Diffusion probabilistic versus generative adversarial models to reduce contrast agent dose in breast MRI.

BACKGROUND: To compare denoising diffusion probabilistic models (DDPM) and generative adversarial networks (GAN) for recovering contrast-enhanced breast magnetic resonance imaging (MRI) subtraction images from virtual low-dose subtraction images. METHODS: Retrospective, ethically approved study. DDPM- and GAN-reconstructed single-slice subtraction images of 50 breasts with enhancing lesions were compared to original ones at three dose levels (25%, 10%, 5%) using quantitative measures and radiologic evaluations. Two radiologists stated their preference based on the reconstruction quality and scored the lesion conspicuity as compared to the original, blinded to the model. Fifty lesion-free maximum intensity projections were evaluated for the presence of false-positives. Results were compared between models and dose levels, using generalized linear mixed models. RESULTS: At 5% dose, both radiologists preferred the GAN-generated images, whereas at 25% dose, both radiologists preferred the DDPM-generated images. Median lesion conspicuity scores did not differ between GAN and DDPM at 25% dose (5 versus 5, p = 1.000) and 10% dose (4 versus 4, p = 1.000). At 5% dose, both readers assigned higher conspicuity to the GAN than to the DDPM (3 versus 2, p = 0.007). In the lesion-free examinations, DDPM and GAN showed no differences in the false-positive rate at 5% (15% versus 22%), 10% (10% versus 6%), and 25% (6% versus 4%) (p = 1.000). CONCLUSIONS: Both GAN and DDPM yielded promising results in low-dose image reconstruction. However, neither of them showed superior results over the other model for all dose levels and evaluation metrics. Further development is needed to counteract false-positives. RELEVANCE STATEMENT: For MRI-based breast cancer screening, reducing the contrast agent dose is desirable. Diffusion probabilistic models and generative adversarial networks were capable of retrospectively enhancing the signal of low-dose images. Hence, they may supplement imaging with reduced doses in the future. KEY POINTS: • Deep learning may help recover signal in low-dose contrast-enhanced breast MRI. • Two models (DDPM and GAN) were trained at different dose levels. • Radiologists preferred DDPM at 25%, and GAN images at 5% dose. • Lesion conspicuity between DDPM and GAN was similar, except at 5% dose. • GAN and DDPM yield promising results in low-dose image reconstruction.

Fibroglandular tissue segmentation in breast MRI using vision transformers: a multi-institutional evaluation.

Accurate and automatic segmentation of fibroglandular tissue in breast MRI screening is essential for the quantification of breast density and background parenchymal enhancement. In this retrospective study, we developed and evaluated a transformer-based neural network for breast segmentation (TraBS) in multi-institutional MRI data, and compared its performance to the well established convolutional neural network nnUNet. TraBS and nnUNet were trained and tested on 200 internal and 40 external breast MRI examinations using manual segmentations generated by experienced human readers. Segmentation performance was assessed in terms of the Dice score and the average symmetric surface distance. The Dice score for nnUNet was lower than for TraBS on the internal testset (0.909 ± 0.069 versus 0.916 ± 0.067, P < 0.001) and on the external testset (0.824 ± 0.144 versus 0.864 ± 0.081, P = 0.004). Moreover, the average symmetric surface distance was higher (= worse) for nnUNet than for TraBS on the internal (0.657 ± 2.856 versus 0.548 ± 2.195, P = 0.001) and on the external testset (0.727 ± 0.620 versus 0.584 ± 0.413, P = 0.03). Our study demonstrates that transformer-based networks improve the quality of fibroglandular tissue segmentation in breast MRI compared to convolutional-based models like nnUNet. These findings might help to enhance the accuracy of breast density and parenchymal enhancement quantification in breast MRI screening.

Using Machine Learning to Reduce the Need for Contrast Agents in Breast MRI through Synthetic Images.

Background Reducing the amount of contrast agent needed for contrast-enhanced breast MRI is desirable. Purpose To investigate if generative adversarial networks (GANs) can recover contrast-enhanced breast MRI scans from unenhanced images and virtual low-contrast-enhanced images. Materials and Methods In this retrospective study of breast MRI performed from January 2010 to December 2019, simulated low-contrast images were produced by adding virtual noise to the existing contrast-enhanced images. GANs were then trained to recover the contrast-enhanced images from the simulated low-contrast images (approach A) or from the unenhanced T1- and T2-weighted images (approach B). Two experienced radiologists were tasked with distinguishing between real and synthesized contrast-enhanced images using both approaches. Image appearance and conspicuity of enhancing lesions on the real versus synthesized contrast-enhanced images were independently compared and rated on a five-point Likert scale. P values were calculated by using bootstrapping. Results A total of 9751 breast MRI examinations from 5086 patients (mean age, 56 years ± 10 [SD]) were included. Readers who were blinded to the nature of the images could not distinguish real from synthetic contrast-enhanced images (average accuracy of differentiation: approach A, 52 of 100; approach B, 61 of 100). The test set included images with and without enhancing lesions (29 enhancing masses and 21 nonmass enhancement; 50 total). When readers who were not blinded compared the appearance of the real versus synthetic contrast-enhanced images side by side, approach A image ratings were significantly higher than those of approach B (mean rating, 4.6 ± 0.1 vs 3.0 ± 0.2; P < .001), with the noninferiority margin met by synthetic images from approach A (P < .001) but not B (P > .99). Conclusion Generative adversarial networks may be useful to enable breast MRI with reduced contrast agent dose. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Bahl in this issue.

Clinical evaluation of cylindrical regional suppression in dynamic contrast-enhanced breast MRI: An intra-individual comparison study on image quality and lesion conspicuity.

PURPOSE: To evaluate the effect of a cylindrical regional-suppression technique (CREST) on image quality and lesion conspicuity in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) of the breast. METHOD: This was a comparative study of 67 women with 44 lesions who underwent breast DCE-MRI with CREST (CREST-DCE) and had a previous DCE-MRI without CREST (conv-DCE) available. Two radiologists assessed image quality parameters and lesion conspicuity using five-point Likert scales. In an intra-individual comparison, the effects of CREST on image quality (strong degradation to strong improvement) were assessed. Moreover, both radiologists identified the post-contrast phase, which benefited the most from using CREST in direct comparison. The statistical analysis included the Wilcoxon signed-rank test. RESULTS: Cardiac motion-rated artefacts were significantly reduced in CREST-DCE compared to conv-DCE (3.6 ± 1.2 [CREST-DCE] vs 2.1 ± 0.8 [conv-DCE], p < 0.001). At the axilla, the visualisation of anatomical structures (3.9 ± 1.0 vs 2.3 ± 1.2, p < 0.001) and the skin contour (4.3 ± 0.8 vs 3.0 ± 1.1, p < 0.001) were significantly improved in CREST-DCE, whereas ghosting artefacts were significantly less pronounced (3.8 ± 1.1 vs 2.4 ± 1.0, p < 0.001). The parasternal region was similarly assessable using both techniques (4.3 ± 1.1 vs 4.2 ± 1.2, p = 0.47). In direct comparison, CREST-DCE images were classified as "improved" in 54/67 and "equivalent" in 13/67 exams. The effects of CREST were found to be most pronounced in the very early post-contrast phase (32/67). The lesion conspicuity was rated similar for CREST and conv-DCE (4.7 ± 0.7 vs 4.8 ± 0.2, p = 0.18). CONCLUSIONS: CREST appears to be an effective tool to reduce cardiac motion-related artefacts and, therefore, may improve image quality in breast DCE-MRI without impairing lesion conspicuity.

Artificial Intelligence for Clinical Interpretation of Bedside Chest Radiographs.

Background Supine chest radiography for bedridden patients in intensive care units (ICUs) is one of the most frequently ordered imaging studies worldwide. Purpose To evaluate the diagnostic performance of a neural network-based model that is trained on structured semiquantitative radiologic reports of bedside chest radiographs. Materials and Methods For this retrospective single-center study, children and adults in the ICU of a university hospital who had been imaged using bedside chest radiography from January 2009 to December 2020 were reported by using a structured and itemized template. Ninety-eight radiologists rated the radiographs semiquantitatively for the severity of disease patterns. These data were used to train a neural network to identify cardiomegaly, pulmonary congestion, pleural effusion, pulmonary opacities, and atelectasis. A held-out internal test set (100 radiographs from 100 patients) that was assessed independently by an expert panel of six radiologists provided the ground truth. Individual assessments by each of these six radiologists, by two nonradiologist physicians in the ICU, and by the neural network were compared with the ground truth. Separately, the nonradiologist physicians assessed the images without and with preliminary readings provided by the neural network. The weighted Cohen κ coefficient was used to measure agreement between the readers and the ground truth. Results A total of 193 566 radiographs in 45 016 patients (mean age, 66 years ± 16 [SD]; 61% men) were included and divided into training (n = 122 294; 64%), validation (n = 31 243; 16%), and test (n = 40 029; 20%) sets. The neural network exhibited higher agreement with a majority vote of the expert panel (κ = 0.86) than each individual radiologist compared with the majority vote of the expert panel (κ = 0.81 to ≤0.84). When the neural network provided preliminary readings, the reports of the nonradiologist physicians improved considerably (aided vs unaided, κ = 0.87 vs 0.79, respectively; P < .001). Conclusion A neural network trained with structured semiquantitative bedside chest radiography reports allowed nonradiologist physicians improved interpretations compared with the consensus reading of expert radiologists. © RSNA, 2022 Supplemental material is available for this article. See also the editorial by Wielpütz in this issue.

Feasibility study of 2D Dixon-Magnetic Resonance Fingerprinting (MRF) of breast cancer.

Purpose: Application of MRF to evaluate the feasibility of 2D Dixon blurring-corrected MRF (2DDb-cMRF) to differentiate breast cancer (BC) from normal fibroglandular tissue (FGT). Methods: Prospective study on 14 patients with unilateral BC on 1.5 T system/axial T2w-TSE sequence, 2DDb-cMRF, B1 map, dynamic contrast-enhanced (DCE) T1-w GE-series. Mean T1 and T2 values and standard deviations were computed in the BC-/FGT-ROI on pre-/post-contrast MRF-maps and their differences were tested by two-tailed student t-test.Accuracy and repeatability of MRF were evaluated in a phantom experiment with gelatin with Primovist surrounded by fat.The T1 reduction between pre-/post-contrast MRF-maps was correlated to DCE signal enhancement in the last image post-contrast through the Pearson´s correlation coefficient (r) and for the phantom validation experiment through the Lin's concordance correlation coefficient (CCC).Visual evaluation of cancers on MRF-Maps was performed by rating each MRF-Map by 3 radiologists. Results: T1- and T2-MRF values of BC vs. FGT were for T1 and T2 pre-contrast respectively: 1147 ± 1 ms vs. 1052 ± 9 ms (p = 0.007) and 83 ± 1 ms vs. 73 ± 1 ms (p = 0.03); post-contrast respectively: 367.3 ± 121.5 ms vs. 690.3 ± 200.3 ms (p = 0.0005) and 76.9 ± 11.5 ms vs. 69.8 ± 15.2 ms (p = 0.12). r was positive (FGT r = 0.7; BC r = 0.6). CCC was 0.999 for T1 and 0.994 for T2. In the T1- and T2-MRF-Maps before contrast respectively (7,7,8)/14 and (5,9,8)/14 cancers were visible to the readers; afterwards, (11,12,12)/14 and (5,6,11)/14. Conclusions: MRF is promising for distinction between BC and FGT as well as for analyzing pre-/post-contrast T1 changes. However, its potential for differential diagnosis warrants further studies.

Digital breast tomosynthesis-guided vacuum-assisted biopsy of suspicious calcifications at different sites within one breast: Is biopsy of more than one location needed?

OBJECTIVE: To investigate how often biopsy of two sites of morphologically similar or equally suspicious calcifications within the same breast yield differing histopathologic results, and how this may affect clinical management. MATERIALS AND METHODS: We identified patients with two or more sites of calcifications categorized as Breast Imaging Reporting and Data System (BI-RADS) ≥ 4b within the same breast who underwent digital breast tomosynthesis-guided vacuum-assisted biopsy (DBT-guided VAB). We analyzed how often biopsy of two distinct sites yielded the same or differing histopathologic findings. The histopathologic findings were dichotomized into "actionable" and "non-actionable", depending on the respective further management. We then analyzed how often the consecutive management would have been the same or different. RESULTS: Of 206 women undergoing DBT-guided VAB at our institution within 24 months, 21 consecutive patients (54 ± 10.2 years; range: 35-71) underwent DBT-guided VAB of two distinct sites of calcifications. Management of histologic findings was the same (both sites actionable or both sites non-actionable) in 12/21 (57 %), different in the remaining 9/21 patients (43 %). Of the nine patients whose differing histologic findings would have led to different clinical management, 4/9 had a high-risk lesion (atypical ductal hyperplasia n = 3, papilloma with epithelial atypia n = 1) vs benign changes (adenosis n = 4), 2/9 had high-grade DCIS vs benign changes (adenosis n = 1, fat necrosis n = 1), and 3/9 had invasive cancer (luminal A n = 2, luminal B n = 1) with high-grade DCIS vs pure high-grade DCIS. CONCLUSIONS: Multiple sites of calcifications within the same breast, even when morphologically similar or equally suspicious, may represent different histopathologic findings with different clinical management implications. Accordingly, in the presence of suspicious calcifications at multiple distinct sites within the same breast, biopsy of more than one site of calcification should be considered.

Background parenchymal enhancement in contrast-enhanced MR imaging suggests systemic effects of intrauterine contraceptive devices.

OBJECTIVES: Levonorgestrel-releasing intrauterine contraceptive devices (LNG-IUDs) are designed to exhibit only local hormonal effects. There is an ongoing debate on whether LNG-IUDs can have side effects similar to systemic hormonal medication. Benign background parenchymal enhancement (BPE) in dynamic contrast-enhanced (DCE) MRI has been established as a sensitive marker of hormonal stimulation of the breast. We investigated the association between LNG-IUD use and BPE in breast MRI to further explore possible systemic effects of LNG-IUDs. METHODS: Our hospital database was searched to identify premenopausal women without personal history of breast cancer, oophorectomy, and hormone replacement or antihormone therapy, who had undergone standardized DCE breast MRI at least twice, once with and without an LNG-IUD in place. To avoid confounding aging-related effects on BPE, half of included women had their first MRI without, the other half with, LNG-IUD in place. Degree of BPE was analyzed according to the ACR categories. Wilcoxon-matched-pairs signed-rank test was used to compare the distribution of ACR categories with vs. without LNG-IUD. RESULTS: Forty-eight women (mean age, 46 years) were included. In 24/48 women (50% [95% CI: 35.9-64.1%]), ACR categories did not change with vs. without LNG-IUDs. In 23/48 women (48% [33.9-62.1%]), the ACR category was higher with vs. without LNG-IUDs; in 1/48 (2% [0-6%]), the ACR category was lower with vs. without LNG-IUDs. The change of ACR category depending on the presence or absence of an LNG-IUD proved highly significant (p < 0.001). CONCLUSION: The use of an LNG-IUD can be associated with increased BPE in breast MRI, providing further evidence that LNG-IUDs do have systemic effects. KEY POINTS: • The use of levonorgestrel-releasing intrauterine contraceptive devices is associated with increased background parenchymal enhancement in breast MRI. • This suggests that hormonal effects of these devices are not only confined to the uterine cavity, but may be systemic. • Potential systemic effects of levonorgestrel-releasing intrauterine contraceptive devices should therefore be considered.