Introduction

Breast MRI is an established clinical imaging modality for high-risk screening, diagnosis, pre-operative staging and neoadjuvant therapy response evaluation. The most common clinical indication was diagnostic evaluation (40.3%), followed by screening (31.7%) [1]. Passage of the breast density notification law has had a major impact on MRI utilization, showing increases from 8.5% to 21.1% in non-high-risk women after the law in California went into effect [2]. The increasing popularity of breast MRI has led to the fast accumulation of large breast MRI database. This offers a great opportunity to address unanswered clinical questions regarding the use of breast density, e.g. whether the volumetric density can be incorporated into risk models to improve the prediction accuracy [3], or be used as a surrogate biomarker to predict hormonal treatment efficacy [4,5]. Since MRI is a three-dimensional (3D) imaging modality with distinctive tissue contrast, it can be used to measure the fibroglandular tissue (FGT) volume. However, because many imaging slices are acquired in one MRI, an efficient, objective, and reliable segmentation method is needed. In a previous study we have developed an automatic segmentation method using the Fully-Convolutional Residual Neural Network (FC-RNN), commonly noted as U-net, with non-fat-sat T1-weighted MRI of 286 patients [6]. In this study we tested the segmentation performance of this developed method in 56 breasts of 28 healthy volunteers, each woman acquired using 4 different scanners.

Methods

The initial dataset used for training included 286 patients with unilateral invasive breast cancer (median age, 49 years; range, 30–80 years), and only the contralateral normal breast was analyzed. MRI was performed on a 3T Siemens Trio-Tim scanner, and the pre-contrast T1-weighted images without fat suppression were used for segmentation. The ground truth was generated using a template-based segmentation method [7]. Deep learning segmentation was performed using the U-net, which is a fully connected convolutional residual network [8], and consists of convolution and max-pooling layers at the descending part (the left component of U), and convolution and up-sampling layers at ascending part (the right component of U). The final segmentation model was developed using 10-fold cross-validation. The independent validation dataset included 28 healthy volunteers (age 20–64, mean 35 years old). Each subject was scanned using four different MR scanners in two institutions, including GE Signa-HDx 1.5T, GE Signa-HDx 3T (GE Healthcare, Milwaukee, WI), Philips Achieva 3.0T TX (Philips Medical Systems, Eindhoven, Netherlands) and Siemens Symphony 1.5T TIM (Siemens, Erlangen, Germany). Non-contrast T1-weighted images without fat suppression were used for segmentation. Since both left and right breasts were normal, they were analyzed separately, so there was a total of 56 breasts. The validation was done using the 56 breasts acquired by each scanner first, and then using all 224 breasts acquired by all 4 scanners together. The ground truth was generated using the template-based method for comparison. The segmentation performance was evaluated using the Dice Similarity Coefficient (DSC) and the overall accuracy based on all pixels. In addition, the Pearson’s correlation was applied to evaluate the correlation between the U-net prediction output and the ground truth.

Results

The DSC and accuracy for each scanner was calculated separately, and then combined for all 4 scanners together. The results are shown in Table 1. Figures 1 and 2 illustrate the segmentation results of two women with different breast morphology. The correlation between the U-net prediction output and ground truth for breast volume is shown in Figure 3. The obtained results for four different scanners were similar. The correlation coefficient r was high, in the range of 0.96-0.98. In each figure, the fitted line was very close to the unity line, and the slope was close to 1. The segmentation result for FGT volume is shown in Figure 4. The FGT segmentation results for MRI acquired using 4 different scanners were similar. The correlation coefficient r was very high, in the range of 0.97-0.99. However, using the unity line as reference, the U-net segmented FGT volume was lower compared to the ground truth, as in the two case examples demonstrated in Figures 1 and 2.

Discussion

The results suggest that deep learning segmentation using U-net is feasible to perform fully automatic segmentation for the breast and FGT and yield reasonable accuracy compared to the ground truth segmented by using a template-based method verified by a radiologist. Over the last decade, segmentation of the breast and FGT on MRI has been studied using semiautomatic to automatic approaches with some operator inputs. The processing time for these methods varies from minutes to more than half an hour, which is partially due to the need for the post-segmentation manual correction. The testing in this study using independent validation datasets allows us to evaluate whether the developed segmentation method can be applied widely to other MRI datasets acquired using different imaging sequences on different MR scanners. This fully-automatic deep learning-based segmentation method may provide an accurate and efficient means to quantify FGT volume for evaluation of breast density in very large datasets. The results can potentially be used for breast cancer risk assessment, and for evaluating the response in women receiving hormonal therapy or chemoprevention.

Acknowledgements

This study is supported in part by NIH R01 CA127927, R21 CA208938, and a Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2017R1D1A1B03035995).

References

[1] Wernli et al. Patterns of breast magnetic resonance imaging use in community practice. JAMA internal medicine. 2014;174(1):125-132. [2] Ram et al. Impact of the California Breast Density Law on Screening Breast MR Utilization, Provider Ordering Practices, and Patient Demographics. Journal of the American College of Radiology. 2018;15(4):594-600. [3] Kerlikowske et al. Combining quantitative and qualitative breast density measures to assess breast cancer risk. Breast Cancer Research. 2017;19(1):97. [4] Lundberg et al. Association of infertility and fertility treatment with mammographic density in a large screening-based cohort of women: a cross-sectional study. Breast Cancer Research. 2016;18(1):36. [5] Chen et al. Reduction of breast density following tamoxifen treatment evaluated by 3-D MRI: preliminary study. Magnetic resonance imaging. 2011;29(1):91-98. [6] Zhang et al. Automatic Breast and Fibroglandular Tissue Segmentation Using Deep Learning by A Fully-Convolutional Residual Neural Network. Presented at the Joint Annual Meeting ISMRM-ESMRMB, Paris, France, June 16-21, 2018; Program Number: 2420. [7] Lin et al. Template‐based automatic breast segmentation on MRI by excluding the chest region. Medical physics. 2013;40(12). [8] Ronneberger et al. U-net: Convolutional networks for biomedical image segmentation. Paper presented at: International Conference on Medical image computing and computer-assisted intervention 2015.