Model-based image reconstruction with wavelet sparsity regularization for through-plane resolution restoration in T₂...

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Image Reconstruction
• Signal Processing

Background:

• High through-plane resolution (1 mm) T2-weighted spin-echo (T2SE) MRI of the prostate is crucial for diagnosis.
• Traditional linear reconstruction methods like Tikhonov (TK) regularization may not optimally preserve signal-to-noise ratio (SNR) while maintaining resolution.

Purpose of the Study:

• To develop and evaluate a model-based image reconstruction method using wavelet sparsity (WS) regularization.
• To compare the performance of WS regularization against TK regularization for prostate T2SE MRI.

Main Methods:

• A WS-regularized reconstruction model was developed using T2SE multislice scan data.
• Reconstructions were tested in phantoms and calibrated in subjects before evaluation in 16 prostate MRI exams.
• WS reconstructions were compared to TK reconstructions, assessing SNR, contrast, and sharpness.

Main Results:

• WS reconstruction demonstrated statistically superior SNR, contrast, and sharpness compared to TK.
• The prostate-to-muscle signal ratio, an indicator of noise reduction, increased in all WS-reconstructed studies.
• WS reconstruction maintained the high 1 mm through-plane resolution.

Conclusions:

• Model-based WS-regularized reconstruction consistently improves SNR in high-resolution prostate T2SE MRI.
• WS regularization offers an advantage over linear TK reconstruction for image quality in prostate MRI.
• This method enhances diagnostic quality by improving key image metrics.