What type of study is this?

This is a Quantitative Study study.

September 17, 2025

Nex2Nex – An unsupervised deep learning method for denoising MR images

Puntos clave

The unsupervised method significantly improves the SNR of MR images using intrinsic noise learning.
Two radiologists rated the DWI images as having 'good' diagnostic quality following denoising.
The approach eliminates the need for accurate MRI noise modeling for effective denoising.
Training models on low SNR MRI data becomes feasible with this new denoising solution.

Resumen

Motivation: Supervised deep learning (DL) denoising methods for MR images requires appropriate modelling of MRI noise. Propose an approach learning MR intrinsic noise without the requirement of modelling the MRI noise accurately. Goal(s): Propose an unsupervised DL denoising solution for improving SNR of MR images, without requirement of accurate MRI noise modelling. Approach: The proposed unsupervised DL denoising approach utilized the multiple excitations of MR images acquired to learn the intrinsic MRI noise. Results: Proposed method was evaluated on data acquired at structural and DWI MRI data. Two experienced radiologists rated the DWI denoising as 'good' image quality which were of diagnostic image quality. Impact: Proposed method enables training MRI DL denoising models without requirement to accurately model MRI noise. This approach becomes helpful while training denoising methods for low SNR MRI data.

Me gusta

Guardar