Related Experiment Video
Updated: Aug 15, 2026
15:48
Tracking the Mammary Architectural Features and Detecting Breast Cancer with Magnetic Resonance Diffusion Tensor Imaging
Published on: December 15, 2014
Deep learning-based reconstruction for multi-shot breast DWI: Quantitative and qualitative improvements
Jin Joo Kim1, Jin You Kim1, Lee Hwangbo1
1Department of Radiology, Biomedical Research Institute, Pusan National University Hospital, Pusan National University School of Medicine, Busan, the Republic of Korea.
European Journal of Radiology
|August 13, 2026
Summary
Deep learning reconstruction significantly enhances breast MRI diffusion-weighted imaging quality and reduces scan time. This advanced technique shows promise for improved breast cancer diagnosis and patient care.
Area of Science:
- Radiology
- Medical Imaging
- Artificial Intelligence in Medicine
Background:
- Multi-shot diffusion-weighted imaging (DWI) is crucial for breast MRI.
- Standard readout-segmented echo-planar imaging (rs-EPI) is time-consuming and prone to artifacts.
- Deep learning (DL) offers potential for faster and higher-quality image reconstruction.
Purpose of the Study:
- To compare the effectiveness of DL-reconstructed (DLR) multi-shot DWI with standard rs-EPI for breast MRI.
- To evaluate improvements in image quality, artifact reduction, and scan time.
- To assess the diagnostic performance of DLR in detecting breast cancer.
Main Methods:
- Retrospective analysis of 146 women with breast cancer undergoing preoperative MRI.
- Comparison of rs-EPI and prototype DLR multi-shot DWI sequences.
- Quantitative analysis of signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and apparent diffusion coefficient (ADC).
- Qualitative assessment of image quality, artifact suppression, and lesion conspicuity by two radiologists.
- Validation in an independent cohort of 39 patients.
Main Results:
- DLR demonstrated significantly higher SNR, CNR, and lesion contrast compared to rs-EPI (p < 0.05).
- DLR achieved a 29% reduction in acquisition time.
- Qualitative scores for image quality and lesion conspicuity were higher with DLR.
- Improved artifact suppression was noted with DLR at b = 800 s/mm².
- Validation cohort confirmed improvements in SNR, CNR, and overall image quality.
Conclusions:
- DL-based reconstruction significantly enhances quantitative and qualitative image quality in multi-shot breast DWI.
- DLR reduces scan time, making breast MRI more efficient.
- These findings support the clinical integration of DL reconstruction for improved breast MRI protocols.