Searcharxiv⌕ Search

arXiv subjects

Brice Fernandez

Publications and source records attributed to Brice Fernandez.

3 recordsLinked to original sources

SNIF: a Self-Navigator Intelligent Filter correcting off-center and spin-history artifacts in quasi-random radial MRI

For self-navigation and self-gating in MRI, self-navigators are extracted from the complex values of the k-space center, called DC-signal, through a temporal frequency analysis. However, this signal often carries artifacts. This paper introduces a Self-Navigator Intelligent Filter (SNIF) that models the physical origin of the k-space center artifacts to properly estimate them and thus more accurately remove them. The artifact affecting the DC-signal $S_{DC}$ generally has two main origins: the jitter of the k-space center (''off-center artifact'') and the cumulative history of the spins (''spin-history artifact''). Both add noise-like artifacts to the signal. The off-center effect can be modeled as a perturbative signal $S_0$ and the spin-history effect as a convolution kernel $h$ so that the measured DC-signal is $S_{DC} = S_m + S_0 \star h$, where $S_m$ is the motion-related component which effectively contains the expected self-navigators. Methods to compute $h$ and $S_0$ were developed, which gives the possibility to extract $S_m$ from $S_{DC}$. They were implemented and evaluated on thoracic UTE datasets acquired on a healthy volunteer. Qualitative visual comparison and quantitative spectral analysis of the DC-signal before and after SNIF demonstrate that the method improves both the accuracy and the precision of the extracted respiratory and cardiac self-navigators. By considering the physical origin of the DC-signal artifacts, SNIF makes both respiratory and cardiac self-navigators more sensitive to motion, which supports finer MRI motion tracking.

eess.IV↗

Zero-shot Bias Correction: Efficient MR Image Inhomogeneity Reduction Without Any Data

In recent years, deep neural networks for image inhomogeneity reduction have shown promising results. However, current methods with (un)supervised solutions require preparing a training dataset, which is expensive and laborious for data collection. In this work, we demonstrate a novel zero-shot deep neural networks, which requires no data for pre-training and dedicated assumption of the bias field. The designed light-weight CNN enables an efficient zero-shot adaptation for bias-corrupted image correction. Our method provides a novel solution to mitigate the biased corrupted image as iterative homogeneity refinement, which therefore ensures the considered issue can be solved easier with stable convergence of zero-shot optimization. Extensive comparison on different datasets show that the proposed method performs better than current data-free N4 methods in both efficiency and accuracy.

eess.IV↗

Prepared MR Elastography

By decoupling motion and spatial encoding, magnitude contrast MR Elastography could be performed for the first time at ultrashort echo times (12 $μ$s). On the basis of a motion-sensitizing magnetization preparation, the available total magnetic moment is sensitized to the motion induced in the tissues so the information can be efficiently carried over by the MR signal magnitude when the selected imaging pulse sequence is applied. The new paradigm allows also for shorter total acquisition times as demonstrated here in a set of homogeneous and heterogeneous phantoms with up to 5-fold acceleration factors. Summary of Main Findings/Short Synopsis Magnitude contrast MR Elastography was developed on the basis of a motionsensitizing magnetization preparation to subsequently make use of any type of imaging sequence, like UTE or ZTE, to mechanically characterize tissues, otherwise inaccessible with standard MRE.

physics.med-ph↗