Integrating Tomosynthesis With Multimodality Imaging

Assessment of Tomosynthesis Findings

The first step in using tomosynthesis properly relies on careful assessment of the tomosynthesis images. One must thoroughly evaluate the tomosynthesis findings, both in terms of morphology and location, before considering use of other modalities. As discussed previously, this involves careful scrolling through images in the standard projections and localizing findings to a specific region of the breast. Beyond what can be achieved with 2D mammography, when findings are questionable on a tomosynthesis projection, one can more precisely hone in on the corresponding area in another projection to determine how true a finding is. Upon careful review, most real lesions can be seen in both standard projections, although sometimes they are more apparent on one view compared with the other. The craniocaudal (CC) view is generally the most useful projection because findings are often more difficult to perceive on the mediolateral oblique (MLO) view, particularly if there is dense tissue in the upper-outer quadrant.

If a potentially worrisome finding is detected, one should ask if it is new or stable. Review of prior studies, even if 2D only, can often yield information as to whether a lesion has likely been present previously or not. Spot tomosynthesis views can sometimes be useful to further assess very subtle or one-view-only findings. If a subtle finding is again reproduced on spot view(s), the degree of suspicion that a true finding exists is heightened. Conversely, if a “soft” finding is not reproduced, further imaging may not be necessary. However, as with 2D mammography, beware of spotting away potentially significant findings, such as architectural distortion. With careful assessment of the standard CC and MLO tomosynthesis images, the level of suspicion of the finding is often already well established before any additional imaging is performed. The principal value of additional imaging is to determine management and facilitate biopsy, if necessary.

Targeted Ultrasound to Assess Tomosynthesis Findings

Targeted ultrasound is the primary modality used for secondary assessment of tomosynthesis findings, both for its negative predictive value (ie, a normal ultrasound helps to establish a questionable tomosynthesis finding as likely normal) and PPV (a correlative characteristically benign or suspicious sonographic finding helps to direct management and determine final Breast Imaging Reporting and Data System [BI-RADS] assessment). However, correlating tomosynthesis and ultrasound imaging can sometimes present challenges.

Distance from the nipple in centimeters on ultrasound is often different from what is measured on mammography and therefore not completely reliable in correlating with tomosynthesis images. Partly this is due to differences in measuring algorithms, measuring directly from the nipple to the lesion within the breast (mammography) versus to the overlying skin surface (ultrasound). In addition, in mammographic imaging the breast is pulled out and compressed, often yielding a total distance from nipple to chest wall in centimeters much greater than in the natural state. In the supine or posterior-oblique position used for ultrasound scanning, in which the breast tissue flattens out against the chest wall, the total distance may be a fraction of what it is on mammography. Knowledge of the approximate clock face and location of the lesion in the anterior, middle, or posterior breast is useful for honing in on the area for targeted scanning. Landmarks, such as cysts, large calcifications, or lymph nodes, in the vicinity may aid in detection of a subtle sonographic finding. It is important to carefully correlate the size and morphology of the lesion on mammography and ultrasound and avoid satisfaction of search. When using ultrasound to search for a spiculated mass detected on tomosynthesis, do not be satisfied when encountering a benign-appearing cyst. In cases in which detection and correlation of imaging findings are performed, the radiologist should always scan the patient and not rely solely on the technologists’ findings. This is essential to be confident that lesion location and appearance are concordant.

If a sonographic finding is identified in the expected location of a subtle tomosynthesis lesion, use of a skin marker, such as a BB, placed on the skin directly overlying the ultrasound findings with subsequent full or spot tomosynthesis images, can be very helpful to correlate findings between modalities. This provides a more confident determination as to whether the sonographic and tomosynthesis findings match and facilitates determination of the degree of suspicion. Such spot views can be performed in any projection. Spot views in the projection in which the finding was originally best seen are preferred in terms of optimally demonstrating correlation of the findings. Tangential views may be preferable for more superficial lesions but are not always necessary for deeper lesions. For deeper lesions, one must be cognizant that the BB is placed on the closest skin surface during ultrasound scanning when the patient is supine or in the semioblique position and that the breast is compressed against the pectoralis muscle by the ultrasound transducer. Therefore the lesion may appear farther away than expected on mammography, depending on the depth of the finding, size of the breast, and mammographic positioning. The BB should be used only as an approximate guide to assess whether the sonographic finding is in the general vicinity of the tomosynthesis finding.

In challenging tomosynthesis cases, such as an architectural distortion due to radial scars or invasive lobular carcinomas, subtle sonographic findings may be expected ( Fig. 11.1 ). Lesions may be correctly identified on rigorous targeted ultrasound scanning that might otherwise have been overlooked. In some cases these subtle lesions do not always present with the common obvious sonographic findings. They may be seen as vague hypoechoic or isoechoic areas, possibly appearing smaller relative to tomosynthesis, because spiculations are not as obvious on ultrasound. Meticulous scanning can often further define a lesion as either real or not, as well as fine-tune the level of suspicion.

If a sonographic correlate is not initially identified for a convincing suspicious tomosynthesis finding, the backward BB method can be used. In this process an open window (needle localization) paddle is used in the projection at the closest skin surface, and a tomosynthesis spot image is obtained. The finding can be identified, and a mark can be made on the overlying skin. Targeted ultrasound can then be performed with more precise localization and may permit identification of otherwise overlooked subtle sonographic findings ( Fig. 11.2 ).

If a sonographic correlate is established, then the next step is to determine whether or not the combination of findings is suspicious. If both tomosynthesis and ultrasound indicate a benign finding, then the patient can usually return to routine screening. This minimizes costs and patient anxiety that may be associated with a recommendation for short-interval diagnostic imaging follow-up. If there is uncertainty about correlation of imaging findings or if either tomosynthesis or ultrasound findings are not definitive, then such cases may be appropriately classified as probably benign with a recommendation for short-term imaging follow-up. Not unexpectedly, the percentage of BI-RADS 3 recommendations may decrease over time because of the improved diagnostic information provided by tomosynthesis and high-quality ultrasound, which together often permit a more accurate classification of lesions as benign, probably benign, or suspicious. This is a major benefit of tomosynthesis that stems from the careful use of the technology and meticulous hands-on ultrasound scanning.

If the tomosynthesis and/or ultrasound findings are suspicious, then biopsy will be indicated. Core needle biopsy is most easily performed using sonographic guidance. Following ultrasound-guided core needle biopsy, a marker clip should be placed and a post-biopsy mammogram should be obtained to provide the direct proof of sonographic and tomosynthesis correlation ( Fig. 11.3 ). If the marker is found not to reside at the tomosynthesis finding or if no sonographic correlate was originally found, stereotactic biopsy under tomosynthesis guidance (if available) would be indicated. Otherwise, additional imaging with MRI would be appropriate. If MRI is performed and is negative in the region in question, mammographic follow-up would likely be reasonable. If MRI shows a suspicious correlative finding, then biopsy could be performed under MR- or tomosynthesis-guided biopsy, if available.

Numerous studies demonstrate that tomosynthesis reduces screening recalls; however, there are some instances where benign breast findings are accentuated on tomosynthesis. These cases may be infrequent but are important because they can present a challenge for both the novice and seasoned breast imager. Such cases include a phenomenon observed in tomosynthesis that could be termed pseudo-architectural distortion . Pseudo-architectural distortion can result from clumped fibroglandular tissue with crossing vascular and trabecular structures and, unlike most true architectural distortions produced by malignancies or complex sclerosing lesions, is usually perceived in only one tomosynthesis view. Some women with heterogeneously dense tissue will have areas mimicking a true architectural distortion.

With tomosynthesis, the perception of architectural distortion is heightened, potentially leading to false-positive recalls and additional diagnostic work-ups. In these cases spot tomosynthesis views may be performed in the projection in which the finding was originally identified. If the finding is not reproduced, it was likely summation artifact. Examination of the surrounding architecture of the tissue is crucial to ensure the lesion in question has indeed been included within the area covered by the compression paddle. Combined 2D and tomosynthesis spot views are usually preferred in this setting because this allows better evaluation of surrounding tissue landmarks to ensure the proper region was captured by the compression paddle. Frequently, spot compression views will result in effacement of the pseudodistortion. In patients with dense tissue, ultrasound of the region in question may still be warranted for complete assessment because some subtle malignant distortions can efface even on tomosynthesis spot compression views. Negative sonographic findings can lead to two different management scenarios. If completely normal tissue is noted in the region on ultrasound, it can increase confidence that the questioned tomosynthesis finding was not real. In such cases routine or short-interval follow-up can be recommended. However, if the tomosynthesis findings are concerning, then tomosynthesis-guided biopsy or further imaging with MRI may be indicated ( Fig. 11.4 ).

Tomosynthesis and Evaluation of Extent of Disease

Tomosynthesis better estimates the size of malignant lesions, as well as detects additional sites of disease obscured on 2D mammography. This is particularly true of small lesions and lesions in heterogeneous and extremely dense tissue. Even in nondense tissue, additional cancers may be detected ( Fig. 11.5 ). If a malignant lesion has been diagnosed, the rest of the remaining breast tissue and the contralateral breast should be carefully examined. Additional lesions may be detected, and targeted ultrasound and biopsies may be performed that may have a significant effect on management, potentially precluding the need for MRI and all the additional costs and associated procedures.