Pith. sign in

REVIEW 3 major objections 5 minor 14 references

Uncertainty quantification for White Matter Hyperintensity segmentation detects silent failures and improves automated Fazekas quantification

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Uncertainty maps from a stochastic segmentation-network ensemble reveal small deep-white-matter lesions that the deterministic model silently misses, and adding uncertainty features to automated Fazekas grading lifts deep-WMH balanced…

desk verdict Solid UQ benchmark for WMH segmentation, but the headline Fazekas improvement lacks statistical support—overlapping 95% CIs make it compatible with noise. read the letter →

arxiv 2411.17571 v2 pith:W5UM62TB submitted 2024-11-26 eess.IV cs.CVcs.LG

Ben Philps , Maria del C. Valdes Hernandez , Chen Qin , Una Clancy , Eleni Sakka , Susana Munoz Maniega , Mark E. Bastin , Angela C.C. Jochems
show 3 more authors
Joanna M. Wardlaw Miguel O. Bernabeu Alzheimers Disease Neuroimaging Initiative
This is my paper · ORCID
classification eess.IVcs.CVcs.LG
keywords uncertaintyquantificationwhitematterhyperintensitysegmentationFazekasscorepredictiondeepensemblesstochasticnetworksbrainMRIsilentfailuredetectionqualitycontrol
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

White matter hyperintensities (WMH), bright spots in brain MRI that mark small-vessel disease, are usually graded by radiologists on the four-point Fazekas scale, while automated segmentation models often miss the smallest deep-white-matter lesions without warning. This paper claims that uncertainty quantification (UQ) closes that gap: the uncertainty map of a combined stochastic-segmentation-network and deep-ensemble model (SSN-Ens) flags small isolated WMH clusters that the deterministic baseline leaves unsegmented, and marks stroke lesions that look like WMH as ambiguous. It also proposes a Fazekas classifier that combines spatial features from WMH probability maps with features from UQ maps, reporting a deep-WMH balanced accuracy of 0.74 versus 0.67 with spatial features alone and 0.62 with volume alone, with better calibration (root Brier 0.65 versus 0.74). The same UQ features improve automated detection of poor-quality segmentations. If the paper is right, uncertainty maps are not a by-product of segmentation but a clinically usable signal for completeness, severity grading, and quality control.

What carries the argument

The load-bearing object is SSN-Ens: an ensemble of Stochastic Segmentation Networks, where each member places a low-rank Gaussian distribution over logits to capture spatially correlated aleatoric uncertainty, and the ensemble of differently initialised members captures epistemic uncertainty. The UQ map is the predictive entropy of the averaged sample distribution, computed from ten inference samples. The second piece of machinery is the Fazekas feature extractor: rings at 5, 10, and 15 mm from the ventricles, within which volume, connected-component counts, and confluence features are computed from thresholded WMH probability and UQ maps, selected by recursive feature elimination and fed to a regularised logistic regression. The UQ features carry the downstream claim because they represent ambiguous penumbral tissue and small deep lesions that volume or mean-segmentation features do not encode.

What would settle it

Have several independent expert raters re-grade the Challenge and ADNI scans and re-run the Fazekas comparison against consensus or multi-rater labels; if the deep-WMH balanced accuracy advantage of SSN-Ens UQ features over volume-only features (0.74 versus 0.62) shrinks or disappears under those labels, the claimed benefit of uncertainty features for Fazekas grading is an artifact of a single rater's policy.

Watch

Extended reading notes

Core claim

The paper's central claim is that voxelwise uncertainty maps carry information the mean segmentation throws away, and that this information is useful in the clinic. Concretely, it claims that SSN-Ens, an ensemble of ten Stochastic Segmentation Networks that each model spatially correlated aleatoric uncertainty while the ensemble covers epistemic uncertainty, produces uncertainty maps that identify small isolated deep WMH clusters the deterministic baseline neither segments nor flags as uncertain, highlight ambiguity between WMH and similar-looking stroke lesions, and supply spatial features that improve automated Fazekas severity grading. On the out-of-distribution Challenge data, SSN-Ens attains the highest Dice (0.70 for the standard model, 0.72 with the nnUNet backbone) and the lowest top-sample absolute volume difference (12.3% and 11.0%). In the Fazekas task its UQ features raise deep-WMH balanced accuracy from 0.62 (volume only) and 0.67 (spatial features only) to 0.74, with a root Brier score of 0.65 versus 0.74 for volume only, and the same features lift poor-segmentation detection from 0.74 to 0.82 balanced accuracy.

Load-bearing premise

The Fazekas grades used as ground truth for the Challenge and ADNI datasets were produced by a single expert rater for this study, with no second rater or adjudication, so the headline improvement from uncertainty features would be unsupported if those labels are noisy or follow a different policy than other raters.

Editorial extensions

If this is right

  • Uncertainty maps can be used as a lesion-level completeness check: regions of high uncertainty are worth a second look for small deep WMH that the segmentation missed.
  • Automated Fazekas grading should use spatial features rather than total WMH volume: spatial features alone raise deep-WMH balanced accuracy from 0.62 to 0.67, and SSN-Ens UQ features raise it to 0.74 with a better root Brier score.
  • Segmentation quality control can run on UQ features: the QC classifier reaches 0.82 balanced accuracy and a 0.50 root Brier score with SSN-Ens UQ features, versus 0.74 and 0.60 with volume only.
  • Sample diversity is what makes a stochastic method useful: high-diversity methods such as SSN-Ens bracket the true WMH volume in their sample distribution even for small lesions, while low-diversity methods such as Probabilistic U-Net and independent-voxel SSN do not.
  • Combining aleatoric and epistemic uncertainty modelling outperforms either alone, and the SSN-Ens advantage in Dice, top-sample AVD, and missed-lesion coverage persists when the method is re-implemented on an nnUNet backbone.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The ventricle-ring plus uncertainty-feature recipe is a general template for automating other ordinal visual rating scales in neuroimaging, such as perivascular-space severity, wherever a clinician grades from spatial pattern rather than total lesion volume.
  • A direct testable extension, which the paper only gestures at in its future-work section, is longitudinal: voxels flagged as uncertain but unsegmented in normal-appearing white matter may be the sites of future WMH, and follow-up scans could confirm this.
  • The MSS3 rater-policy finding implies that any automated Fazekas system trained on one institution's labels will systematically under- or over-grade elsewhere even on identical anatomy; training on multi-rater consensus labels, which this study does not do, would be the necessary remedy and a natural next experiment.
  • Because only total predictive entropy is used, the quality-control signal may be stronger if decomposed into epistemic versus aleatoric components; re-running the QC task on each component separately would show where the improvement actually comes from.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This paper benchmarks several uncertainty quantification (UQ) methods for white matter hyperintensity (WMH) segmentation and evaluates their downstream utility. The authors train models on an in-house CVD dataset and evaluate on the MICCAI WMH Challenge as an out-of-distribution set, with additional qualitative and downstream analyses on ADNI and MSS3. They report that an ensemble of Stochastic Segmentation Networks (SSN-Ens) improves Dice and top-sample volume estimation, that UQ maps can highlight small deep WMH missed by the deterministic baseline, that spatial features extracted from WMH probability and UQ maps improve automated Fazekas classification over volume-only and segmentation-only features, and that UQ features improve detection of poor-quality segmentations. The experiments include six-fold cross-validation, multiple metrics, bootstrapped confidence intervals, hyperparameter sweeps, and both standard U-Net and nnU-Net backbones.

Significance. The paper addresses a clinically relevant problem and contains several strengths: it compares a wide range of UQ techniques on real out-of-distribution data, ships code, uses public Challenge ground truth for segmentation benchmarking, and includes qualitative analysis of silent failures and stroke-lesion confusion. If the Fazekas claim were statistically robust, the paper would make a valuable contribution by showing that UQ maps carry actionable information beyond the mean segmentation. However, the principal novel claim—that adding UQ features improves Fazekas classification—rests on overlapping confidence intervals and labels produced by a single rater, so the current evidence is weaker than the abstract suggests.

major comments (3)
  1. [Table 4 and Section 3.5.3] The abstract's headline claim that incorporating WMH uncertainty information improves Fazekas classification and calibration is not supported by the reported statistics. For DWMH, SSN-Ens with UQ has balanced accuracy 0.74 (95% CI 0.66, 0.81) versus 0.71 (0.63, 0.77) without UQ, and root Brier 0.65 (0.60, 0.70) versus 0.68 (0.65, 0.72). For PV WMH the corresponding values are 0.73 (0.66, 0.79) versus 0.72 (0.65, 0.79) and 0.64 (0.60, 0.69) versus 0.65 (0.61, 0.70). All pairwise intervals overlap substantially. Because the bootstrap resamples the same train/test splits for every feature set, a paired difference test or a bootstrap confidence interval on the difference is straightforward and should be reported. Without this, the claimed improvement is a point estimate compatible with zero effect. The same concern applies to the QC results in Table 5, where the balanced-accuracy intervals for SSN-Ens with and without UQ overlap (0.82 (0.71, 0.91) versus 0.79 (0.70, 0.87)).
  2. [Section 3.1 and Section 4.2.1] Fazekas labels for the Challenge and ADNI datasets were generated by a single expert rater for this study, and no inter-rater reliability or adjudication is reported. The MSS3 evaluation is then interpreted through a post hoc 'rater policy shift' explanation (Section 4.2.1), supported only by volume distributions per class rather than by any independent rating of the same images. If the labels are noisy or carry rater-specific criteria, the relative ranking of volume-only, spatial-feature, and UQ-feature classifiers in Table 4, and the conclusion that 'the model does generalise well to the MSS3 dataset given the rater policy disagreement', are not established. At minimum, the authors should report a second rater on a subset of images, or explicitly frame the single-rater design and the MSS3 interpretation as limitations and soften the corresponding conclusions.
  3. [Abstract and Table 2] The abstract states that SSN-Ens 'yields the highest Dice and lowest Absolute Volume Difference % (AVD) score'. Table 2 shows that Deep Ensembles has a lower mean AVD (42.7%) than SSN-Ens (43.8%); SSN-Ens's advantage is in Top AVD (12.3% versus 19.2%). The text should specify 'lowest Top AVD' or otherwise reconcile the claim with the table, since the current wording is contradicted by the paper's own results.
minor comments (5)
  1. [Section 3.4.3] The notation in Eqs. (12) and (13) is confusing: the two standard-deviation conventions are described in prose, but the equations do not explicitly define the subscripts on the metric E. Please define E_{r,s} clearly so the reader can follow which average is taken over model runs versus subjects.
  2. [Section 4.1] The first sentence of Section 4.1 reads 'All methods outperform the baseline SEnt Dice score of for the standard models' and appears to be missing the baseline value; the sentence should be corrected.
  3. [Figure 2 caption] The caption contains the typo 'Periventriclar' and should read 'Periventricular'.
  4. [Section 4.2.1] The first sentence of Section 4.2.1 refers to 'PVWM' where the intended abbreviation is 'PVWMH'.
  5. [Section 3.6] The definition of 'poor quality' as the bottom 20th percentile of the SEnt baseline Dice is arbitrary but useful; please state explicitly that the QC experiment measures detection of low Dice relative to the deterministic baseline, not segmentation quality in general.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; results are benchmarked against external ground truth and self-citations are not load-bearing.

full rationale

The paper's derivation chain is not circular. Segmentation performance (Dice, AVD, per-component F1, sUEO, GED) is evaluated against the public MICCAI Challenge ground truth for the out-of-distribution test set and against CVD ground truth for in-distribution metrics, i.e., external targets not constructed from the models' own outputs. The Fazekas classifier is trained to predict visual scores; for Challenge and ADNI these were generated by an expert rater for this study (Section 3.1), which is a labeling limitation (no inter-rater reliability reported) but not a circular reduction, since the labels are not a function of the extracted features. The QC task defines 'poor quality' by SEnt Dice (Section 3.6), but this is an external performance measure of a specific model, and the features are predictive inputs; it is a narrow operationalization rather than an input-output identity. The only self-citations (Philps et al., 2023, 2024) concern loss-function choice and annotator-policy effects; neither is load-bearing, and the MSS3 'rater policy' explanation is corroborated by the volume-distribution analysis in Fig. 7. No equation reduces to an input by construction, and no fitted parameter is renamed as a prediction. The paper is self-contained against external benchmarks.

Assumptions & free parameters 14 free parameters · 7 assumptions · 0 invented entities

The paper introduces no invented entities. The central results rest on standard UQ approximations, a set of hyperparameter choices, and several labeling assumptions, the most fragile being the single-rater Fazekas labels and the post hoc MSS3 rater-policy explanation.

free parameters (14)
  • empty_slice_retention_proportion = 0.1
    Tuned from {0.05, 0.1, 0.3, 0.5, 0.9, 1.0} on validation data; controls class balance in training slices for all models.
  • mc_dropout_rate = 0.2
    Tuned from {0.05, 0.1, 0.2, 0.3, 0.4}; applied to all layers, selected for sample diversity and Dice.
  • ensemble_size = 10
    Tuned from {3, 5, 10, 20, 30}; 10 chosen as a balance of performance and training cost.
  • ssn_covariance_rank = 25
    Tuned from {5, 10, 12, 15, 18, 20, 25, 30, 40, 50}; low ranks reduce expressiveness, higher ranks give marginal gains.
  • evidential_kl_weight = 0.05
    Tuned from {0, 0.01, 0.05, 0.1, 0.5, 1} with a ramp over 50 epochs.
  • p_unet_kl_beta = 1
    Tuned on a logarithmic scale in [0, 10].
  • p_unet_latent_dim = 12
    Tuned from {6, 12, 24, 36}; 12 found as effective as larger sizes.
  • fazekas_probability_threshold_t = 0.2
    Binarization threshold for WMH probability and UQ maps; tuned from {0.1, 0.2, 0.3}; reported main result uses t=0.2.
  • fazekas_feature_count_k = 18 (Fazekas), 9 (QC)
    Number of features kept after recursive feature elimination; k=18 for Fazekas with range [10,26], k=9 for QC with range [6,12].
  • logistic_regression_weight_decay = 10
    Regularization strength for Fazekas and QC logistic regression classifiers.
  • poor_quality_dice_threshold = 0.57
    Dice below 0.57 defines poor-quality segmentations as the bottom 20th percentile of the CVD test set.
  • per_method_learning_rate_and_weight_decay = not reported
    Each UQ method tuned over learning rates {0.01, 0.001, 0.0001} and weight decay [0, 1e-2]; final chosen values are not stated.
  • wmh_probability_threshold = 0.5
    Fixed binarization threshold for WMH probability maps in Section 3.4.3.
  • monte_carlo_sample_count = 10
    Number of stochastic inferences used for uncertainty maps, Top Dice/AVD, and GED; chosen by protocol, not optimized.
assumptions (7)
  • domain assumption MC-dropout approximates the Bayesian posterior predictive via random dropout masks (Eq. 1).
    Used in Section 2.2.2 to justify uncertainty estimates from test-time dropout; this approximation can be miscalibrated.
  • domain assumption Evidential deep learning Dirichlet formulation and KL penalty follow Li et al. (2023a) (Eqs. 2-4).
    Used in Section 2.2.4; the posterior concentration parameterization beta=(e+1)^2 is assumed valid for WMH.
  • domain assumption SSN low-rank covariance captures spatially coherent aleatoric uncertainty (Eq. 6/10).
    Used in Section 2.2.5; voxel independence is assumed broken by the low-rank plus diagonal covariance.
  • domain assumption Fazekas labels for Challenge and ADNI generated by a single expert rater are reliable enough to serve as ground truth.
    Introduced in Section 3.1; no inter-rater reliability or adjudication is reported.
  • ad hoc to paper The MSS3 performance drop is caused by annotator policy shift rather than model failure.
    Section 4.2.1 infers this from per-class WMH volume distributions (Fig. 7), without a second rating of MSS3.
  • domain assumption Lesion instances can be represented as 3D connected components of binarized ground truth, and IOU > 0 counts as detection.
    Used in Section 3.4.3 for lesion-wise silent failure analysis; the detection criterion is permissive for small lesions.
  • domain assumption Combined CVD, ADNI, and Challenge data can be pooled for training the Fazekas classifier despite acquisition and cohort differences.
    Section 3.5.2 uses a 75/25 split over the pooled set; no domain-adversarial correction is applied.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Uncertainty quantification for White Matter Hyperintensity segmentation detects silent failures and improves automated Fazekas quantification." pith.science (2026). https://pith.science/paper/W5UM62TB

@misc{pith2026241117571,
  author       = {Pith},
  title        = {Pith review of: Uncertainty quantification for White Matter Hyperintensity segmentation detects silent failures and improves automated Fazekas quantification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W5UM62TB}},
  note         = {Machine review of arXiv:2411.17571}
}
read the original abstract

White Matter Hyperintensities (WMH) are key neuroradiological markers of small vessel disease present in brain MRI. Assessment of WMH is important in research and clinics. However, WMH are challenging to segment due to their high variability in shape, location, size, poorly defined borders, and similar intensity profile to other pathologies (e.g stroke lesions) and artefacts (e.g head motion). In this work, we assess the utility and semantic properties of the most effective techniques for uncertainty quantification (UQ) in segmentation for the WMH segmentation task across multiple test-time data distributions. We find UQ techniques reduce 'silent failure' by identifying in UQ maps small WMH clusters in the deep white matter that are unsegmented by the model. A combination of Stochastic Segmentation Networks with Deep Ensembles also yields the highest Dice and lowest Absolute Volume Difference % (AVD) score and can highlight areas where there is ambiguity between WMH and stroke lesions. We further demonstrate the downstream utility of UQ, proposing a novel method for classification of the clinical Fazekas score using spatial features extracted from voxelwise WMH probability and UQ maps. We show that incorporating WMH uncertainty information improves Fazekas classification performance and calibration. Our model with (UQ and spatial WMH features)/(spatial WMH features)/(WMH volume only) achieves a balanced accuracy score of 0.74/0.67/0.62, and root brier score of 0.65/0.72/0.74 in the Deep WMH and balanced accuracy of 0.74/0.73/0.71 and root brier score of 0.64/0.66/0.68 in the Periventricular region. We further demonstrate that stochastic UQ techniques with high sample diversity can improve the detection of poor quality segmentations.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 10 canonical work pages

  1. [3]

    Stroke J

    Eur. Stroke J. 6 (1), 81–88. http://dx.doi.org/10.1177/2396987320929617, Publisher: SAGE Publications. Cutillo, C.M., Sharma, K.R., Foschini, L., Kundu, S., Mackintosh, M., Mandl, K.D., 2020. Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency. NPJ Digit. Med. 3 (1), 1–5. Czolbe, S., Arnavaz, K.,...

    arXiv 2025

  2. [6]

    V olume Only: Only the estimated volume from the SEnt model is used as a feature

    SSN Ens w/ UQ Heatmap of Balanced Accuracy Score for Target: WMH Deep 0.62 0.64 0.66 0.68 0.70 0.72 Balanced Accuracy Score Figure (E.1) Balanced Accuracy Score of the Fazekas classification models for Deep WMH for each setting of hyper-parameters t (the threshold at which model outputs are binarized) and k (the remaining number of features included in th...

    work page

  3. [8]

    Implicit Anatomical Rendering for Medical Image Segmentation with Stochastic Experts

    Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol. 12 (8), 822–838. http://dx.doi.org/10.1016/S1474-4422(13)70124-8, Publisher: Elsevier, URL https: //www.thelancet.com/journals/laneur/article/PIIS1474-4422(13)70124-8/fulltext. Medical Image Analysis 105 (2025) 103697 18 B. Ph...

    work page Pith review arXiv 2025

  4. [10]

    V olume Only: Only the estimated volume from the SEnt model is used as a feature

    SSN Ens w/ UQ Heatmap of Root Brier Score for Target: WMH Deep 0.66 0.68 0.70 0.72 0.74 Root Brier Score Figure (E.2) Root Brier Score of the Fazekas classification models for Deep WMH for each setting of hyper-parameters t (the threshold at which model outputs are binarized) and k (the remaining number of features included in the logistic regression mode...

    work page

  5. [11]

    V olume Only: Only the estimated volume from the SEnt model is used as a feature

    SSN Ens w/ UQ Heatmap of Balanced Accuracy Score for Target: WMH PV 0.720 0.725 0.730 0.735 0.740 0.745 0.750 0.755 0.760 Balanced Accuracy Score Figure (F.1) Balanced Accuracy Score of the Fazekas classification models for PV WMH for each setting of hyper-parameters t (the threshold at which model outputs are binarized) andk (the remaining number of feat...

    work page

  6. [12]

    V olume Only: Only the estimated volume from the SEnt model is used as a feature

    SSN Ens w/ UQ Heatmap of Root Brier Score for Target: WMH PV 0.63 0.64 0.65 0.66 0.67 Root Brier Score Figure (F.2) Root Brier Score of the Fazekas classification models for PV WMH for each setting of hyper-parameters t (the threshold at which model outputs are binarized) and k (the remaining number of features included in the logistic regression model af...

    work page

  7. [13]

    V olume Only: Only the estimated volume from the SEnt model is used as a feature

    SSN Ens w/ UQ Heatmap of Balanced Accuracy Score for QC Task 0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82 0.83 Balanced Accuracy Score Figure (G.1) Balanced Accuracy Score of the QC classification models for each setting of hyper-parameters t (the threshold at which model outputs are binarized) and k (the remaining number of features included in the logistic r...

    work page

  8. [14]

    V olume Only: Only the estimated volume from the SEnt model is used as a feature

    SSN Ens w/ UQ Heatmap of Root Brier Score for QC Task 0.50 0.52 0.54 0.56 0.58 Root Brier Score Figure (G.2) Root Brier Score of the QC classification models for each setting of hyper-parameters t (the threshold at which model outputs are binarized) andk (the remaining number of features included in the logistic regression model after recursive feature el...

    work page

Show all 14 references
  1. [2011]

    Stroke 42 (7), 1917–1922

    White matter hyperintensity penumbra. Stroke 42 (7), 1917–1922. Mehrtens, H.A., Gonzalez, C., Mukhopadhyay, A., 2022. Improving robustness and cal- ibration in ensembles with diversity regularization. In: DAGM German Conference on Pattern Recognition. Springer, pp. 36–50. Moji...

    2025

  2. [2013]

    White matter lesion assessment in patients with cognitive impairment and healthy controls: reliability comparisons between visual rating, a manual, and an automatic volumetrical MRI method—the gothenburg MCI study. J. Aging Res. 2013 (1), 198471. Ovadia, Y., Fertig, E., Ren, J...

    2013

  3. [2018]

    Considerations on accuracy, pattern and possible underlying factors of brain microbleed progression in older adults with absence or mild presence of vascular pathology. J. Int. Med. Res. 46 (9), 3518–3538. Harrison, J., Willes, J., Snoek, J., 2024. Variational bayesian last la...

    2024 arXiv

  4. [2021]

    COVID-19 infection segmentation from chest CT images based on scale uncertainty. In: Clinical Image-Based Procedures, Distributed and Collaborative Learning, Artificial Intelligence for Combating COVID-19 and Secure and Privacy- Preserving Machine Learning: 10th Workshop, CLIP...

    2021

  5. [2022]

    A national network of safe havens: Scottish perspective. J. Med. Internet Res. 24 (3), e31684. http://dx.doi.org/10.2196/31684, Company: Journal of Medical Internet Research Distributor: Journal of Medical Internet Research Institution: Journal of Medical Internet Research Lab...

    doi

  6. [2023]

    In: International Conference on Medical Image Computing and Computer-Assisted Intervention

    Robust t-loss for medical image segmentation. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, pp. 714–724. Gootjes, L., Teipel, S., Zebuhr, Y., Schwarz, R., Leinsinger, G., Scheltens, P., Möller, H.J., Hampel, H., 2004. Reg...

    2004

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.