Page 383 - Biomedical Engineering and Design Handbook Volume 2, Applications
P. 383

BREAST IMAGING SYSTEMS: DESIGN CHALLENGES FOR ENGINEERS  361

                            Also widely varying among investigators are their criteria used for differentiating benign from
                          malignant lesions. Criteria vary from simplistic models that report the percent of lesion enhancement
                          at 2 min following contrast injection to more sophisticated physiologic models that take into account
                          the initial T1 characteristics of a lesion and calculate Gd concentration as a function of time in order
                          to extract pharmacokinetic parameters. Thus, wide variability in the accuracy cited by these investi-
                          gators for differentiating benign from malignant lesions has been reported (66 to 93 percent) (Daniel
                          et al., 1998; Esserman et al., 1999; Gilles et al., 1994; Hickman et al., 1994; Hylton, 1999; Orel et al.,
                          1994; Perman et al., 1994; Van Vaals et al., 1993). Despite the many differing techniques, it is clear
                          that there is a tendency for cancer to enhance more rapidly than benign lesions after bolus injection
                          of Gd chelate. However, it is also clear that overlap exists in dynamic curves between cancerous and
                          noncancerous lesions, resulting in false-negative diagnosis in all reported series and false-positive
                          diagnosis in many. The development and availability of Breast MRI CAD Systems has improved the
                          interpretation of BMRI tremendously. These software tools perform automated subtraction of con-
                          trast enhanced scans obtained at sequential time points during the MRI examination from the initial
                          precontrast scan. The resulting images depict any portion of the breast or a lesions that enhanced and
                          removes all non-enhancing (normal) structures; the data from the sequential images are then used to
                          calculate the temporal time intensity curves that numerically demonstrate the inflow and egress of
                          contract media into/out of breast lesions. In addition, MIP images and other 3-dimensional image
                          reconstructions are easily accomplished to assist the interpreting physician.
                            Alternative approaches to characterizing enhancing lesions on breast MRI include extracting
                          architectural features that describe breast lesions. The superior soft tissue contrast of MRI and use
                          of higher spatial resolution techniques have prompted investigations in this area and the development
                          of a lexicon for interpreting and reporting breast MRI scans (Gilles et al., 1994; Nunes et al., 1997;
                          Orel et al., 1994; Schnall et al., 2001; American College of Radiology, 2003). Such an advancement
                          would improve the widespread general reliability and comparability among breast MRI examinations
                          performed from one institution to another. Clearly, the relative importance of spatial and temporal res-
                          olution in this regard requires further evaluation.
                            Other reported investigations of breast MRI suggest that MRI demonstrates more extensive
                          cancer than indicated by mammography or predicted by clinical breast examination. Several inves-
                          tigators have now demonstrated that MRI can detect breast cancer that is mammographically occult
                          (Harms et al., 1993; Heywang et al., 1989; Sardanelli et al., 2007a; Sardanelli et al., 2007b; Port
                          et al., 2007; Lehman et al., 2005; Leach et al., 2005; Kuhl et al., 2005; Kriege et al., 2004; Warner
                          et al., 2004) and suggest that MRI may have a role as a screening examination for patients with a high
                          genetic predisposition to breast cancer and in those populations of women having extremely radio-
                          dense breast tissue on x-ray mammography. The American Cancer Society recently published its guide-
                          lines for breast screening with MRI as an adjunct to mammography, wherein they defined their specific
                          recommendations, the lifetime risk of populations comprising their specific recommendations, and the
                          strength of the evidence on which they based their recommendations (Saslow et al., 2007). An annual
                          MRI screening study along with an annual screening mammogram for women having a BRCA muta-
                          tion that have not been tested themselves for the BRCA mutation, and women having a lifetime risk of
                          20 to 25 percent or greater for developing breast cancer as defined by the BRCAPRO (Parmigiani et al.,
                          1998; Berry et al., 1997) or other breast cancer estimation models that are dependent on family history.
                          This recommendation is based on evidence from nonrandomized screening trials and observational stud-
                          ies (Sardanelli et al., 2007a; Sardanelli et al., 2007b; Port et al., 2007; Lehman et al., 2005; Leach et al.,
                          2005; Kuhl et al., 2005; Kriege et al., 2004; Warner et al., 2004).
                            The recommendations for annual MRI breast screening along with annual mammography for
                          other populations is based on expert concensus opinion regarding the lifetime risk for breast
                          cancer. These populations include women who underwent radiation to the chest between the age of
                          10 and 30 years and those diagnosed with having Li-Fraumeni syndrome, Cowden syndrome, and
                          Bannayan-Riley-Ruvalcaba syndrome, and individuals with  first-degree relatives diagnosed with
                          these syndromes. The panel found insufficient evidence to recommend for or against MRI screening
                          in women whose lifetime risk for developing breast cancer is 15 to 20 percent as defined by
                          BRCAPRO or other models largely dependent on family history, women with a prior pathologic diag-
                          nosis of lobular carcinoma in situ, atypical lobular hyperplasia, or in situ or invasive breast cancer.
   378   379   380   381   382   383   384   385   386   387   388