Research Presentation Session: Breast

RPS 1802 - Advances in breast MRI techniques

March 2, 09:30 - 11:00 CET

7 min
Abbreviated breast magnetic resonance imaging (MRI) protocols in a very high-risk population: friend or foe?
Tamara Suaris, London / United Kingdom
Author Block: L. Metaxa, V. Papalouka, N. Vidyaprakash, S. Lee, S. Dani, T. Suaris; London/UK
Purpose: AbbMRI aims to provide a cost-effective study without compromising diagnostic accuracy. We aim to assess the diagnostic performance of AbbMRI in high-risk breast cancer screening.
Methods or Background: Four experienced breast radiologists retrospectively reviewed the AbbMRI protocol in 236 MRI breast studies performed for very high-risk between January 2018 – 2021. Readers were blinded to prior imaging, history, and patient outcomes. The protocol consisted of six sequences: pre-contrast T1W, T2W axial, two dynamic postcontrast, subtracted axial sequences and MIP reconstruction. Results were compared with existing reports of the full protocol (19 sequences).
Results or Findings: From the 236 MRI studies, four invasive malignancies and one B3 lesion (radial scar) were identified. The average time to read the AbbMRI was 69 seconds (range: 28 to 240 seconds; 95% CI: 63, 74 seconds).
The inter-rater rate (IRR) between the abbreviated and full protocol was: 91.5%. The recall rate for AbbMRI was 9.5%, and 6.1% for full protocol. Sensitivity achieved by AbbMRI (100%) was equal to the full protocol (no cancers were missed). NPV were high for both reading sessions (100%). Specificity and PPV of the abbreviated MRI was lower compared to the full protocol (91.2 versus 94.7, and 8.9 versus 13.8%, respectively) (p value .01, a=0.5).
Conclusion: Abbreviated MRI is feasible in screening MRI of high risk patients, with sensitivity/CDR comparable to the full MRI protocol, but with a small increase in the recall rates. Abbreviated breast MRI can be considered as a safe alternative to conventional MR studies in UK very high-risk screening, reducing MR waiting times/backlog.
Limitations: Identified limitations were: (1) this was a retrospective study, with data for the full protocol collected from existing reports, where the radiologists were not blinded to the patients' personal history; (2) the abbreviated MRI was not double read.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: This study was approved by the Director of Research.
7 min
Evaluation of pure DCIS with abbreviated breast MRI
Nurbanu Basdogan, Istanbul / Turkey
Author Block: A. Arslan, N. Basdogan, C. S. Topal, F. Ezberci; Istanbul/TR
Purpose: The most reliable imaging method for detecting breast cancer is magnetic resonance imaging (MRI) of the breast. 'Abbreviated MRI' imaging protocol was developed to shorten scanning and evaluation time to screen large populations. The aim of our study is to determine the lesion detection rate with the abbreviated MRI protocol in patients with a histopathological diagnosis of pure DCIS (ductal carcinoma in situ without any invasive components) in the breast
Methods or Background: Patients who underwent breast MRI between May 2021 and October 2023 in our institution were reviewed retrospectively by two radiologists (a radiologist with nine years of breast MRI experience, and a radiology resident with two years of MR experience). All the MRIs were carried out using a 1.5T MRI machine (GE Optima MR450w). Patients diagnosed with DCIS by biopsy were included in the study. Cases who had a history of breast cancer, had invasive tumour foci in addition to DCIS, and cases who received neoadjuvant chemotherapy, were excluded from the study. Abbreviated MRI protocol includes T1WI without contrast, T1WI with first minute contrast, and one minute subtraction image.
Results or Findings: A total of 36 DCIS diagnosed by biopsy were evaluated in our centre. Eleven cases were excluded from the study because they had invasive tumour foci in addition to DCIS. Except for five of the 25 pure DCIS, the others showed contrast enhancement in the first minute of contrast and subtraction images (sensitivity 80.0%). Of the five cases, one case had low-grade DCIS, two cases had intermediate-type DCIS, and two cases had high-grade DCIS. The mean age of the patients included in the study was 51.72 (39-80) years.
Conclusion: Pure DCIS could be detected with the high sensitivity abbreviated MRI protocol.
Limitations: The limited number of patients and the retrospective nature of the research are limitations of this study.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by Umraniye Reserch and Training Hospital.
7 min
Abbreviated vs full protocol in breast MRI: a meta-analysis comparing their diagnostic performances
Ottavia Battaglia, Milan / Italy
Author Block: O. Battaglia, F. Pesapane, A. Rotili, O. D'Ecclesiis, S. Penco, G. Signorelli, L. Nicosia, S. Gandini, E. Cassano; Milan/IT
Purpose: In the last decade, the interest in magnetic resonance imaging abbreviated protocols (AB-MRI) has grown as shorter scan times translate into cost savings, enhanced patient tolerance and improved accessibility.
The aim of this meta-analysis is to compare the diagnostic performances of AB-MRI and full protocol MRI (FP-MRI) in breast cancer assessment.
Methods or Background: We conducted a systematic review and meta-analysis from 01/11/2019 up to 12/12/2022. A biavariate model was used to calculate the summary estimates of sensitivity and specificity. Random effect models were used to calculate summary AUC and 95% CIs. Probability distributions for negative and positive predictive values were obtained.
Results or Findings: Out of 367 studies, a total of 11 studies were included. The summary sensitivity estimated from the AB-MRI compared to the FP-MRI is slightly lower (86.2%, 95% CI: 82.1-89.6 and 93.7%, 95% CI: 88.7-96.6, respectively). Summary estimates (SEs) of specificity are similar (79.7%, 95% CI: 73.2-84.8 for AB-MRI and 78.0%, 95% CI: 70.6-84.0 for FP-MRI). A better accuracy has emerged for FP-MRI (84.4% for AB-MRI and 87.6% for FP-MRI). The SEs of positive predictive value is similar (22.3% for AB-MRI and 20.4% for FP-MRI).
Conclusion: AB-MRI protocols have emerged as a potential alternative to overcame FP-MRI limits, offering increased efficiency and cost-effectiveness by reducing acquisition and reading times, improving accessibility, and enhancing the patient experience with similar diagnostic performances.
Further research is needed to optimise the AB-MRI and to better define its indications. This includes investigating the role of machine learning algorithms to enhance lesion detection and characterisation within the AB-MRI. Long-term outcome studies are needed to evaluate AB-MRI’s impact on patient outcomes, including detection rates and interval cancers.
Limitations: An identified limitation was the heterogeneity of the proposed AB-MRI protocols in the included studies.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: No information provided by the submitter.
7 min
Enhancing the Kaiser score for unenhanced breast MRI
Nina Pötsch, Vienna / Austria
Author Block: N. Pötsch, P. Clauser, P. Kapetas, T. H. Helbich, P. A. Baltzer; Vienna/AT
Purpose: Unenhanced abbreviated breast MRI protocols aim to reduce scanning time and costs while enhancing access to breast MRI. However, in order to further exploit the advantages of the Kaiser score, a well-established clinical decision rule for breast MRI, it must be adapted to unenhanced protocols.
Methods or Background: We studied 120 patients who underwent breast MRI between 2021 and 2023 for further evaluation of suspicious or equivocal findings in conventional imaging (BI-RADS 0, 3-5 in mammography and ultrasound) or for staging purposes (BI-RADS 6). All patients underwent biopsy for histologic verification or were followed for a minimum of 12 months to confirm lesion stability. MRI scans were conducted using 1.5T or 3T scanners using dedicated breast coils and a protocol in line with international recommendations additionally including DWI and ADC. Lesion characterisation relied solely on T2w and DWI/ADC-derived features (such as lesion type, margins, shape, internal signal, surrounding tissue findings, ADC value). Statistical analysis was done using decision tree analysis with benign (B2, B3) and malignant (B5a/b) as dependent variables.
Results or Findings: We analysed a total of 161 lesions, including 81 non-mass lesions, with a malignancy rate of 40%. The lesion margins (spiculated, irregular, or circumscribed) were identified as the most critical criterion within the decision tree. Further subclassification considered the ADC value as the second most important criterion. The resulting score demonstrated a strong diagnostic performance with an AUC of 0.840, providing clear rule-in and rule-out criteria.
Conclusion: Lesion characterisation in unenhanced MRI is feasible, with lesion margins and ADC value as the most important criteria.
Limitations: An identified limitation was that our analysis did not address lesion detection.
Funding for this study: This study received no funding.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: Informed consent was waived.
7 min
Role of ultra-fast breast MRI in characterisation of breast masses: a step toward a faster MRI technique
Rasha Karam Mahmoud Mohammed, Mansoura / Egypt
Author Block: R. Karam Mahmoud Mohammed, D. M. Bayoumi, F. A. Shokeir; Mansoura/EG
Purpose: To evaluate the efficacy of parameters derived from ultra-fast breast MRI, maximum slope (MS), initial enhancement phase (IE phase) and time to enhancement (TTE), in differentiation of benign and malignant breast lesions.
Methods or Background: This is a prospective study that included 264 patients with 273 breast lesions, from which 120 were pathologically proven to be benign and 153 were pathologically proven to be malignant. An ultra-fast study was performed between the pre-contrast study and the first postcontrast sequence in the dynamic protocol using the TWIST sequence without fat suppression. The TWIST sequence was composed of a pre-contrast phase followed by 15 postcontrast TWIST series with a temporal resolution of 4.32 seconds each. We put a ROI at the hot spot of the lesion on subtracted images and a wash-in curve was automatically generated. From the ultra-fast wash in curve parameters were manually calculated (TTE and MS).
Results or Findings: TTE was significantly lower in malignant lesions compared to benign ones (median=8 seconds [0-20] versus 20 seconds [8-40], respectively. P value<0.001). However, MS was higher in malignant lesions compared to benign ones (median=15 %/s [2-140] versus 5 %/s [1-17] respectively. P value=<0.001).
TTE cut-off value of 15.5 seconds showed a sensitivity of 78.4%, specificity of 80%, AUC of 0.876, and accuracy of 79%. However, MS, a cut off value of 7.25 %/s showed sensitivity of 82.4%, specificity of 70%, AUC of 0.86 and accuracy of 76.9%.
Conclusion: Ultra-fast breast MRI can be used as a faster and accurate MRI technique for differentiation between benign and malignant breast lesions.
Limitations: Identified limitations were: (1) the relatively small number of cases and (2) the fact that the ultra-fast technique was not compared to the conventional dynamic MRI technique.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by our institution’s ethics committee (Mansoura Faculty of Medicine Institutional Research Board, reference number: R.23.05.2174). Written informed consent was obtained from all patients included in the study.
7 min
Application of the Kaiser score in the assessment of suspicious contrast-enhancing lesions on breast MRI in patients with equivocal conventional imaging findings
Katarina Obradovic, Belgrade / Serbia
Author Block: K. Obradovic, I. Petkovic, D. Pavlovic Stankovic, V. Urban, Z. C. Milosevic; Belgrade/RS
Purpose: To assess whether applying the Kaiser score might downgrade breast MRI BI-RADS categories in patients with equivocal conventional imaging findings (mammography and/or US) and avoid unnecessary biopsies.
Methods or Background: Equivocal conventional imaging findings are one of the indications for problem-solving breast MRI. Generally, a combination of high sensitivity (98–100%) and suboptimal specificity (up to 88%) is inherent to breast MRI, which frequently results in false-positive diagnoses.
Our single-centre study enrolled 61 consecutive patients from January 2017 to December 2018, with equivocal conventional imaging findings and 69 breast lesions on MRI (62 lesions in BI-RADS category 4, seven in BI-RADS category 5), undergoing US-CNB and/or surgery. All MRI examinations were re-evaluated after 4 years of the patients' follow-up, using the Kaiser score by two breast imaging radiologists and one radiology resident, individually and blinded to the pathological diagnosis and previous reading results, with the consensus about different opinions through consultation.
Results or Findings: Histopathology revealed 18 malignant and 51 benign lesions, resulting in a predominance of benignancy of 73.9%. The mean size of lesions on MRI was 15±10 mm; 50 lesions presented as masses (72.5%), 19 as a non-mass enhancement (27.5%), and there was no difference between malignant and benign lesions. The Kaiser score was as follows: 1–4 (BI-RADS 2/3) in 48 lesions (69.6%), 5–7 (BI-RADS 4) in 11 (15.9%), and 8–11 (BI-RADS 5) in ten (14.5%) (Figs. sensitivity and specificity were 100% and 94.1%, respectively, with three false-positive findings and no false-negative findings. The AUC was 0.971 (95% CI: 0.932–1.000).
Conclusion: Applying the Kaiser score to breast MRI, BI-RADS categories are reclassified (particularly BI-RADS 4 to BI-RADS 2/3), which can obviate unnecessary biopsies in breast lesions with equivocal conventional imaging findings.
Limitations: No limitations were identified.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Institute of Oncology and Radiology of Serbia: 1140-01
7 min
Breast multiparametric MRI (MP-MRI) to discriminate between pure ductal carcinoma in situ (DCIS) and microinvasive carcinoma (MIC): the importance of DWI
Federica Cicciarelli, Latina / Italy
Author Block: F. Cicciarelli1, V. Rizzo2, F. Galati3, M. Pasculli3, R. Maroncelli4, F. Pediconi3, C. Catalano3; 1Latina/IT, 2Maglie/IT, 3Rome/IT, 4Guidonia/IT
Purpose: DCIS is a common subtype of breast cancer that is often upgraded to invasive cancer on the final pathology report of the surgical specimen. DCIS has an upgrade rate ranging in MIC from less than 10% to more than 40%. In these cases, histologic upgrade should be considered before surgical planning, to include sentinel lymph node biopsy in view of possible axillary lymph node metastasis.
The aim of our study is to investigate whether breast multiparametric magnetic resonance imaging (MP-MRI) can distinguish between pure DCIS and MIC or invasive ductal cancer (IDC).
Methods or Background: Between January 2018 and November 2022, patients with biopsy-proven DCIS who performed preoperative breast MP-MRI on 3T magnet were enrolled in this retrospective study. MP-MRI features (size, morphology, margins, extent of disease, presence of peritumoural oedema and curve I/T value, type and distribution of enhancement, restriction of diffusion and ADC value) were evaluated. For categorical variables chi-square test was assessed to investigate the association with the pathological outcome. Logistic regression analysis was performed to evaluate their prognostic value.
Results or Findings: Out of 129 patients with biopsy-proven DCIS, 36 presented foci of micro-infiltration on surgical specimens and eight was IDC. The presence of micro-infiltration foci was significantly associated with several MP-MRI features, such as tumour size ≥2 cm (p=0.02), clustered ring enhancement (p<0.001), and segmental distribution (p<0.01). Interestingly, a mean ADC value <1.3 x 10-3 mm2/s revealed to be a prognostic factor for the presence of micro-infiltration foci (p< 0.05).
Conclusion: Breast MP-MRI can predict the presence of micro-infiltration foci in biopsy-proven DCIS, and could be considered a valid tool for therapeutic planning in patients with DCIS.
Limitations: The relativity small number of patients was an identified limitation.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: No information provided by the submitter.
7 min
Screening without mammography using abbreviated breast MRI alone
Stephanie Morscheid, Aachen / Germany
Author Block: S. Morscheid, M. Bode, L. C. Huck, K. Strobel, C. K. Kuhl; Aachen/DE
Purpose: Breast MRI is recommended for supplemental screening for women with dense breasts. However, in women undergoing screening MRI, the additional cancer detection rate attributable to mammography has been shown to be limited. Based on recently published cost-effectiveness analyses, using MRI alone for screening would represent the most cost-effective screening method. The purpose of this study was to report first results on the outcome of women undergoing breast MRI as a stand-alone screening test (without mammography).
Methods or Background: This is an ongoing study which has so far enrolled 1,085 women (aged 40-85 years, median 57 years) without personal history of breast cancer who underwent screening using abbreviated MRI alone. Breast MRI was performed on a 1.5T system with multichannel-coil, according to a standardised protocol. Validation of MRI diagnoses was achieved by follow-up (977/1,085) or biopsy (108/1,085). A further 54 biopsies were performed within the screening period.
Results or Findings: Women underwent a total 2,771 screening MRIs (mean 2.6 studies/participant), covering a total of 3,220 women-years and a subsequent follow-up period of 2,425 women-years. MRI was positive in 392/2,771 examinations (14.1%). Breast cancer was confirmed in 67, yielding a PPV3 (of biopsy) of 41.4%. The overall cancer detection rate was 20.8 per 1000 women-years. The median age of women diagnosed with cancer was 59 years. Cancers were DCIS in 18/67 and invasive cancers in 49/67. Of the invasive cancers, cancers were staged pT1a in 7/49, pT1b in 21/49, pT1c in 16/49, pT2 in 5/49, and as N0 in 42/49 (85.7%); N1 and N1mi in the remaining cases; all were staged M0. 60/67 patients with MRI-detected cancer underwent bilateral preoperative mammography. The MRI-known cancer was occult on mammography in 40/60 (67%).
Conclusion: Abbreviated breast MRI alone appears sufficient for breast cancer screening, ensuring early cancer detection.
Limitations: The pseudo-prospective study design was an identified limitation.
Funding for this study: No external funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of RWTH University Aachen.
7 min
Looking beyond standard dosage in dynamic contrast enhanced breast MRI using a pre-trained deep-learning model
Srivathsa Pasumarthi Venkata, Santa Clara / United States
Author Block: S. Pasumarthi Venkata1, P. Gulaka1, G. Zaharchuk2, R. Chamberlain1; 1Menlo Park, CA/US, 2Stanford, CA/US
Purpose: Gadolinium-based contrast agents (GBCAs) are widely used in breast MRI. Increasing GBCA dosage improves lesion visualisation but safety concerns have been raised regarding long-term GBCA retention. In this work, we repurposed a dose reduction deep-learning (DL) model to synthesise contrast boosted (CB) breast MRI images that have better lesion visualisation without increasing dosage.
Methods or Background: A DL model for dose reduction in brain MRI was previously trained to enhance 10% to 100% dose. This model was repurposed for boosting contrast in breast MRI by inputting pre-contrast and standard postcontrast (SC) sequences to synthesise contrast boosted (CB) sequences. Dynamic contrast enhanced (DCE) breast MRI data from 60 patients (30 internal, 30 Duke Breast MRI public dataset) were used to evaluate CB algorithm. CE images were normalised and co-registered to pre-contrast sequences, and the first time point post-injection was used as the postcontrast. Axial images were manually cropped to remove signals posterior to the pectoral muscles. Left and right sides were processed individually by the model. Lesion-to-background ratio (LBR) and contrast enhancement percentage (CEP) metrics were calculated on SC and CB images. ROIs were manually drawn for calculating quantitative metrics. Enhancement kinetic curves (time post-injection versus signal-intensity [SI]) were drawn with pre-contrast, SC and other CE time points along with the average SI of CB images.
Results or Findings: The mean LBR and CEP of CB (9.63±0.53 and 182.3%) were significantly better (p<0.001) than those of SC images (6.73 ± 0.34 and 93.4%). It was also observed that the DL model improves the lesion enhancement and visibility.
Conclusion: Quantitative and qualitative assessment has shown the feasibility of DL based contrast boosting in breast MRI to improve lesion visualisation.
Limitations: Detailed clinical evaluation required.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: No information provided by the submitter.

This session will not be streamed, nor will it be available on-demand!