Deep learning-denoised 3D-EPI achieves submillimeter whole-brain T2* in under 2 minutes

For multiple sclerosis imaging, 3D-EPI with CAIPIRINHA and deep learning denoising provided 0.65 mm whole-brain T2* in 1 min 56 s; denoising increased PSNR by 3 dB, SSIM by 13%, and lesion-to-vein contrast by 9%.

• Submillimeter (0.65 mm isotropic) whole-brain T2*-weighted imaging was achieved with scan times from 3 min 22 s (R=2) to 1 min 56 s (R=4).
• Deep learning denoising (DnCNN) significantly improved quantitative image quality metrics and lesion-to-vein contrast, as assessed by a blinded neuroradiologist.
• This sequence may help meet updated MS diagnostic criteria requiring detection of the central vein sign.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.