Objective: Concerns about gadolinium retention and potential long-term effects have prompted non-contrast imaging approaches. The purpose of our study is to evaluate the diagnostic performance of a combined approach using non-contrast MRI and syntheti...
Objective: Concerns about gadolinium retention and potential long-term effects have prompted non-contrast imaging approaches. The purpose of our study is to evaluate the diagnostic performance of a combined approach using non-contrast MRI and synthetic contrast-enhanced images in patients with newly diagnosed breast cancer.
Material and Methods: MRI examinations in 279 women (mean age 54.7 years; imaged June 2019 to December 2022 at a single institution) with biopsy proven cancer were retrospectively analyzed. Using an extended dimensional U-Net algorithm, synthetic MRIs were generated from non-contrast sequences (T1-weighted and diffusion-weighted imaging [DWI]). Four radiologists analyzed lesion detectability and suspicion scores on non-contrast MRI alone, synthetic contrast-enhanced MRI alone and combined modalities. Diagnostic performance per breast was calculated.
Results: Among 558 breasts in 279 patients, 259 were negative and 299 had enhancing lesions, including 282 cancers (240 invasive and 42 ductal carcinoma in situ [DCIS]) and 17 benign findings. Synthetic contrast-enhanced MRI images showed sensitivity of 91.1% (1028/1128; 95% CI: 88.3, 93.4) and specificity of 95.5% (1054/1104; 95% CI: 92.6, 97.3). Addition of synthetic contrast-enhanced images to non-contrast MRI improved sensitivity from 92.2% (1040/1128; 95% CI: 89.0, 94.5) to 95.9% (1082/1128; 95% CI: 93.8, 97.4) (P < .001) but decreased specificity from 95.8% (1058/1104; 95% CI: 92.9, 97.6) to 94.4% (1042/1104; 95% CI: 91.1, 96.5) (P =.023). Eight cancers were missed by all readers on non-contrast MRI but were detected by at least one reader on combined MRI. Of these, six were DCIS, all presenting as nonmass enhancement, and two were invasive ductal cancers – one a 4.9 cm nonmass enhancement and the other a 1.5 cm mass.
Conclusion: The addition of synthetic contrast-enhanced images improved breast cancer detection compared to non-contrast MRI alone. This combined approach may be particularly useful as an alternative for patients who are unable to receive contrast agents.