Stephane Loubrie¹, Ana Rodriguez-Soto¹, Lauren Fang¹, Christopher Conlin¹, Maren MS Andreassen², Tyler Seibert^1,3,4, Michael Hahn¹, Vandana Dialani^5,6, Catherine J Wei⁷, Zahra Karimi⁵, Joshua Kuperman¹, Anders Dale^1,8, Etta Pisano^5,9, and Rebecca Rakow-Penner^1,4
1Radiology, UCSD, San Diego, CA, United States, ²Department of Circulation and Medical Imaging, Norwegian University of Science and Technology, Trondheim, Norway, ³Radiation medecine, UCSD, San Diego, CA, United States, ⁴Bioengineering, UCSD, San Diego, CA, United States, ⁵Beth Israel Hospital, Boston, MA, United States, ⁶Harvard Medical School, Bosston, MA, United States, ⁷Commonwealth Radiology Associates, Boston, MA, United States, ⁸Neurosciences, UCSD, San Diego, CA, United States, ⁹American college of radiology, Reston, VA, United States

Diffusion weighted imaging (DWI) holds great potential in improving specificity of findings detected on contrast enhanced breast MRI. Restriction spectrum imaging (RSI), an advanced diffusion imaging model, has potential in discriminating between malignant and fibroglandular breast tissue. In this abstract, we evaluate RSI’s performance in differentiating malignant from benign lesions in a prospective study performed on a breast screening population. All lesions were biopsy proven. The breast RSI model allowed discrimination between malignant, high-risk and low-risk benign lesions and healthy fibroglandular tissue.

Carlotta Ianniello^1,2, Ryan Brown^1,2, Linda Moy^1,2,3, Justin Fogarty¹, Freya Schnabel^3,4, Deborah Axelrod^3,4, and Guillaume Madelin^1,2
1Center for Advanced Imaging Innovation and Research (CAI2R), Department of Radiology, New York University Grossman School of Medicine, New York, NY, United States, ²Center for Biomedical Imaging, Department of Radiology, New York University Grossman School of Medicine, New York, NY, United States, ³Perlmutter Cancer Center, New York University Grossman School of Medicine, New York, NY, United States, ⁴Department of Surgery, New York University Grossman School of Medicine, New York, NY, United States

Sodium (²³Na) MRI has been employed to measure the total sodium concentration (TSC) in breast lesions, which is often elevated compared to healthy breast. The knowledge of the ²³Na relaxation times of the breast, which could be subject to variations in lesions, could increase the accuracy of TSC estimation. In this work we measured the ²³Na relaxation times in six fibroadenoma patients and found a 71% increase in T_2,s and a 28% decrease in T₁ in the lesions compared to healthy tissue. Therefore, using healthy breast relaxation times in TSC quantification could result in errors in lesions.

Jacob S Ecanow¹, David B Ecanow², Nondas Leloudas¹, Bradley Hack¹, and Pottumarthi V Prasad^2
1NorthShore University HealthSystem, Evanston, IL, United States, ²Radiology, NorthShore University HealthSystem, Evanston, IL, United States

We present preliminary data from an ongoing study in individuals referred for biopsy of BIRADS 3, 4, and 5 breast lesions based on mammography, ultrasound, and/or screening MRI.  DTI data was acquired using single shot EPI with 30 different directions and 2 mm isotropic resolution.  MRI data was analyzed using BIT-Motion software.  All cancers showed low ADC and λ1 values compared to control regions.  FA, consistent with prior reports showed minimal difference and so did (λ1-λ3).

Christian Kremser¹, Birgit Amort¹, and Martin Daniaux^1
1Dept. of Radiology, Medical University of Innsbruck, Innsbruck, Austria

Patients undergoing breast MRI frequently have metallic biopsy clips placed within or adjacent to the lesion, producing metallic artifacts, which could lead to misinterpretation. The aim of our study was to investigate the extent of the artifact of four different clips commonly used in our hospital on diffusion-weighted images obtained with a readout-segmented sequence (RESOLVE) and their influence on ADC maps. It is found that artifacts vary in shape and size between different clips, DWI images and ADC-maps. For the marker with the smallest artifact the minimum size of detectable lesion seems to be approximately 10mm.

Arun Somasundaram¹, Mingming Wu¹, Tabea Borde¹, Marcus R. Makowski¹, Eva M. Fallenberg¹, Yu Zhao², and Dimitrios C. Karampinos^1,3
1Department of Diagnostic and Interventional Radiology, School of Medicine, Technical University of Munich, Munich, Germany, ²AI Lab, Tencent, Shenzhen, China, ³Munich Institute of Biomedical Engineering, Technical University of Munich, Munich, Germany

Conventional methods for semi-automatic and rule-based breast segmentation on MR images have a tradeoff between accuracy and segmentation speed. To overcome this, several 2D-deep learning approaches have been proposed, which usually focus on using T1-weighted images to train their models. Our aim is to train a 3D-network for breast segmentation, trained on water-fat MR images, which can be used to measure the breast density based on the proton density fat fraction (PDFF). We show that our model segments both fast and accurately, and can visually outperform our ground truth segmentations while requiring only a few seconds to generate labels.

Joshua D Kaggie¹, Gabrielle C Baxter¹, Mary A McLean^1,2, Ramona Woitek¹, Ferdia A Gallagher^1,2, Andrew D James^3,4, Aneurin J Kennerley^4,5, Rolf F Schulte⁶, William J Brackenbury^3,5, and Fiona J Gilbert^1
1Radiology, University of Cambridge, Cambridge, United Kingdom, ²Cambridge Institute, Cancer Research UK, Cambridge, United Kingdom, ³Biology, University of York, York, United Kingdom, ⁴York Biomedical Research Institute, University of York, York, United Kingdom, ⁵Chemistry, University of York, York, United Kingdom, ⁶GE Healthcare, Munich, Germany

Sodium (²³Na)-MRI was performed using a dual-tuned bilateral ²³Na/¹H breast coil on seven patients with ductal carcinoma, one patient with lobular carcinoma, and ten normal volunteers. Several patients had [²³Na] measurements that were 2-3 times higher than the mean normal fibroglandular [²³Na] measurements. There were no strong correlations between ²³Na-MRI and either DCE or DWI measurements in the patients. These measurements suggest that [²³Na] may provide an additional physiological imaging mechanism of breast cancer.

Alistair Lamb¹, Alan Bainbridge², Tom Parry³, Harriet Rogers⁴, Stuart A Taylor³, Hui Zhang⁵, and Anna Barnes^6
1Department of Medical Physics & Biomedical Engineering, University College London, London, United Kingdom, ²Department of Medical Physics and Biomedical Engineering, University College London Hospitals NHS Foundation Trust, London, United Kingdom, ³Centre for Medical Imaging, University College London, London, United Kingdom, ⁴Institute of Nuclear Medicine, University College London Hospitals NHS Foundation Trust, London, United Kingdom, ⁵Centre for Medical Image Computing, University College London, London, United Kingdom, ⁶King's Technology Evaluation Centre, King’s College London, London, United Kingdom

We investigate the reliability of Whole-Body Imaging ADC estimates from subjects tested and retested within- and between-scanners from different vendors with minimal differences in acquisition protocol and post-acquisition analysis. We show substantial within-subject variation in extracranial ADC estimates within- and between-scanners as measured by Limits of Agreement. We additionally show between-scan variability between scanners is dominated by between-scan variability within a scanner. Furthermore, averaging across subsequent within-scanner examinations does not substantially improve reliability of ADC estimates. We therefore conclude a post-acquisition method for reducing within-scanner variation is required to improve the reliability of ADC estimates.