Page 381 - Biomedical Engineering and Design Handbook Volume 2, Applications
P. 381

BREAST IMAGING SYSTEMS: DESIGN CHALLENGES FOR ENGINEERS  359

                          differentiated from those due to benign processes by their shape. Accurate shape depiction requires
                          resolution superior to that necessary for simple detection. In a similar fashion, malignant and
                          benign masses are often distinguished by the appearance of their borders (margins), with spiculated
                          margins having a much higher correlation with malignancy than smooth margins. To overcome
                          this limitation, developers of breast CT imaging systems are working with x-ray tube developers
                          to build special low kVp and smaller focal spot x-ray tubes for breast CT imaging. A second chal-
                          lenge for breast CT is imaging breast tissue lying adjacent to the chest wall. This problem arises
                          because of the nonzero thickness of the table upon which the patient lies prone, and the difficulty
                          of positioning the x-ray focal spot and active portion of the detector very close to the bottom sur-
                          face of the table. In terms of promoting clinical acceptance of breast CT, an advancement that
                          would add to its clinical utility is the development of techniques for image-guided biopsy, such as
                          those currently available using stereotactic x-ray imaging. Accurate needle placement is likely to
                          be more difficult for  the uncompressed and unrestrained breast than for a compressed breast.
                          However, this technical challenge can likely be overcome using image-guided biopsy approaches
                          similar to those used for MRI-guided breast biopsy.



              12.4.3 MRI
                          At present, mammography is the primary imaging modality used to detect early clinically occult
                          breast cancer. Despite advances in mammographic technique, there are limitations in sensitivity and
                          specificity that remain. These limitations have stimulated exploration into alternative or adjunctive
                          imaging techniques. MRI of the breast provides higher soft tissue contrast than conventional mam-
                          mography. This provides the potential for improved lesion detection and characterization. Studies have
                          already demonstrated the potential for breast MRI to distinguish benign from malignant breast lesions
                          and to detect mammographically and clinically occult cancer (Dash et al., 1986; El Yousef et al., 1984;
                          Stelling et al., 1985). The first studies using MRI to detect both benign and malignant breast lesions
                          concluded that it was not possible to detect and characterize lesions on the basis of signal intensities on
                          T1-and T2-weighted images (Dash et al., 1986; El Yousef et al., 1984; Stelling et al., 1985). However,
                          reports on the use of gadolinium-enhanced breast MRI were more encouraging. Cancers enhance rel-
                          ative to other breast tissues following the administration of intravenous Gd-DTPA (Kaiser and
                          Zeitler, 1989). Indeed, in one study, 20 percent of cancers were seen only after the administration
                          of Gd-DTPA (Heywang et al., 1989). In addition, several studies reported on MRI detection of
                          breast cancer not visible on mammography (Harms et al., 1993; Heywang et al., 1989). The detec-
                          tion of mammographically occult multifocal cancer in up to 30 percent of patients has led to the
                          recommendations that MRI can be successfully used to stage patients who are potential candidates
                          for breast conservation therapy (Harms et al., 1993). Figure 12.5 shows an example of mammo-
                          graphically occult ductal carcinoma in situ (DCIS) identified via gadolinium-enhanced MRI.
                            However, the presence of contrast enhancement alone is not specific for distinguishing malignant
                          from benign breast lesions. Benign lesions frequently enhance after Gd injection on MRI, including
                          fibroadenomas, benign proliferative change, and inflammatory change. To improve specificity, some
                          investigators recommend dynamic imaging of breast lesions to evaluate the kinetics of enhancement
                          (Heywang et al., 1989; Kaiser and Zeitler, 1989; Stack et al., 1990). Using this technique, cancers
                          have been found to enhance intensely in the early phases of the contrast injection. Benign lesions
                          enhance variably but in a more delayed fashion than malignant lesions. Recently, the American
                          College of Radiology has published its reporting lexicon for breast MRI, which incorporates both
                          lesion morphology and kinetic information to diagnose MR-depicted breast lesions (American
                          College of Radiology, 2003).
                            Because x-ray mammography is limited by its relatively low specificity, MRI has been suggested
                          as an adjunct imaging study to evaluate patients with abnormal mammograms. One application of
                          breast MRI may be to reduce the number of benign biopsies performed as a result of screening and
                          diagnostic mammography work-up. To distinguish benign from malignant breast lesions using MRI
                          scanning, most investigators rely on studying the time course of signal intensity changes of a breast
                          lesion after contrast injection. However, among reported studies, the scan protocols varied widely
   376   377   378   379   380   381   382   383   384   385   386