Contrast-enhanced mammography (CEM) is a functional imaging technique that integrates anatomical and vascular information to aid in the detection of breast lesions. By administering intravenous contrast and using dual-energy acquisition methods, CEM enhances the visualisation of abnormal tissue, particularly in individuals with dense breast tissue. While CEM has demonstrated comparable sensitivity to magnetic resonance imaging (MRI), it continues to face limitations in specificity. This shortfall can lead to increased rates of unnecessary patient recalls and benign biopsies. To address this, a recent investigation evaluated both qualitative descriptors, such as lesion conspicuity and quantitative measures of enhancement. The goal was to determine whether these parameters could improve the diagnostic performance of CEM, especially in differentiating between benign conditions, background parenchymal enhancement (BPE) and malignant lesions. Additionally, enhancement patterns on CEM were compared with dynamic contrast-enhanced MRI (DCE-MRI) in a subset of cases to assess concordance.
Lesion Conspicuity and Its Diagnostic Value
Lesion conspicuity, as defined in the updated BI-RADS CEM lexicon, was used by experienced breast radiologists to qualitatively assess lesions. Each lesion was categorised as having low, moderate or high conspicuity based on the degree to which it stood out from the background tissue. A clear relationship emerged between conspicuity levels and lesion histology. Among the low-conspicuity lesions, the majority were found to be normal or benign. Conversely, most malignant lesions demonstrated either moderate or high conspicuity. In fact, all highly conspicuous lesions in the study were confirmed to be invasive cancers. The conspicuity assessment displayed strong interobserver agreement, underscoring its reliability as a diagnostic criterion. Notably, reader agreement was highest for clearly enhancing or non-enhancing lesions and slightly lower in the evaluation of BPE, where subjective perception played a greater role. These findings suggest that conspicuity could serve as a practical marker for malignancy in routine CEM interpretation, potentially reducing the number of false-positive cases and avoiding unnecessary interventions.
Must Read: AI-Integrated Mammography Screening: An Alternative to Double Reading
Quantitative Enhancement Metrics and Pattern Analysis
In addition to the qualitative analysis, the study also focused on quantifying enhancement using two specific time points in the imaging sequence: the early craniocaudal (CC) view and the later mediolateral oblique (MLO) view. These were referred to as CEearly and CElate, respectively. Enhancement was measured using the ratio of the maximum signal within the lesion compared to the background, and the relative signal difference (%RSD) was calculated to assess enhancement kinetics. CEearly proved to be the most reliable metric, providing superior performance in distinguishing malignant from benign lesions and BPE, with specificity reaching 90% in differentiating cancers from BPE. These results indicate that enhancement intensity measured shortly after contrast administration may be particularly valuable in identifying malignant lesions. Furthermore, %RSD was used to classify enhancement into progressive, plateau or wash-out patterns. Progressive enhancement was more common in benign or BPE cases, while plateau and wash-out patterns were frequently associated with malignancies. However, these quantitative metrics were not significantly different between invasive and non-invasive cancers. Non-mass enhancing lesions that were benign typically had lower CEearly values than their malignant counterparts, which supports the use of this measurement in improving lesion characterisation on CEM.
Comparative Agreement with DCE-MRI
A subgroup analysis was conducted for lesions that were imaged using both CEM and DCE-MRI within a short time interval. The comparison aimed to assess the consistency of enhancement patterns across both modalities. The level of agreement was generally moderate, with the strongest concordance observed for lesions displaying a wash-out pattern. Such lesions demonstrated similar behaviour on both CEM and DCE-MRI, indicating that CEM can be a viable alternative when MRI access is limited. For lesions with progressive or plateau enhancement, the agreement between the two imaging methods was lower. This variation suggests that while CEM and DCE-MRI are broadly comparable in detecting certain malignancy-related patterns, discrepancies may arise in more subtle or slowly enhancing lesions. Importantly, the method for calculating enhancement relied on the maximum pixel value within the lesion, rather than the average. This choice was intended to provide a consistent, reproducible measure that reflects the peak signal in heterogeneous lesions, despite the potential influence of noise.
This study highlights the diagnostic benefits of integrating lesion conspicuity assessment with quantitative enhancement analysis in contrast-enhanced mammography. CEearly, representing contrast from the early imaging view, demonstrated the best performance in identifying malignancies and distinguishing them from benign lesions and BPE. Lesion conspicuity showed strong agreement among radiologists and was significantly associated with histological outcomes, supporting its value in routine CEM reporting. Although enhancement patterns and metrics did not significantly differentiate between invasive and non-invasive cancers, the findings support the broader use of enhancement analysis in improving diagnostic specificity. Agreement with DCE-MRI further supports the clinical relevance of CEM, particularly for lesions with wash-out enhancement. These results suggest that standardising enhancement assessment protocols and integrating conspicuity descriptors could enhance diagnostic accuracy and reduce unnecessary procedures in breast imaging.
Source: European Radiology
Image Credit: Freepik
