Improved prostate diffusion imaging using deep learning denoising and phase correction with ultra-high-density coil...

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Imaging
• Radiology

Background:

• Diffusion-weighted imaging (DWI) is crucial for identifying clinically significant prostate cancer.
• High b-value DWI is limited by low signal-to-noise ratio (SNR), high noise, and susceptibility artifacts.

Purpose of the Study:

• To assess the feasibility of enhancing DWI quality using a novel 50-channel pelvic coil.
• To evaluate the combined use of deep learning (DL)-based phase correction and DL-denoising algorithms for DWI improvement.

Main Methods:

• Prospective study of 24 men undergoing prostate MRI.
• DWI acquisition using a prototype 50-channel coil and standard phased array coil.
• Reconstruction with vendor's DL denoising (ARDL) and an investigational DL Phase Correction + ARDL (DLPC+ARDL) algorithm.
• Qualitative scoring by two readers and quantitative SNR/CNR measurements.

Main Results:

• The 50-channel coil with DLPC+ARDL achieved significantly higher SNR (99.70 ± 28.50) and CNR (91.68 ± 44.39) compared to other methods (P < .0001).
• DLPC+ARDL improved prostate border definition, zone distinction, lesion conspicuity, and confidence in detecting extraprostatic extension.
• Inter-reader agreement was fair for acquired DWI (κ=0.34) and lower for synthetic DWI (κ=0.19).

Conclusions:

• The 50-channel pelvic coil combined with DLPC + ARDL reconstruction yields superior DWI image quality.
• This approach significantly outperforms ARDL alone and standard 30-channel coil imaging.
• Improved DWI quality has implications for more accurate prostate cancer detection and characterization.