MORSE-PI -- Flexible and artefact-free image reconstruction for structural and functional QSM and other phase-critical imaging applications
summary
The gist
Phase imaging applications such as QSM are highly sensitive to noise amplifications, phase singularities, and other artefacts…
In short
MORSE-PI is a new image reconstruction method designed to create high-quality phase images for structural and functional brain imaging, like QSM. It tackles noise amplification and artifacts common in challenging scenarios such as ultra-high field MRI (7T) or single-echo scans. By using a Virtual Reference Coil (VRC) based on noise covariance, MORSE-PI produces artifact-free, singularity-free results with high signal-to-noise ratio.
Key concepts
- Virtual Reference Coil (VRC)
- The VRC is a mathematical construct created by combining coil sensitivity maps. Crucially, its elements are correlated using the noise covariance matrix to ensure robust signal support across the entire brain. This VRC helps correct phase offsets in coil sensitivity estimates, leading to cleaner, high-SNR phase images.
- Noise Covariance Matrix
- This matrix quantifies how noise varies across different MRI coils and locations. By using this information, MORSE-PI enhances the spatial correlation between neighboring coil sensitivities before calculating the VRC. This enhancement ensures the VRC has consistent signal support throughout the field of view.
- Phase Singularities
- Phase singularities are specific artifacts that appear in phase images, often caused by noise or reconstruction errors. MORSE-PI is specifically designed to eliminate these singularities, resulting in singularity-free phase maps. This is vital for accurate quantitative measurements like QSM.
- Structural and Functional QSM (SWI/fQSM)
- Quantitative Susceptibility Mapping (QSM) measures magnetic susceptibility in the brain tissue. MORSE-PI is applied to both structural (SWI, QSM) and functional (fQSM) phase-based brain imaging. The method provides the high-quality phase images necessary for these techniques to accurately map tissue properties.
Terminology used across episodes
This episode discusses
- MORSE-PI -- Flexible and artefact-free image reconstruction for structural and functional QSM and other phase-critical imaging applications · Paper Radio
- MORSE: Multiple Orthogonal Reference Sensitivity Encoding · Paper Radio
The paper
MORSE-PI -- Flexible and artefact-free image reconstruction for structural and functional QSM and other phase-critical imaging applications · Read on arXiv
Barbara Dymerska, Oliver Josephs, Benjamin James, Vahid Malekian, Nadine N. Graedel, Martina F. Callaghan
University College London
Phase imaging applications such as QSM are highly sensitive to noise amplifications, phase singularities, and other artefacts, particularly in challenging scenarios such as ultra-high field (7T), under-sampled or single-echo acquisitions. We present a novel image reconstruction method, MORSE-PI, designed to produce high-SNR, artefact-free, and singularity-free phase images for both structural and functional phase-based brain imaging. MORSE-PI extends our previous approach, MORSE, by introducing a Virtual Reference Coil (VRC). The VRC is constructed as a linear combination of coil sensitivity maps, with correlations enhanced between coil elements using the noise covariance matrix. Such a VRC ensures robust signal support across the entire brain and is used to correct phase offsets in the MORSE-derived coil sensitivity estimates, resulting in artefact-free, high SNR phase. Compared to GRAPPA with ASPIRE phase correction, MORSE-PI demonstrates greater robustness to artefacts such as noise amplification and aliasing, and shows improved reproducibility in structural imaging at both 3T and 7T. Unlike ESPIRiT and GRAPPA combined with adaptive coil combination methods, MORSE-PI yields singularity-free phase maps. MORSE-PI enables high-SNR reconstructions even for the most challenging scenarios, such as single-echo EPI at 7T. Its efficient, containerised implementation using the Gadgetron framework supports deployment on the MRI scanner console during measurements. MORSE-PI offers a flexible and computationally efficient solution for generating high-SNR, artefact- and singularity-free phase images in both single- and multi-echo GRE and EPI acquisitions. This makes it particularly well-suited for structural and functional QSM, as well as other phase-based MRI applications. Its robustness and rapid computational time facilitate efficient deployment on scanners across field strengths.
DOI: 10.1007/s10334-026-01434-8
Transcript
Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.
Kai: We're looking at the paper: "MORSE-PI -- Flexible and artefact-free image reconstruction for structural and functional QSM and other phase-critical imaging applications".
Mira: Here's the gist of the paper. Phase imaging applications such as QSM are highly sensitive to noise amplifications, phase singularities, and other artefacts, particularly in challenging scenarios such as ultra-high field (7T), under-sampled or single-echo acquisitions.
Kai: We present a novel image reconstruction method, MORSE-PI, designed to produce high-SNR, artefact-free, and singularity-free phase images for both structural (SWI, QSM) and functional (fQSM) phasebased brain imaging.
Mira: Purpose of the Method MORSE-PI is introduced to address the sensitivity of phase imaging applications like QSM to noise amplifications, phase singularities, and other artefacts in challenging scenarios.
Kai: The method aims to produce high-SNR, artefact-free, and singularity-free phase images for both structural (SWI, QSM) and functional (fQSM) phasebased brain imaging.
Paper summary: Kai: So we're looking at this paper called "MORSE-PI -- Flexible and artefact-free image reconstruction for structural and functional QSM and other phase-critical imaging applications".
Mira: This paper is addressing how sensitive phase imaging techniques are to noise amplification, phase singularities, and other artifacts, especially in tough spots like ultra-high field seven tesla or when you only get a single echo.
Lev: So the main goal of this work is to create a way to get high signal-to-noise ratio images that are free from those kinds of problems for both structural structural imaging like SWI and QSM, and functional ones called fQSM.
Kai: Exactly, they introduce this new method called MORSE-PI to tackle those sensitivity issues head on.
Mira: The core idea is that it produces phase images that are high signal-to-noise ratio, free of artifacts, and without those pesky singularities.
Lev: They claim this works for both structural and functional phase based brain imaging, which is a big deal because those applications really need clean phase information to be useful.
Kai: The method extends a previous approach called MORSE by adding something they call a Virtual Reference Coil or VRC.
Mira: This VRC isn't just some random thing; it's built as a linear combination of the coil sensitivity maps, and they enhance the correlations between those coil elements using the noise covariance matrix.
Lev: They say this Virtual Reference Coil ensures robust signal support across the entire brain and is used to correct phase offsets in those MORSE derived coil sensitivity estimates, which results in artefact-free, high signal-to-noise ratio phase.
Kai: What's interesting is they build this VRC from smooth MORSE derived coil sensitivities instead of using individual coil images for construction.
Mira: This smoothing step is what helps them enhance the intercoil correlation between the coils, which strengthens that Virtual Reference Coil idea.
Lev: So, if you were trying to run this on real hardware, you’d need those smooth sensitivities calculated first before you can build that VRC structure.
Paper summary: Kai: Compared to other methods they looked at, like GRAPPA with ASPIRE phase correction or ESPIRiT and GRAPPA combined methods, MORSE-PI shows better robustness against things like noise amplification and aliasing.
Mira: They also see improved reproducibility in structural imaging when they test it at both three tesla and seven tesla.
Lev: That improved reproducibility is important because when you're trying to build real hardware, you need results that are consistent across different field strengths, not just one perfect setup.
Kai: They specifically point out that unlike ESPIRiT or GRAPPA combined with adaptive coil combination methods, MORSE-PI yields singularity free phase maps.
Mira: That means they are tackling a specific problem in phase imaging where you can get those weird sharp spikes in the data, and MORSE-PI seems to handle that better than what those other combinations do.
Lev: From an engineering standpoint, getting singularity free phase maps is a big win because it simplifies downstream processing; you don't have to spend so much time trying to remove these specific artifacts before you can even start the main reconstruction.
Kai: So, this method is designed to make the whole pipeline smoother from a computational perspective.
Mira: And they show high signal-to-noise ratio reconstructions even in really tough scenarios, like single echo EPI acquisitions at seven tesla.
Lev: That single echo scenario is particularly hard because you lose a lot of information, so achieving high SNR there is quite impressive.
Kai: It seems like the paper shows that MORSE-PI provides high signal-to-noise ratio, fold over free and singularity free phase images across various acquisitions.
Mira: This flexibility means it works for single echo and multi echo structural gradient echo scans as well as functional EPI scans.
Lev: And the method doesn't require you to add an extra reference body coil or multi echo reference scans for this to work, which is a nice thing.
Kai: The method also works with a wide range of B zero map estimation techniques and background field removal methods, and even dipole inversion algorithms and denoising pipelines.
Paper summary: Mira: It's quite versatile when you put it into a larger framework of image processing tools.
Lev: That versatility means it could potentially be integrated into many existing MRI analysis software packages without needing a complete overhaul of the reconstruction pipeline.
Kai: So, the MORSE-PI paper is about developing this flexible reconstruction method that delivers high quality phase images for various acquisitions at different field strengths.
Mira: It’s specifically targeting those hard cases where noise and artifacts usually kill the image quality in QSM or fQSM.
Lev: The main implication for real hardware is that if you can implement this VRC construction smoothly, you get more reliable results from your expensive high-field scanners.
Kai: So it boils down to using a Virtual Reference Coil built cleverly with noise covariance information to correct phase errors and keep the signal strong across the brain.
Mira: It’s about making the reconstruction process less dependent on having perfect coil data or ideal acquisition settings for those tricky phase sensitive measurements.
Lev: The limitation they mention is that you still need that initial noise calibration scan to get all this going, which is something you have to do anyway in many protocols.
Kai: So, it's a method that offers high quality phase images without needing extra scans, as long as you do the initial calibration scan.
Mira: This means it’s a practical tool for researchers wanting better results from QSM and fQSM studies on high field systems.
Lev: It’s about getting more consistent and artifact free data from those challenging acquisitions.
Kai: So, to wrap up this MORSE-PI paper, it's about using a Virtual Reference Coil enhanced by noise covariance to produce singularity free phase images for structural and functional QSM.
Mira: It shows better robustness at both three tesla and seven tesla compared to GRAPPA with ASPIRE phase correction.
Lev: The paper points out that you still need that initial noise calibration scan for this reconstruction method to function correctly.
Kai: That’s the essence of MORSE-PI, focusing on creating a flexible and artefact free way to get phase images for QSM and fQSM.
Conclusion: Kai: So this paper is about MORSE-PI, which stands for something that lets you reconstruct images without those nasty phase artifacts in QSM and fQSM scans.
Mira: Yeah, the authors are Dymerska and others, looking at how to make those phase images really clean across different field strengths.
Lev: They’re aiming for high signal-to-noise ratio results that actually hold up when you take things to seven tesla or when you only get a single echo scan.
Kai: It seems like the title itself says it's about being flexible and artifact free, which is exactly what they claim this method achieves.
Mira: The authors are showing how their approach tackles those noise amplifications and phase singularities that usually ruin structural and functional scans.
Lev: What matters for us is that they demonstrate improved reproducibility in structural imaging at both three tesla and seven tesla, which is a tough comparison to make.
Kai: So what this means for us on the hardware side is that if you’re running an expensive scanner, this method could give you more consistent results without needing extra complex setups.
Mira: It suggests that by using a virtual reference coil built with noise information, we can get high quality phase images without having to rely on perfect coil data or extra reference scans.
Lev: The catch is they still need that initial noise calibration scan, which is something you usually have to do anyway in many protocols.
Kai: And that's the main thing they flag—it’s not a magic fix; it depends on having that calibration measurement first.
Mira: It lays out a solid framework for how we can handle phase-critical imaging more reliably, even in challenging scenarios like single echo EPI scans.
Lev: So it’s less about inventing a whole new physical principle and more about clever mathematical correction using the noise covariance matrix.
Kai: It’s basically an advanced way to correct the phase offsets by building this smart virtual reference coil that understands the noise correlations.
Mira: It really shows how important those underlying assumptions are, because if your initial assumptions about the noise model are off, you lose that high SNR.
Lev: This makes me think about how we need to ensure our error correction models account for those non-ideal noise environments in real hardware testing.
Kai: So it's a solid step forward in making these complex imaging techniques more usable for actual clinical and research settings at higher fields.
More episodes
- 2610.34250-Yb/Sr optical clock comparison at 2.08E-18 uncertainty over a 1350-km fiber link
- 2610.36514-Integer electron-count cuts in variational second-order reduced-density-matrix theory. I. Molecular dissociation with fixed real-space and orbital partitions
- 2611.14082-Two atoms with strong light-induced dipolar interactions
- 2611.14040-Characterization of BNLs 20 and 30 micrometer thick LGADs for the Electron Ion Collider
- 2611.13399-Lift-element-based dynamic network of vortex gust-airfoil interactions
- 2611.13476-Hopfion Superlattices in 3D Twisted Moir'e Light Fields
- 2611.13642-Investigation of the timing performance of double-sided p-type 3D silicon sensors produced by CNM
- 2611.13830-SiPM optical readout of a PICOSEC Micromegas precise timing detector
- 2611.14152-Where are the shortest-path nodes in spatial networks?
- 2611.14219-Sub-Ice: semi-automated mapping of basal channels from ice shelf surface elevation data