Improving the diagnostic performance of contrast-enhanced mammography through lesion conspicuity and enhancement quantification.
No Thumbnail Available
Authors
Allajbeu I.
Nanaa M.
Manavaki R.
Papalouka V.
Bene I.
Payne N.
Giannotti E.
van Nijnatten T.
KilburnToppin F.
Healy N.
Check for full-text access
Issue Date
2025
Type
Article
Language
Keywords
Alternative Title
Abstract
Objectives: To analyze qualitative and quantitative enhancement of breast lesions on CEM and their impact on specificity and overall diagnostic performance in predicting malignancy. A secondary objective was to compare lesion enhancement patterns between CEM and contrast-enhanced (CE)-MRI. Method(s): The cohort included screening and symptomatic cases from CEM research studies (December 2016-March 2023) with an identifiable lesion. Three breast radiologists independently assessed lesion conspicuity as low, moderate, or high, based on the BI-RADS CEM lexicon. Lesion enhancement was quantified by drawing two regions of interest representing lesion and background parenchyma, to calculate contrast enhancement from the early (CEearly) and late (CElate) views. Area-under-the-curve (AUC) was used to assess diagnostic performance, with thresholds determined using the maximum Youden index. Cohen's kappa was used to measure agreement between CEM and DCE-MRI enhancement patterns. p-values < 0.05 were deemed statistically significant. Result(s): From 503 CEM studies, 143 BI-RADS 2-5 lesions were analyzed. Lesion conspicuity was significantly associated with lesion histology (p ) views. Area-under-the-curve (AUC) was used to assess diagnostic performance, with thresholds determined using the maximum Youden index. Cohen's kappa was used to measure agreement between CEM and DCE-MRI enhancement patterns. p-values < 0.05 were deemed statistically significant. Result(s): From 503 CEM studies, 143 BI-RADS 2-5 lesions were analyzed. Lesion conspicuity was significantly associated with lesion histology (p early, CElate), and enhancement patterns on CEM recombined images. CEearly performed better in differentiating malignant from benign lesions or background parenchymal enhancement (BPE), with AUC values of 0.83 and 0.88 and 90% specificity in distinguishing BPE from cancers. There was fair/moderate agreement between lesion enhancement patterns on CEM and DCE-MRI (Cohen's kappa = 0.35, p < 0.001), with a higher agreement for lesions exhibiting a wash-out pattern (Cohen's kappa = 0.5, p < 0.001). Conclusion(s): Both conspicuity and quantification of lesion enhancement can improve CEM specificity in predicting malignancy, with CEearly offering the best diagnostic performance. Key Points: Question Quantifying lesion enhancement conspicuity on contrast-enhanced mammography (CEM) has demonstrated potential in differentiating malignancy from benign lesions and BPE. Finding Contrast from the early recombined view (CEearly) performed better in discriminating malignant from benign lesions and BPE, with 90% specificity for BPE vs cancers. Clinical relevance Conspicuity and quantification of lesion enhancement on CEM can improve the specificity and overall diagnostic performance of CEM in cancer detection. Implementation of conspicuity thresholds in routine CEM interpretation could potentially reduce unnecessary recalls and benign biopsies. Copyright © The Author(s) 2025.
Description
Citation
Publisher
License
Journal
European radiology
Volume
35