Page 385 - Biomedical Engineering and Design Handbook Volume 2, Applications
P. 385
BREAST IMAGING SYSTEMS: DESIGN CHALLENGES FOR ENGINEERS 363
diagnostic information complimentary to that of x-ray mammography (Allen et al., 2000; Buscombe
et al., 2001; Palmedo et al., 1998; Scopinaro et al., 1997).
Studies have been made comparing the performance of scintimammography to that of contrast-
enhanced MRI as adjuncts to x-ray mammography. In a study of 49 patients comparing contrast MRI
with conventional, large-camera scintimammography, Imbriaco et al. found no statistically significant
difference in either sensitivity or specificity (Imbriaco et al., 2001). However, comparing contrast MRI
with scintimammography using a dedicated breast gamma camera (see the section “Dedicated
Cameras”) imaging 33 indeterminate lesions, Brem et al. found comparable sensitivity, but signifi-
cantly greater specificity (71 percent) for scintimammography relative to MRI (25 percent) (Brem
et al., 2007). Like MRI, scintimammography is particularly useful for women with radiodense breasts,
for whom mammographic interpretation can be difficult.
Technical challenges associated with scintimammography are (1) positioning the camera close to
the breast, (2) dealing with the significant scatter radiation arising from gamma rays emitted from
regions of the heart and liver, and (3) correcting for contrast degradation due to partial volume aver-
aging and attenuation of gamma rays emitted from the lesion. The first of these issues is driven by
the fact that the spatial resolution of cameras with parallel hole collimators is approximately a linear
function of the distance between source and collimator. The difficulty of positioning general-purpose
gamma cameras close to the breast has led to the fact that while scintimammography using large
gamma cameras has excellent sensitivity for tumors larger than about 1 cm, sensitivity is generally
poor for smaller, nonpalpable, or medially located lesions. This has been a primary incentive for the
development of dedicated gamma cameras for breast imaging.
Dedicated Cameras. One reason for the low sensitivity of conventional scintimammography for
small lesions is that it is difficult to position the conventional Anger cameras close to the breast. This is
due to their large size, and their appreciable inactive borders. The result is that the lesion-to-collimator
distance can often exceed 20 cm. At this distance, the spatial resolution of conventional gamma cameras
with high-resolution parallel hole collimators can be 15 to 20 mm. Thus counts originating from the
lesion are smeared out over a large area of the image, and small lesions, providing few counts, are lost.
One potential method for improving sensitivity is the development of smaller gamma cameras with nar-
row inactive borders, permitting camera placement adjacent to the chest wall, and near the breast.
Several groups have developed such dedicated breast gamma cameras (Cinti et al., 2003; Hruska and
O’Connor, 2006; Majewski et al., 2001; More et al., 2006; Pani et al., 1998; Pani et al., 2004; Stebel et al.,
2005). Early clinical evaluation of these dedicated systems have shown promising results (Brem et al.,
2007; Brem et al., 2005; Scopinaro et al., 1999).
Positron Emission Mammography. In positron emission mammography (PEM), positron-emitting
radiotracers are utilized, rather than single gamma ray emitters. Radiotracer location within the
breast is determined by detection of the pair of simultaneously emitted 511-keV gamma rays, resulting
when the positron annihilates with an electron in the breast. Timing coincidence circuitry is used
to identify gamma rays originating from a single annihilation event. To date, breast PET has been
based primarily on assessment of either glucose metabolic rate via 2-[ F]fluoro-2-deoxyglucose
18
(FDG) or of estrogen or progestin receptor density using 16a-[ F]fluoroestradiol (FES). Primarily
18
because of its limited sensitivity for small lesions, whole-body PET has limited application in tumor
detection (Mankoff and Eubank, 2006). The primary clinical breast cancer applications for whole-body
FDG-PET are loco-regional staging (determination of the extent of cancer spreading), and monitoring
of loco-regional recurrence following therapy (Avril and Adler, 2007; Eubank, 2007). A European
multicenter trial evaluating whole-body FDG-PET for staging recurrent breast cancer demonstrated an
overall sensitivity and specificity for both palpable and nonpalpable breast lesions of 80 and 73 per-
cent, respectively (Bender et al., 1997). In a prospective study of 144 patients (185 breast lesions),
Avril et al. tested the diagnostic ability of FDG-PET using whole-body PET scanners in patients with
masses suggestive of breast cancer (Avril et al., 2000). Histological evaluation of 185 tumors showed a
sensitivity of 68.2 and 91.9 percent for pT1 and pT2 tumors, respectively. The overall specificity was
only 75.5 percent. However, the positive predictive value (fraction of cases positive on PET that were
shown by histology to be malignant) was 96.6 percent. The authors cited partial volume effects (which
worsen as the tumor-to-voxel size decreases) and variation of metabolic activity among tumor types as