Beyond MSE: Rician Likelihood Denoising for Self-Supervised Cardiac $T2$ and $T1\rho$ MRI

2026-09-24

AI Summary

This study addresses the bias introduced by conventional MSE loss in self-supervised cardiac MRI denoising, which neglects Rician noise characteristics and leads to distortions in both images and parametric maps, compounded by the clinical absence of low-noise ground truth data. To overcome these limitations, this work proposes a self-supervised denoising framework based on Rician maximum likelihood estimation as an alternative to blind-spot networks. By formulating denoising as maximum likelihood estimation under a known Rician distribution, the method circumvents the restrictive independent and identically distributed noise assumption, enabling unbiased signal recovery. The proposed approach yields unbiased denoised images and high-fidelity T2/T1ρ parametric maps, achieving performance comparable to supervised baselines while substantially reducing scan costs. Ultimately, it establishes a novel paradigm for ground-truth-free medical image restoration.

Abstract

Magnetic resonance imaging involves an inherent trade-off among spatial resolution, acquisition time, and noise. This trade-off contributes to long scan times and high cost. Deep learning has improved image denoising, but cardiac MRI remains difficult because high-resolution, rapid acquisitions generally lack corresponding low-noise ground truth. Self-supervised denoising offers a potential solution by learning from noisy image pairs or even single noisy acquisitions. However, we show that Noise2Void-style blind-spot denoising, which uses a mean squared error (MSE) loss and assumes zero-mean, independent and identically distributed (i.i.d.) noise, is poorly suited to MR magnitude images. When applied to short-axis $T2$-weighted and $T1\rho$-weighted cardiac MRI with synthetic Rician noise, it produces biased denoised images and biased parametric maps of $T2$ and $T1\rho$. To address this limitation, we formulate self-supervised denoising as maximum likelihood estimation under a known Rician noise model. This yields unbiased denoisers that are competitive with supervised baselines.

Problem

• Cardiac MRI
• Self-supervised denoising
• Rician noise
• Mean squared error
• Parametric mapping

Innovation

• Self-supervised denoising
• Rician likelihood
• Cardiac MRI
• Maximum likelihood estimation
• Noise2Void