution in the direction of com-
pression compared to the other two coordinate directions. This incomplete sampling also results in image
artifacts, especially if simple backprojection via shifting and summing is used for reconstruction. The
optimum acquisition strategy is thus dependent on the reconstruction technique utilized, and possibly on
breast type (Suryanarayanan et al., 2000; Wu et al., 2003; Wu et al., 2004; Zhang et al., 2006).

Breast CT. In recent years, several academic groups have begun the development of dedicated
breast CT scanners (Boone et al., 2001; Crotty et al., 2007; Ning et al., 2003). Relative to using a
conventional whole-body CT scanner to image the breasts, the primary advantage of a dedicated
scanner is the potential reduction of radiation dose from the levels typical of whole-body CT (10 to
20 mGy) to those typical of mammography (1.0 to 3.0 mGy). In dedicated breast CT, complete angu-
lar (helical) sampling around the breast is used, thereby reducing the artifacts present in tomosyn-
thesis and producing more isotropic spatial resolution. In virtually all prototype systems, the
woman is prone with the uncompressed breast pendant though a hole in the table. A flat-panel
detector whose edge is aligned with the chest wall provides cone beam geometry, and in most cases
the entire breast is imaged in a single helical acquisition and x-ray exposure. Boone et al. have
developed a prototype scanner capable of obtaining 500 views over 360° in 17 s, while delivering
a total radiation dose to an average-sized breast equal to that of a conventional two-view planar
mammographic study (Boone et al., 2001; Boone et al., 2005). An industrial x-ray tube with a sta-
tionary tungsten target is operated at 80 kVp, a tube voltage significantly higher than that used in
mammography (25 to 35 kVp). Compared to two-dimensional (2D) FFDM, the system has modest
spatial resolution; in the coronal plane the MTF falls to 0.10 at 0.7 to 1.7 cycles/mm, depending on
the location within the field of view. This is in part because of a relatively large focal spot size (0.4
mm) and effective detector element size (0.388 mm), and the thickness of the CsI converter (600
mm) (Kwan et al., 2007). Nevertheless, images demonstrate high contrast depiction of anatomical
structures in the breast, and a second prototype system is now being tested in a pilot clinical trial.
Compared to breast MRI, breast CT may have advantages in terms of shorter acquisition
time and potentially elimination of the need for an injected contrast agent. One challenge
facing breast CT is improving the spatial resolution without unacceptable dose increase.
Higher spatial resolution is an important goal not just because of the importance of microcalcifi-
cation detection, but also because microcalcifications indicative of malignant processes are often