REVIEW 2 major objections 4 minor 2 references
In vivo imaging of central nervous system fluid spaces using synchrotron radiation-based micro computed tomography
T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash
Pith's one-line read Intravital synchrotron micro-CT achieves whole-brain imaging of mouse CSF spaces at 6.3 µm resolution in 10-second scans.
desk verdict The in vivo SRμCT method is real and worth refereeing, but the abstract's 'whole-brain at 6.3 μm' claim is not supported by the stated detector field of view; the whole-brain demonstration is at 8 μm. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing instrument is the intravital SRµCT setup: a modular, 3D-printed radio-transparent mouse holder with multi-point fixation and heating, coupled to beamline cameras and monochromators that tune hard X-ray energy to the contrast agent's K-edge and record 5–15 ms exposures, so a full tomogram takes 10–32 s. Around this, the authors build remote physiological monitoring (temperature, oxygenation, end-tidal CO2, ECG), optional ventilation-cardiac synchronization, retrospective cardiac gating, and quantitative calibration that maps reconstructed linear attenuation coefficients to contrast-agent concentration. This combination is what makes whole-organ, micrometer-scale, time-resolved imaging of moving fluid spaces possible in a living mouse.
What would settle it
Measure intracranial pressure continuously during and after a lateral-ventricle infusion of 1–5.75 µl at the rates used here; if CSF pressure rises substantially above the physiological range or ventricular volume changes during the infusion phase, the observed four-phase contrast dynamics could be driven by the infusion itself rather than by endogenous CSF production and flow.
Extended reading notes
Core claim
The central claim is that intravital SRµCT can image mouse CNS fluid spaces at micrometer resolution with whole-brain field of view while the animal is alive. Using a portable, radio-transparent heated holder and fast monochromatic hard X-ray acquisitions, the authors achieved uniform 6.3 µm voxel imaging of the entire brain and resolved the ventricles, choroid plexus, and cranial and spinal subarachnoid spaces, with contrast-to-noise ratio rising from 0.2 to 7.9 after infusion of a barium nanoparticle agent. They calibrated reconstructed attenuation coefficients to barium concentration and mapped the four-phase spread and clearance of contrast agent after intraventricular and intracisternal infusions. They also showed that ventricular volume falls by about 37% within minutes of euthanasia, that the cardiac cycle deforms the nasopharynx by about 8 µm, and that the choroid plexus moves tens of microns over minutes. Taken together, these demonstrations establish SRµCT as a bridge between multiphoton microscopy and MRI for in vivo mouse brain imaging.
Load-bearing premise
The load-bearing premise is that a mouse under anesthesia, surgically fitted with a tracheal tube and CSF cannula, ventilated, and infused with up to 5.75 µl of nanoparticles still behaves like the intact living animal, so the measured solute spread and tissue motion reflect natural CSF physiology rather than preparation artifacts.
Editorial extensions
If this is right
- CSF solute transport can be mapped over the entire brain at sub-minute temporal resolution, providing volumetric data to validate or refute computational models of CSF flow and clearance.
- The two infusion routes (lateral ventricle and cisterna magna) produce distinct concentration dynamics, so SRµCT can distinguish local infusion-dominated transport from later diffusion- and convection-driven spread.
- Because ventricular volume drops by about 37% shortly after death, quantitative anatomy derived from ex vivo or post-mortem imaging may systematically underestimate live CSF space dimensions.
- Retrospective cardiac gating resolves tissue motion tied to the cardiac cycle, such as the roughly 8 µm nasopharyngeal deformation, and can be extended to search for cardiac-driven brain motion in mice.
- Native-state acquisitions can delineate major CSF compartments without contrast agent or surgery, offering a minimally perturbing baseline for physiological studies.
Reading between the lines
- If intracranial pressure is not monitored, the 1–5.75 µl infusions used here could themselves alter CSF volume and pressure; a testable extension is to record ICP during infusion and compare the observed four-phase dynamics against pressure-matched controls.
- The peri-mortem shrinkage result implies published ventricular volumes from ex vivo µCT or histology of the same mouse strains may carry a systematic bias, which could be corrected with a live-to-post-mortem conversion factor.
- With faster detectors and tighter gating, the method could resolve whether arterial pulsations drive brain motion in mice, a question the paper leaves open because no cyclic ventricular-wall displacement was detected.
- The quantitative concentration maps could seed inverse models of CSF advection-diffusion, since they provide whole-field boundary data that point measurements cannot.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper presents a method for in vivo synchrotron radiation-based micro computed tomography (SRμCT) of mouse CNS fluid spaces, built around a portable holder and life-support system deployed at three synchrotron facilities. It demonstrates native-state and contrast-enhanced whole-organ imaging, quantitative mapping of a barium nanoparticle contrast agent over time, peri-mortem ventricular shrinkage, and motion quantification via retrospective cardiac gating and non-gated time series. The authors report whole-brain imaging at 6.3 μm voxel size and dynamic imaging with scan times of 10–32 s, positioning SRμCT as a bridge between multiphoton microscopy and MRI for mouse brain studies.
Significance. If the technical claims hold, SRμCT fills a genuine gap: no other modality currently offers micrometer-scale resolution and whole-organ coverage in the living mouse brain. The paper's strengths include multi-facility validation, openly shared hardware designs and code, calibrated attenuation-to-concentration mapping, FSC-based resolution estimation, and a clear description of the peri-mortem changes that motivate in vivo imaging. These are concrete, reproducible contributions. The main weakness is that the headline 'whole-brain at 6.3 μm' claim is not supported by the reported detector fields of view, and the physiological interpretation of post-infusion solute dynamics rests on unverified assumptions about the preparation's fidelity to native CSF physiology.
major comments (2)
- [Abstract; Results 'Anatomic imaging...'; Methods 'Native-state imaging at ESRF ID17'; Supplementary Table 1] The claim of whole-brain imaging at 6.3 μm uniform voxel size is contradicted by the stated geometry. The 6.3 μm acquisition (Mouse19) has an effective field of view of 2560 × 780 pixels, i.e., 16.1 mm × 4.9 mm, and the dynamic 6.45 μm acquisitions (Mouse63, Mouse50) have fields of view of 2560 × 700 pixels, i.e., 16.5 mm × 4.5 mm. These vertical extents are smaller than the typical dorsoventral extent of the adult mouse brain (≈5.5–6.5 mm), and no vertical tiling or stitching is described for any of these scans. The only dataset whose field of view plausibly covers the whole brain is the 8.0 μm SPring-8 acquisition (JP28, 16.4 mm × 12.0 mm). The abstract, the Results section, and the Discussion therefore overstate the demonstrated coverage: whole-brain coverage at micrometer resolution is shown at 8 μm, while the 6.3–6.45 μm acquisitions cover only a sub-volume of the brain. Please revise the claims to state that whole-brain coverage is demonstrated at 8.0 μm, or document how the 6.3 μm volume covers the entire brain (e.g., with explicit stitching details).
- [Results 'Mapping spatiotemporal solute distribution...'; Methods 'Intra-cerebroventricular infusion at ESRF ID17'] The manuscript interprets the post-infusion contrast agent distribution as governed by 'the natural interplay of diffusion and convection' (Results) and uses this to argue that SRμCT provides unique data on CSF dynamics. However, the infused volumes (1–5.75 μl) are comparable to or larger than the cited mouse CSF production rate (0.3–0.7 μl/min), and intracranial pressure is not monitored. During the infusion phase the authors correctly attribute transport to the infusion pump, but the post-infusion phase is presented without direct evidence that the preparation does not substantially alter CSF volume, pressure, or flow. To support the 'natural' interpretation, the paper should either provide ICP measurements or a sham-infusion control, or explicitly soften the language to describe solute distribution under the conditions of this anesthetized, ventilated, and infused preparation rather than as unperturbed physiology.
minor comments (4)
- [Supplementary Table 1] The infusion rate and volume for Mouse63 and Mouse50 appear to be swapped: Supplementary Table 1 lists Mouse63 as 0.5 μl/min and 2.5 μl and Mouse50 as 0.2 μl/min and 1 μl, whereas the Methods text states that the intra-cerebroventricular infusion (Mouse63) used 0.2 μl/min for 1 μl and the intra-cisterna magna infusion (Mouse50) used 0.5 μl/min for 2.5 μl.
- [Methods, FSC paragraph under 'Intra-cerebroventricular infusion at ESRF ID17'] The sentence describing the FSC threshold cites reference 53 (Rudick et al., 1982), a CSF clearance study, rather than the FSC methodology reference (van Heel & Schatz, reference 57), and the Savitzky-Golay filter 'window width of 5088' appears to be a typographical error. Please correct the citation and the filter width.
- [Methods, 'Retrospective cardiac-gated imaging at ESRF ID17'] The text reports '2 188 projections per bin' for 18 bins, which sums to approximately 39,400 projections, not the stated 60,000 total acquisitions; please clarify whether some projections were excluded (e.g., outside the minimum cycle duration) or correct the arithmetic.
- [Discussion and Methods] Each quantitative demonstration (e.g., Mouse63 for ICV dynamics, Mouse50 for ICM, CA019 for choroid plexus motion, JP34 for peri-mortem shrinkage) is based on a single animal. This is acceptable for a proof-of-principle methods paper, but the manuscript should state explicitly that these are n=1 demonstrations and should not be interpreted as population-level results.
Circularity Check
No significant circularity: the paper is a methods/capability demonstration whose imaging, calibration, and motion-quantification steps are externally anchored and not derived from their own outputs.
full rationale
This is an imaging methods paper rather than a derivation paper, and I found no load-bearing step that reduces to its own inputs. The paper's central claim is a demonstrated capability: whole-brain SRµCT imaging at 6.3–8 µm voxel size, with calibrated contrast-agent concentration mapping and motion quantification. The concentration calibration (Results, Fig. 2g) is a linear fit of attenuation coefficient to known barium standards; this is standard measurement practice, not a hidden parameter that is later renamed as a prediction. No 'prediction' language is used for the fitted concentrations, and the calibration does not feed back into the claim being tested. The spatial-resolution estimates use Fourier shell correlation against the 1-bit threshold, an external criterion, and are reported as measured values rather than assumed inputs. The gating analysis cites Fardin et al. for the retrospective gating approach, but that citation is a methodological building block and not a self-citation chain carrying the paper's central conclusion; the observed nasopharyngeal motion is an experimental finding, not a consequence of the gating algorithm's assumptions. The peri-mortem ventricular shrinkage and choroid plexus displacement are quantified directly from segmentations of the acquired volumes, with no fitted model being used to generate the measured numbers. I also reviewed the in-scope limitation statements: the paper openly notes the scarcity of suitable beamlines, the loss of physical interpretability of δ/β when tuned, the lack of ground truth for nasopharyngeal displacement, and the absence of detected cyclic ventricular-wall motion. These are limitations or correctness risks, not circularities. The skeptic's field-of-view concern about the 6.3 µm 'whole-brain' claim is a geometric consistency question about detector FOV versus brain size, which is a factual/correctness issue independent of circularity and does not involve any input being used to derive itself. Overall, no circular step was identified, and the appropriate score is 0.
Assumptions & free parameters
free parameters (3)
- Paganin delta/beta ratio =
200
- Calibration line slope and intercept =
slope = 0.0202 ml/(mg cm), intercept = 0.158 cm^-1
- Segmentation thresholds and morphological filter radii =
various (Otsu threshold, 4.5-12.5 pixel radii)
assumptions (5)
- domain assumption Anesthetized, ventilated mice maintain physiologic CNS fluid spaces representative of the living state.
- domain assumption X-ray linear attenuation scales linearly with barium concentration in vivo as calibrated in vitro.
- domain assumption Retrospective cardiac gating correctly assigns projections to cardiac phases based on ECG timing.
- domain assumption Paganin phase retrieval with a single delta/beta ratio adequately models the tissue mixture, improving CNR without introducing artifacts that affect interpretation.
- standard math The Fourier shell correlation 1-bit threshold estimates effective spatial resolution.
Cite this review
Pith. "Pith review of In vivo imaging of central nervous system fluid spaces using synchrotron radiation-based micro computed tomography." pith.science (2026). https://pith.science/paper/JXHLAYLY
@misc{pith2026250703186,
author = {Pith},
title = {Pith review of: In vivo imaging of central nervous system fluid spaces using synchrotron radiation-based micro computed tomography},
year = {2026},
howpublished = {\url{https://pith.science/paper/JXHLAYLY}},
note = {Machine review of arXiv:2507.03186}
}
abstract
Current approaches to in vivo imaging of the mouse central nervous system (CNS) do not offer a combination of micrometer resolution and a whole-brain field of view. To address this limitation, we introduce an approach based on synchrotron radiation-based hard X-ray micro computed tomography (SR$\mu$CT). We performed intravital SR$\mu$CT acquisitions of mouse CNS fluid spaces at three synchrotron radiation facilities. Imaging was conducted on both anesthetized free-breathing and ventilated animals, with and without retrospective cardiac gating. We achieved whole-brain imaging at 6.3 $\mu$m uniform voxel size, observed the distribution of cerebrospinal fluid (CSF) contrast agent over time and quantified choroid plexus movement. SR$\mu$CT bridges the gap between multiphoton microscopy and magnetic resonance imaging, offering dynamic imaging with micrometer-scale resolution and whole-organ field of view. Intravital SR$\mu$CT will play a crucial role in validating and integrating hypotheses on CSF dynamics and solute transport by providing unique data that cannot be acquired otherwise.
Reference graph
Works this paper leans on
-
[1]
Mittone, A., Manakov, I., Broche, L., Jarnias, C., Coan, P. & Bravin, A. Characterization of a sCMOS-based high-resolution imaging system. J. Synchrotron Rad 24, 1226-36 (2017)
work page 2017
-
[2]
Mirone, A., Gouillart, E., Brun, E., Tafforeau, P. & Kieffer, J. PyHST2: an hybrid distributed code for high speed tomographic reconstruction with iterative reconstruction and a priori knowledge capabilities. Nucl Instrum Methods Phys Res B 324, 41–48 (2013)
work page 2013
Reviewed August 6, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.