Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Selective Phase-Aware Training of nnU-Net for Robust Breast Cancer Segmentation in Multi-Center DCE-MRI

Beyza Zayim, Aissiou Ikram, Boukhiar Naima

TL;DR

Multicenter DCE-MRI breast tumor segmentation faces substantial cross-center variability due to scanners and protocols. The authors propose a selective phase-aware training framework within nnU-Net to mitigate quality- and center-induced domain shifts, systematically evaluating data quality and temporal phases on a 1506-case MAMA-MIA dataset. Key findings show that including lower-quality ISPY data or applying CLAHE can harm performance, while a quality-filtered, multi-phase training on DUKE and NACT data achieves the best validation Dice (~0.72), underscoring the value of data-centric strategies. The work highlights the need for automated quality assessment and domain adaptation to realize robust, generalizable segmentation in heterogeneous clinical settings.

Abstract

Breast cancer remains the most common cancer among women and is a leading cause of female mortality. Dynamic contrast-enhanced MRI (DCE-MRI) is a powerful imaging tool for evaluating breast tumors, yet the field lacks a standardized benchmark for analyzing treatment responses and guiding personalized care. We participated in the MAMA-MIA Challenge's Primary Tumor Segmentation task and this work presents a proposed selective, phase-aware training framework for the nnU-Net architecture, emphasizing quality-focused data selection to strengthen model robustness and generalization. We employed the No New Net (nnU-Net) framework with a selective training strategy that systematically analyzed the impact of image quality and center-specific variability on segmentation performance. Controlled experiments on the DUKE, NACT, ISPY1, and ISPY2 datasets revealed that including ISPY scans with motion artifacts and reduced contrast impaired segmentation performance, even with advanced preprocessing, such as contrast-limited adaptive histogram equalization (CLAHE). In contrast, training on DUKE and NACT data, which exhibited clearer contrast and fewer motion artifacts despite varying resolutions, with early phase images (0000-0002) provided more stable training conditions. Our results demonstrate the importance of phase-sensitive and quality-aware training strategies in achieving reliable segmentation performance in heterogeneous clinical datasets, highlighting the limitations of the expansion of naive datasets and motivating the need for future automation of quality-based data selection strategies.

Selective Phase-Aware Training of nnU-Net for Robust Breast Cancer Segmentation in Multi-Center DCE-MRI

TL;DR

Multicenter DCE-MRI breast tumor segmentation faces substantial cross-center variability due to scanners and protocols. The authors propose a selective phase-aware training framework within nnU-Net to mitigate quality- and center-induced domain shifts, systematically evaluating data quality and temporal phases on a 1506-case MAMA-MIA dataset. Key findings show that including lower-quality ISPY data or applying CLAHE can harm performance, while a quality-filtered, multi-phase training on DUKE and NACT data achieves the best validation Dice (~0.72), underscoring the value of data-centric strategies. The work highlights the need for automated quality assessment and domain adaptation to realize robust, generalizable segmentation in heterogeneous clinical settings.

Abstract

Breast cancer remains the most common cancer among women and is a leading cause of female mortality. Dynamic contrast-enhanced MRI (DCE-MRI) is a powerful imaging tool for evaluating breast tumors, yet the field lacks a standardized benchmark for analyzing treatment responses and guiding personalized care. We participated in the MAMA-MIA Challenge's Primary Tumor Segmentation task and this work presents a proposed selective, phase-aware training framework for the nnU-Net architecture, emphasizing quality-focused data selection to strengthen model robustness and generalization. We employed the No New Net (nnU-Net) framework with a selective training strategy that systematically analyzed the impact of image quality and center-specific variability on segmentation performance. Controlled experiments on the DUKE, NACT, ISPY1, and ISPY2 datasets revealed that including ISPY scans with motion artifacts and reduced contrast impaired segmentation performance, even with advanced preprocessing, such as contrast-limited adaptive histogram equalization (CLAHE). In contrast, training on DUKE and NACT data, which exhibited clearer contrast and fewer motion artifacts despite varying resolutions, with early phase images (0000-0002) provided more stable training conditions. Our results demonstrate the importance of phase-sensitive and quality-aware training strategies in achieving reliable segmentation performance in heterogeneous clinical datasets, highlighting the limitations of the expansion of naive datasets and motivating the need for future automation of quality-based data selection strategies.
Paper Structure (13 sections, 3 equations, 3 figures, 1 table)

This paper contains 13 sections, 3 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related work
  3. Methods
  4. Dataset
  5. Preprocessing
  6. Model Architecture and Training Strategy
  7. Evaluation Metrics
  8. Postprocessing
  9. Simulation results
  10. Discussions
  11. Conclusion
  12. Acknowledgments
  13. Disclosure of Interests.

Figures (3)

  • Figure 1: Segmentation pipeline overview.
  • Figure 2: DCE-MRI images before (top) and after CLAHE preprocessing (bottom). Red circles show CLAHE-induced artifacts across different centers.
  • Figure 3: Dice scores across experimental configurations. Left: Performance on all the datasets. Right: ISPY2-specific results showing consistent performance patterns.