Related Experiment Video
Updated: May 16, 2025
06:08
A Cognitive Fusion-guided Prostate Biopsy Using Multiparametric Magnetic Resonance Imaging and Transrectal Ultrasound
Published on: March 21, 2025
108
Accelerating prostate rs-EPI DWI with deep learning: Halving scan time, enhancing image quality, and validating in
Peipei Zhang1, Zhaoyan Feng1, Shu Chen2
1Department of Radiology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, No.1095 Jie Fang Avenue, Hankou, Wuhan 430030, PR China.
Magnetic Resonance Imaging
|May 14, 2025
Summary
Deep learning super-resolution significantly cuts prostate MRI scan times by over 50% while maintaining high image quality. This technology, using models like MSSNet, shows promise for faster, clearer prostate diffusion-weighted imaging in clinics.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiology
Background:
- Prostate diffusion-weighted imaging (DWI) is crucial for diagnosis.
- Reducing scan time in DWI is essential for patient comfort and reducing motion artifacts.
- High-resolution imaging is vital for accurate prostate cancer detection and characterization.
Purpose of the Study:
- To assess the feasibility and effectiveness of deep learning (DL) super-resolution for prostate DWI.
- To reduce scan time in readout-segmented echo-planar imaging (rs-EPI) DWI.
- To maintain or improve image quality and quantitative accuracy (ADC values) during scan time reduction.
Main Methods:
- Retrospective and prospective analysis of prostate rs-EPI DWI data.
- Application of DL super-resolution models, including Multi-Scale Self-Similarity Network (MSSNet).
- Quantitative evaluation using Structural Similarity Index (SSIM), Peak Signal-to-Noise Ratio (PSNR), Normalized Root Mean Squared Error (NRMSE), Apparent Diffusion Coefficient (ADC) values, and Signal-to-Noise Ratio (SNR).
Main Results:
- MSSNet achieved superior image reconstruction with high SSIM (0.9798), PSNR, and NRMSE.
- DL approach reduced rs-EPI DWI scan time by 54.4% with image quality comparable to high-resolution ground truth (HRGT).
- Strong correlation (within 5% difference) between DL-reconstructed and ground truth ADC values; significant SNR enhancement observed across models, with MSSNet excelling.
Conclusions:
- DL-based super-resolution, particularly MSSNet, is effective for prostate rs-EPI DWI.
- This technology successfully reduces scan time while preserving/enhancing image quality and quantitative accuracy.
- DL super-resolution presents a promising advancement for clinical prostate MRI applications.
