An investigation of tomosynthesis with regard to the diagnostic efficacy of spot compression mammography

Both DBT in standard views and CSC are clinically available for patients who have ambiguous findings in FFDM, such as vague masses, architectural distortions, and asymmetries.^11,12,13,14. DBT in standard views not only shows more detailed structures of lesions but also helps to localize findings in one view. Therefore, tomosynthesis improves the clinical accuracy of mammography by increasing both sensitivity and specificity^17,18At the same time, CSC also helps improve clinical accuracy by reducing glandular overlap^12,13. TSC combines the advantages of tomosynthesis and point compression and is therefore expected to have better diagnostic efficiency. The results of this study show that the diagnostic efficacy using CSC data is significantly improved compared with TSC data, especially in patients with dense breast tissue. This suggests that TSC may be a better choice than CSC for the evaluation of patients with equivocal findings in FFDM.

We examined cases assigned different BI-RADS categories on CSC and TSC images. For cases with malignant lesions, both readers assigned higher BI-RADS categories to TSC images than to CSC images. TSC may show blurred or spiky edges of masses that are obscured on CSC or architectural distortions around asymmetric lesions that are not obvious on CSC. Most cases of benign lesions were assigned lower BI-RADS categories by both readers on TSC images than on CSC images. Analysis showed that the edge of the circumscribed mass was better seen on TSC images than on CSC images. Another reason was that TSC revealed associated benign features, such as the mass’s “halo sign,” a ring of compressed low-density adipose tissue around the mass. The advantage of TSC may be its ability to resolve tissue overlap and better visualize the lesion margin and surrounding structures.

It was observed that three cases of benign lesions were assigned a higher BI-RADS category by a less experienced reader on TSC images than on CSC images. All three cases showed asymmetries, suggesting that some asymmetries are difficult to detect for less experienced readers. Previous studies^19,20 suggest that training for less experienced radiologists is important to improve the learning curve and fully exploit the benefits of tomosynthesis.

Our results show that for an experienced reader (with 10 years of experience), the diagnostic performance was significantly better when using TSC images than when using CSC images, including higher AUC, accuracy, specificity, PPV, and NPV. For a less experienced reader (with 4 years of experience), the diagnostic efficacy was still better when using TSC images compared with CSC images, with higher AUC and accuracy. Although the use of TSC improved the diagnostic efficacy of both readers, a certain level of experience may be required to realize its full potential. This shows that standardized training of radiologists is essential for the application of TSC in clinical practice.

The study cohort mainly consisted of patients with dense breast tissue. The data collected from patients with dense breast tissue showed that the improvement in diagnostic efficacy using TSC data is significant compared with CSC data. However, the sample size of patients with non-dense breast tissue was too small to perform meaningful statistical analysis. Therefore, the diagnostic efficacy of TSC for non-dense breasts is still unclear, which requires further investigation with a larger sample size.

Our study shows that TSC has a dose that is about 20% higher than CSC, but it is still within the safety limits. Compared to the significant improvement in diagnostic performance, the slight and tolerable dose increase is well justified.

Our research results are clinically significant for the following reasons: (1) the study preliminarily demonstrates that CSC-based tomosynthesis can continue to exert its advantages, expands the application scenarios of tomosynthesis, and verifies the feasibility and radiation dose of TSC; (2) it can help optimize the examination process by proving that TSC may be a better choice for evaluating patients with equivocal findings in FFDM and with dense breast compared with CSC, especially in the areas where mammography is the primary examination technique for breast disease; and (3) Tagliafico et al.¹¹ found that the diagnostic accuracy of DBT in standard views is at least equal to that of CSC. Since our results show that the diagnostic efficiency of TSC is better than that of CSC, we speculate that the efficiency of TSC is also better than that of DBT in standard views, although this needs to be verified in our next study.

Limitations of this study include its retrospective design and small sample size. We hope to conduct further investigations with a larger number of patients. In addition, few patients in our center underwent both DBT in standard views and TSC; therefore, TSC and DBT in standard views could not be directly compared. We would like to investigate this further with a larger volume of data. In addition, all images were acquired using a mammography device from a single manufacturer. A future multicenter study may provide better results.

In conclusion, TSC offers better diagnostic efficacy compared to CSC at a slightly increased dose. It is a promising alternative to CSC for the evaluation of patients with equivocal findings in FFDM.