Abstract:Purpose: Segmentation of the breast region in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is essential for the automatic measurement of breast density and the quantitative analysis of imaging findings. This study aims to compare various deep learning methods to enhance whole breast segmentation and reduce computational costs as well as environmental effect for future research. Methods: We collected fifty-nine DCE-MRI scans from Stavanger University Hospital and, after preprocessing, analyzed fifty-eight scans. The preprocessing steps involved standardizing imaging protocols and resampling slices to ensure consistent volume across all patients. Using our novel approach, we defined new breast boundaries and generated corresponding segmentation masks. We evaluated seven deep learning models for segmentation namely UNet, UNet++, DenseNet, FCNResNet50, FCNResNet101, DeepLabv3ResNet50, and DeepLabv3ResNet101. To ensure robust model validation, we employed 10-fold cross-validation, dividing the dataset into ten subsets, training on nine, and validating on the remaining one, rotating this process to use all subsets for validation. Results: The models demonstrated significant potential across multiple metrics. UNet++ achieved the highest performance in Dice score, while UNet excelled in validation and generalizability. FCNResNet50, notable for its lower carbon footprint and reasonable inference time, emerged as a robust model following UNet++. In boundary detection, both UNet and UNet++ outperformed other models, with DeepLabv3ResNet also delivering competitive results.

What problem does this paper attempt to address?

The paper attempts to address the problem of breast region segmentation in Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) to achieve automatic measurement of breast density and quantitative analysis of imaging findings. Specifically, the study aims to compare various deep learning methods to improve the performance of whole breast segmentation, reduce computational costs, and minimize environmental impact, providing support for future research work. ### Main Objectives: 1. **Improve Segmentation Performance**: Identify the best-performing model for the breast region segmentation task by comparing different deep learning architectures. 2. **Reduce Computational Costs**: Optimize the training and inference time of the models to reduce resource consumption. 3. **Minimize Environmental Impact**: Evaluate the carbon footprint of different models and select those with a lower environmental impact. ### Research Background: - **Breast Cancer Risk Assessment**: Accurate breast region segmentation is crucial for assessing breast cancer risk. - **Medical Imaging Technology**: MRI plays a significant role in high-quality visualization of the whole breast, but interpreting breast MRI images is challenging due to noise and artifacts from surrounding anatomical structures. - **Deep Learning Applications**: In recent years, deep learning technology has made significant progress in medical image analysis, particularly excelling in segmentation tasks. ### Methods: - **Dataset**: Collected 59 DCE-MRI scan data from Stavanger University Hospital, analyzed 58 scans after preprocessing. - **Preprocessing**: Standardized imaging protocols and resampled slices to ensure consistent volume across all patients. - **Model Evaluation**: Evaluated seven deep learning models, including UNet, UNet++, DenseNet, FCNResNet50, FCNResNet101, DeepLabv3ResNet50, and DeepLabv3ResNet101. - **Cross-Validation**: Employed 10-fold cross-validation to ensure the robustness of model validation. ### Results: - **Segmentation Performance**: UNet++ performed best in Dice score, while UNet showed excellent validation and generalization capabilities. - **Boundary Detection**: UNet and UNet++ outperformed other models in boundary detection, with DeepLabv3ResNet also showing competitive performance. - **Computational Efficiency**: FCNResNet50 excelled with a lower carbon footprint and reasonable inference time. - **Environmental Impact**: FCNResNet50 had the lowest carbon footprint, followed by DeepLabv3ResNet50. ### Discussion: - **Model Performance Differences**: UNet and UNet++ excelled in segmentation and boundary detection, but the complexity of UNet++ might lead to longer training times. - **Impact of Network Depth**: Deeper networks (e.g., ResNet101) can learn more complex features but may overfit with smaller datasets. - **Performance of DenseNet**: Although DenseNet's connectivity pattern aids feature reuse, its lack of skip connections resulted in poorer segmentation performance. Overall, this study provides valuable references for breast region segmentation by comparing different deep learning architectures, helping to improve segmentation accuracy, reduce computational costs, and minimize environmental impact.

Comparative Analysis of Deep Learning Architectures for Breast Region Segmentation with a Novel Breast Boundary Proposal

Deep Learning-Based Automatic Segmentation for Size and Volumetric Measurement of Breast Cancer on Magnetic Resonance Imaging.

Edge of discovery: Enhancing breast tumor MRI analysis with boundary-driven deep learning

Assessing Breast Cancer Volume Alterations Post-Neoadjuvant Chemotherapy Through DenseNet-201 Deep Learning Analysis on DCE-MRI

3D Breast Cancer Segmentation in DCE-MRI Using Deep Learning With Weak Annotation

Breast Region Segmentation being Convolutional Neural Network in Dynamic Contrast Enhanced MRI.

Deep learning methods for lesion detection on mammography images: a comparative analysis

A coarse-to-fine and automatic algorithm for breast diagnosis on multi-series MRI images

Joint Dense Residual and Recurrent Attention Network for DCE-MRI Breast Tumor Segmentation

Segmentation of whole breast and fibroglandular tissue using nnU-Net in dynamic contrast enhanced MR images

A publicly available deep learning model and dataset for segmentation of breast, fibroglandular tissue, and vessels in breast MRI

Breast Regions Segmentation Based on U-net++ from DCE-MRI Image Sequences

Radiologist-level Performance by Using Deep Learning for Segmentation of Breast Cancers on MRI Scans

Lesion Segmentation in Dynamic Contrast Enhanced MRI of Breast.

Joint-phase attention network for breast cancer segmentation in DCE-MRI

An improved semantic segmentation for breast lesion from dynamic contrast enhanced MRI images using deep learning

Deep Learning-Based Segmentation of Locally Advanced Breast Cancer on MRI in Relation to Residual Cancer Burden: A Multi-Institutional Cohort Study

Breast Delineation in Full-Field Digital Mammography Using the Segment Anything Model

Deep learning-based breast region extraction of mammographic images combining pre-processing methods and semantic segmentation supported by Deeplab v3+

A Spatio-Temporal Fully Convolutional Network For Breast Lesion Segmentation In Dce-Mri

HRCUNet: Hierarchical Region Contrastive Learning for Segmentation of Breast Tumors in DCE‐MRI