Skip to main content

MRI Quantitative Susceptibility Mapping (QSM)

By Troy Zhou, PhD, DABR, DABSNM
• 16 min read

Introduction

Quantitative susceptibility mapping (QSM) is an MRI post-processing technique that reconstructs the magnetic susceptibility of tissue at each voxel from gradient-echo phase data, turning a traditionally discarded signal into a quantitative biomarker. Where conventional MRI reads the magnitude of the signal, QSM mines its phase, which encodes the local magnetic field shifts produced by iron, calcium, blood products, and myelin, and inverts those field shifts back into the susceptibility values that caused them. 1, 2

The appeal is that susceptibility is a genuine physical tissue property with a sign: paramagnetic materials such as non-heme iron and deoxygenated blood increase the local field, while diamagnetic materials such as calcium and myelin decrease it. 3, 6 That sign lets QSM do something routine sequences cannot — reliably separate a calcification from a microbleed, or measure the iron burden of the deep gray nuclei as a number rather than a shade of gray. 3, 4 Because of this, QSM has become a leading advanced-imaging method for brain iron quantification, venous and oxygenation imaging, and the study of neurodegeneration. 4, 5

QSM is also unforgiving of sloppy physics. It is the output of a chain of reconstruction steps applied to phase data from a carefully specified acquisition, and the answer it produces is only as good as the field homogeneity, phase stability, and pipeline discipline behind it. 1, 6 This guide explains the physics of QSM, the inverse problem at its core, how it differs from susceptibility-weighted imaging (SWI), its clinical uses, and the acquisition, reconstruction, and quality-control practices that make its numbers defensible. DRPS supports MRI programs through its MRI physics testing and accreditation support services across Florida, Maryland, Virginia, Washington DC, California, Nevada, Pennsylvania, New York, New Jersey, and Delaware.

Topic Explanation

Susceptibility, phase, and the field it creates

Magnetic susceptibility, , describes how much a material magnetizes in an applied field. In a scanner's main field , a tissue with susceptibility develops a magnetization that perturbs the local field by a small amount, typically a few parts per million. Protons precess at the Larmor frequency, which is proportional to the local field, so a field offset accrues as a measurable phase over the echo time of a gradient-echo sequence:

where is the gyromagnetic ratio, is the local field offset, and TE is the echo time. 2, 6 The measured phase, therefore, is a window onto the field map — and the field map is what a susceptibility distribution produces. QSM's job is to run that relationship backward: from measured phase to field map, and from field map to susceptibility.

Why SWI is not enough

Susceptibility-weighted imaging exploits the same phase information, but only qualitatively: it high-pass-filters the phase and multiplies it into the magnitude image to darken veins, microbleeds, and iron-rich structures. 2 The result is visually striking and clinically useful, but it has real limitations. SWI contrast is nonlocal (blooming extends beyond the true source), orientation-dependent, echo-time-dependent, and processing-dependent, and it cannot tell a paramagnetic source from a diamagnetic one by inspection. 1, 2 QSM removes the blooming and nonlocality by solving for the actual source distribution, and because it reports a signed quantitative value, it resolves the calcium-versus-blood ambiguity that defeats SWI. 3

The pipeline, not a single reconstruction

QSM is best understood as a multi-step pipeline rather than one reconstruction. The standard steps are:

  1. Phase unwrapping — recovering continuous phase from the wrapped values the scanner stores.
  2. Background-field removal — stripping the large field contributions from sources outside the region of interest (air-tissue interfaces, the sinuses, the skull) to isolate the local tissue field.
  3. Dipole-field inversion — solving the ill-posed inverse problem that converts the local field map into a susceptibility map. 1, 6

Each step has multiple published algorithms, and the choices propagate into the final numbers, which is why standardization matters so much. 1

Key Technical Principles

The dipole model and the ill-posed inversion

The central physics of QSM is that the field perturbation from a susceptibility distribution is the convolution of that distribution with a magnetic dipole kernel. In Fourier space, convolution becomes multiplication, and the relationship between the normalized field shift and susceptibility takes the compact form:

where is the dipole kernel and is the component of the spatial frequency along . 1, 6 Recovering susceptibility means dividing by — but wherever . Solving for that angle:

This is the magic angle: a double cone in k-space where the kernel vanishes and the inversion is undefined. Naive division amplifies noise and produces the characteristic streaking artifact of QSM, so the inverse problem must be regularized or otherwise constrained. 1, 6

How the dipole inversion is solved

Several families of methods address the ill-posed inversion, trading accuracy, artifacts, and acquisition burden.

Dipole-inversion approach How it works Strengths Limitations
Thresholded k-space division (TKD) Caps or thresholds the near-zero kernel values before dividing Fast, simple Residual streaking; susceptibility underestimation
Iterative / regularized (e.g., morphology-enabled dipole inversion, MEDI) Adds spatial priors from the magnitude image to constrain the solution Reduced streaking; good anatomy Requires parameter tuning; iterative cost
Multi-orientation (COSMOS) Combines acquisitions at several head orientations to fully sample the kernel Reference-quality, low artifact Impractical clinically (multiple scans, repositioning)
Deep-learning inversion Learns the inversion from training data Fast at inference; strong artifact suppression Needs validated training; generalization and QC concerns

Single-orientation, regularized methods dominate clinical research use, while multi-orientation COSMOS serves mainly as a gold-standard reference. 1, 6 The consensus recommendation for clinical brain research favors standardized, well-characterized single-orientation pipelines so that results are reproducible across sites. 1

Units, reference regions, and reproducibility

Susceptibility is reported relative to a reference, because MRI measures field differences, not absolute susceptibility. Values are expressed in ppm or ppb and are typically referenced to a region such as cerebrospinal fluid or whole-brain mean. 1, 3 The practical consequence is that a QSM value is only interpretable alongside its reference region and pipeline, and cross-site or longitudinal comparisons require the acquisition, reconstruction, and reference to be held constant. QSM reconstruction challenges and consensus efforts exist precisely to reduce this variability. 1

Acquisition requirements

QSM is usually built on a 3D multi-echo gradient-echo acquisition with flow compensation, a monopolar readout, full coverage of the region of interest, and preservation of the phase images. 1 Echo times must be chosen so phase evolves adequately without excessive wrapping or signal loss, and field strength matters because susceptibility-induced phase scales with . Consensus guidance specifies these acquisition parameters so that the downstream pipeline can succeed. 1 The dependence on phase stability and field homogeneity is why QSM is so sensitive to the same fundamentals a facility already tests: B0 homogeneity, shimming, and geometric fidelity.

Clinical Impact

Brain iron and neurodegeneration

The most mature application of QSM is brain iron quantification. Non-heme iron accumulates with age in the deep gray nuclei — the globus pallidus, putamen, caudate, substantia nigra, and red nucleus — and QSM measures that paramagnetic burden quantitatively. 3, 4 Altered iron patterns have been studied in Parkinson disease, Alzheimer disease, and multiple sclerosis, where QSM also probes demyelination and chronic active lesions. 4, 5 QSM has even been applied to developmental and pediatric neuroscience: studies have reported lower deep-gray-nuclei iron content in children with autism spectrum disorder and with attention-deficit/hyperactivity disorder, illustrating how a quantitative susceptibility biomarker can be used to compare groups. 9, 10

Calcium versus hemorrhage

Because paramagnetic and diamagnetic sources carry opposite signs, QSM resolves one of the classic ambiguities of routine MRI. A hyperdense or hypointense focus that could be either calcification or a small bleed shows a positive susceptibility if it is a blood product and a negative susceptibility if it is calcium. 3, 6 The following summary links the physics sign to common tissue sources.

Susceptibility sign Representative sources Example clinical use of QSM
Paramagnetic (positive ) Non-heme iron, ferritin/hemosiderin, deoxyhemoglobin, methemoglobin Brain iron quantification; microbleed and hemorrhage characterization; venous imaging
Diamagnetic (negative ) Calcium, myelin Distinguishing calcification from hemorrhage; myelin-related study

Veins and oxygenation

Deoxygenated hemoglobin is paramagnetic, so venous blood has a distinct susceptibility relative to surrounding tissue. QSM depicts cerebral veins with high conspicuity and, because the susceptibility difference depends on the oxygen extraction fraction, it can be used to estimate venous oxygenation — a quantitative window on cerebral metabolism that qualitative imaging cannot provide. 2, 4

Practical Optimization Tips

Getting trustworthy QSM is an exercise in physics discipline from acquisition through reporting.

1. Protect the acquisition

Use a validated 3D multi-echo gradient-echo protocol with flow compensation and monopolar readout, confirm the phase images are reconstructed and saved, and keep echo times, resolution, and field strength fixed across a study or program. 1

2. Standardize the pipeline

Fix the phase-unwrapping, background-field-removal, and dipole-inversion algorithms and their parameters, and document the versions. Changing any stage mid-study can shift values even when the scanner is unchanged. 1, 6

3. Define and record the reference region

Choose a susceptibility reference (for example, CSF or whole-brain mean), apply it consistently, and report it. A QSM value without its reference is not interpretable. 1, 3

4. Keep the MRI fundamentals healthy

Because QSM lives on phase, it inherits the scanner's field-homogeneity and geometric behavior. Maintain shimming and B0 homogeneity, verify geometric accuracy, and keep signal-to-noise high; these are the same parameters your routine MRI QC already covers.

5. Validate before clinical reliance

Treat any QSM implementation — especially a deep-learning inversion — as a method that must be validated against references and monitored, not a black box to be trusted on first use. 1

Common pitfalls to avoid

  • Confusing QSM with SWI. They share input but differ fundamentally; SWI is qualitative and nonlocal, QSM is quantitative and local. 1, 2
  • Comparing values across different pipelines or references. Numbers are pipeline- and reference-specific. 1
  • Ignoring the magic-angle streaking. Unregularized inversion produces artifacts that can mimic or obscure pathology. 6
  • Neglecting B0 and geometric QC. Poor field homogeneity corrupts phase and therefore susceptibility.
  • Over-interpreting absolute numbers. QSM reports relative susceptibility; interpret within the study's own framework. 3

Regulatory Considerations

MRI is non-ionizing, so it falls outside the federal and state radiation-machine programs that govern X-ray equipment, but it is firmly within the accreditation and quality-management framework that payers and accreditors require. QSM does not change those obligations; it depends on them.

Key frameworks to reference:

  • ACR MRI Accreditation Program — establishes image-quality, phantom, and quality-control expectations, including the routine MRI QC program and the requirement for an annual MRI equipment performance evaluation by a qualified medical physicist or MR scientist. 7
  • ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of MRI — defines the scope and content of that physicist evaluation, covering the field homogeneity, geometric accuracy, and signal-to-noise parameters that advanced phase-based methods such as QSM rely upon. 8
  • The Joint Commission — diagnostic imaging standards (for example, EC.02.04.03) require that imaging equipment performance be evaluated and maintained, reinforcing the same QC foundation. 7

Advanced quantitative MRI does not create a separate regulatory category, but it does raise the clinical consequence of QC failures: a drifting shim or a geometric distortion that might be tolerable for qualitative reading can corrupt a quantitative susceptibility value. For facilities offering or developing QSM, the practical compliance posture is to keep the required MRI QC program and annual physics evaluation rigorous and well-documented, and to treat the QSM pipeline itself as a validated, version-controlled method. This connects directly to MRI physics testing, accreditation support, and medical physicist consulting.

Frequently Asked Questions (FAQs)

What is quantitative susceptibility mapping (QSM)?

QSM is an advanced MRI technique that reconstructs the underlying magnetic susceptibility of tissue, voxel by voxel, from the phase of a gradient-echo acquisition. Unlike susceptibility-weighted imaging, which produces a qualitative contrast, QSM yields a quantitative map in parts per million (ppm) or parts per billion (ppb), so iron, calcium, blood products, and myelin can be measured rather than merely visualized.

How is QSM different from susceptibility-weighted imaging (SWI)?

SWI combines magnitude and filtered phase to create a qualitative image that emphasizes veins, microbleeds, and iron, but its contrast depends on echo time, orientation, and processing, and it is nonlocal and nonquantitative. QSM solves an inverse problem to recover the actual susceptibility value at each voxel, which is local, more reproducible, and quantitative, and which distinguishes paramagnetic from diamagnetic sources by their sign.

Why does QSM require solving an ill-posed inverse problem?

A susceptibility distribution produces a magnetic field perturbation described by a dipole convolution. Recovering susceptibility from the measured field means dividing by the dipole kernel in Fourier space, but that kernel is zero on a cone at the magic angle of about 54.7 degrees, so the inversion is ill-posed and naive division produces severe streaking artifacts. Regularized or multi-orientation methods are used to stabilize the solution.

What can QSM measure clinically?

QSM is used to quantify brain iron in the deep gray nuclei, to distinguish calcification (diamagnetic) from hemorrhage (paramagnetic), to characterize microbleeds and cerebral veins, to estimate venous oxygenation, and to study demyelination and neurodegeneration. It is primarily a brain research and advanced clinical tool, though applications outside the brain are developing.

What acquisition does QSM need?

QSM is typically built on a 3D multi-echo gradient-echo acquisition with flow compensation, monopolar readout, full brain coverage, and saved phase data. Consensus recommendations specify echo-time ranges, resolution, and field-strength considerations so that phase is sampled adequately and the processing pipeline can recover susceptibility reliably.

Why is a reproducible processing pipeline important for QSM?

QSM is not a single number from the scanner; it is the output of a multi-step pipeline — phase unwrapping, background-field removal, and dipole inversion — and each step has algorithm choices that affect the result. For values to be comparable across time, scanners, and sites, the acquisition, reconstruction method, and a susceptibility reference region must be standardized and documented.

Does QSM affect a facility's MRI physics and accreditation obligations?

QSM does not replace required MRI quality control, but it raises the stakes on the fundamentals it depends on — B0 homogeneity, phase stability, geometric accuracy, and signal-to-noise. Facilities still owe the routine ACR MRI quality-control program and the annual MRI equipment performance evaluation by a qualified medical physicist; a robust QC foundation is what makes an advanced technique like QSM trustworthy.

Key Takeaways

  • QSM reconstructs quantitative tissue magnetic susceptibility from gradient-echo phase, converting a usually discarded signal into a measurable biomarker in ppm or ppb. 1, 2
  • It is quantitative and local where SWI is qualitative and nonlocal, and its signed values separate paramagnetic iron and blood from diamagnetic calcium and myelin. 1, 3
  • The core challenge is an ill-posed dipole inversion whose kernel vanishes at the magic angle (~54.7°), requiring regularized or multi-orientation solutions. 1, 6
  • QSM is a pipeline, not a scanner output, so acquisition, reconstruction, and reference region must be standardized for reproducibility. 1
  • Its leading uses are brain iron quantification, calcium-versus-hemorrhage discrimination, and venous/oxygenation imaging. 3, 4
  • QSM depends on MRI fundamentals and accreditation-level QC — B0 homogeneity, geometric accuracy, and the annual physicist evaluation. 7, 8

Conclusion

Quantitative susceptibility mapping turns the phase of an MRI signal — long treated as a nuisance — into a quantitative map of a real tissue property. Its power comes from physics: susceptibility has a sign, so QSM can measure brain iron, tell calcium from blood, and probe venous oxygenation in ways qualitative imaging cannot. Its fragility comes from the same physics: the field-to-source inversion is ill-posed, the answer is the product of a multi-step pipeline, and every value is tied to a reference region and a set of algorithm choices. For a facility, the lesson is that advanced quantitative MRI does not loosen the requirement for disciplined physics — it tightens it. Reproducible QSM rides on a healthy magnet, a standardized acquisition, a version-controlled pipeline, and the routine MRI quality-control program and annual physics evaluation that keep the fundamentals honest.

How DRPS Can Help

Diagnostic Radiation Physics Services helps MRI facilities keep the physics foundation that advanced quantitative imaging depends on. This includes the required MRI physics testing and annual equipment performance evaluation, B0 homogeneity and geometric-accuracy assessment, signal-to-noise and phase-stability checks, acceptance testing of new systems, and accreditation support for ACR MRI accreditation and Joint Commission readiness. For programs developing quantitative or research MRI methods such as QSM, our medical physicist consulting can help validate and document the acquisition and processing chain.

DRPS supports facilities across our service locations, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware.

Quantitative imaging is only as trustworthy as the quality program beneath it — and that is exactly where an independent medical physicist adds value.

Related Resources

References

  1. Bilgic B, Costagli M, et al.; QSM Consensus Organization Committee. Recommended implementation of quantitative susceptibility mapping for clinical research in the brain: A consensus of the ISMRM electro-magnetic tissue properties study group. Magn Reson Med. 2024;91(5):1834-1862. doi:10.1002/mrm.30006. PubMed
  2. Liu C, Li W, Tong KA, Yeom KW, Kuzminski S. Susceptibility-weighted imaging and quantitative susceptibility mapping in the brain. J Magn Reson Imaging. 2015;42(1):23-41. doi:10.1002/jmri.24768. PubMed
  3. Harada T, Kudo K, Fujima N, et al. Quantitative Susceptibility Mapping: Basic Methods and Clinical Applications. RadioGraphics. 2022;42(4):1161-1176. doi:10.1148/rg.210054. PubMed
  4. Fushimi Y, Nakajima S, Sakata A, et al. Value of Quantitative Susceptibility Mapping in Clinical Neuroradiology. J Magn Reson Imaging. 2024;59(6):1914-1929. doi:10.1002/jmri.29010. PubMed
  5. Vinayagamani S, Sheelakumari R, Sabarish S, et al. Quantitative Susceptibility Mapping: Technical Considerations and Clinical Applications in Neuroimaging. J Magn Reson Imaging. 2021;53(1):23-37. doi:10.1002/jmri.27058. PubMed
  6. Deistung A, Schweser F, Reichenbach JR. Overview of quantitative susceptibility mapping. NMR Biomed. 2017;30(4):e3569. doi:10.1002/nbm.3569. PubMed
  7. American College of Radiology. MRI Accreditation Program Requirements. acr.org
  8. American College of Radiology; American Association of Physicists in Medicine. ACR–AAPM Technical Standard for Diagnostic Medical Physics Performance Monitoring of Magnetic Resonance Imaging (MRI) Equipment. acr.org
  9. Tang S, Zhang G, Ran Q, et al. Quantitative susceptibility mapping shows lower brain iron content in children with attention-deficit hyperactivity disorder. Hum Brain Mapp. 2022;43(8):2495-2502. doi:10.1002/hbm.25798. PubMed
  10. Tang S, Xu Y, Liu X, et al. Quantitative susceptibility mapping shows lower brain iron content in children with autism. Eur Radiol. 2021;31(4):2073-2083. doi:10.1007/s00330-020-07267-w. PubMed