Pith. sign in

REVIEW 4 major objections 5 minor 26 references

Towards Scalable and Robust White Matter Lesion Localization via Multimodal Deep Learning

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Combining T1 and FLAIR in one network yields the best white-matter lesion segmentation, while joint lesion-region training degrades both tasks.

desk verdict A clean, well-organized WMH segmentation benchmark whose headline claim—multimodal input significantly beats FLAIR-only—rests on a 0.02 Dice gap with no significance test. read the letter →

arxiv 2506.22041 v1 pith:DQCR7QQN submitted 2025-06-27 eess.IV cs.CV

Julia Machnio , Sebastian N{o}rgaard Llambias , Mads Nielsen , Mostafa Mehdipour Ghazi This is my paper · ORCID
classification eess.IVcs.CV
keywords multimodaldeeplearningwhitematterhyperintensitylesionsegmentationanatomicallocalizationmulti-taskmissingmodalitiesFLAIRMRIT1-weighted
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

This paper proposes a deep learning framework that segments white matter hyperintensities (WMH) and their anatomical locations directly in native space from FLAIR and T1 MRI. It claims that feeding both contrasts as separate channels into one 3D U-Net gives the best lesion Dice (0.74), beating FLAIR alone (0.72) and T1 alone (0.59), and that training the two modalities as interchangeable inputs trades a little accuracy (0.67) for the ability to make predictions when one sequence is missing. It also reports that jointly predicting lesion and white-matter region masks in one network degrades both tasks (lesion Dice 0.43, region Dice 0.29), interpreting this as representational conflict between tasks. These results matter because automated, registration-free localization of lesions by anatomical region could make WMH analysis more scalable and clinically deployable.

What carries the argument

The load-bearing mechanism is a 3D U-Net with four training configurations: FLAIR-only, T1-only, concatenated FLAIR and T1 channels, and an interchangeable-modality setup, all trained with a composite cross-entropy plus Dice loss and heavy MRI-specific augmentation. For anatomical supervision, the method registers the JHU MNI White Matter Atlas Type II to each subject's T1 scan using the extracted white-matter region, yielding 34 native-space region labels that also define the regional lesion burden. This design lets the paper attribute differences in Dice to input configuration and task setup, isolating the effect of multimodal fusion and the cost of joint learning.

What would settle it

Evaluate the same four input configurations on the untouched official WMH challenge test split: if FLAIR-only reaches or exceeds the concatenated model's 0.74 Dice, the claimed multimodal advantage collapses. Separately, compare the affine-registered atlas region labels against manual anatomical parcellations on a subset of subjects to test whether the region-wise lesion estimates are trustworthy.

Watch

Extended reading notes

Core claim

The paper's central discovery is that multimodal fusion and task modularity are both load-bearing for WMH analysis. Concatenating T1 and FLAIR in a shared model outperforms either modality alone, while an interchangeable-training model preserves clinically useful robustness to missing inputs at a modest cost. In the multi-task setting, uniting lesion segmentation and region segmentation in a single network substantially reduces performance relative to separate models, which the paper attributes to optimization conflicts or representational interference between tasks. The anatomical localization results therefore support the use of separate or carefully designed task-specific models over naive joint training.

Load-bearing premise

The paper assumes the affine registration of the JHU MNI White Matter Atlas Type II to each subject's T1 via the extracted white-matter region produces anatomically correct labels, without validating against manual annotations, so any registration error would propagate into every region-wise lesion estimate.

Editorial extensions

If this is right

  • If both T1 and FLAIR are available, a concatenated-input model should be the default choice for WMH lesion segmentation, since it reaches Dice 0.74 versus 0.72 for FLAIR alone.
  • A modality-interchangeable model can run inference with only FLAIR or only T1, giving a practical fallback for clinical or multi-site protocols where one sequence is missing, with accuracy within about 0.07 Dice of the best model.
  • The multi-task joint model can output lesion burden per anatomical region in a single pass, but at a large cost (lesion Dice 0.43, region Dice 0.29), so naive joint training should be avoided unless architectural changes reduce task interference.
  • Anatomical region segmentation itself is stable across input types, around 0.75 Dice, suggesting that FLAIR-only pipelines can still localize lesions to regions when T1 is absent.

Reading between the lines

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

  • Beyond the paper, the multi-task drop may stem from the entangled supervision signal: because regional lesion labels are the product of lesion and region masks, the network receives no clean gradient for either task until both are partly solved, so a two-branch architecture with independent binary outputs could recover much of the lost Dice.
  • Beyond the paper, the affine atlas registration is the least controlled link in the localization pipeline, so an independent validation against manual anatomical parcellations would likely revise the region-wise burden numbers and clarify whether the 0.29 region Dice reflects learning failure or label noise.
  • Beyond the paper, treating FLAIR and T1 as interchangeable views could serve as a built-in domain-adaptation strategy: training the network to ignore modality identity may make it less sensitive to scanner- or site-specific contrast differences, which the multi-site WMH dataset could test directly.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. This paper presents a 3D U-Net framework for white matter hyperintensity (WMH) segmentation and anatomical localization in native space. It compares four input configurations: FLAIR-only, T1-only, concatenated FLAIR and T1, and a modality-interchangeable training scheme, and additionally explores a multi-task model that jointly predicts lesion and anatomical region masks. Experiments on the MICCAI WMH Segmentation Challenge dataset report that concatenated multimodal input achieves the highest lesion Dice score (0.74), that the modality-interchangeable setup trades accuracy for robustness (0.67), and that joint multi-task learning degrades performance (lesion Dice 0.43, region Dice 0.29). The paper concludes that multimodal fusion improves segmentation and that multi-task learning suffers from representational conflict between tasks.

Significance. If the findings are statistically robust, the practical contribution is real: native-space processing avoids explicit registration for lesion segmentation, and the modality-interchangeable configuration addresses a clinically common missing-modality scenario. The multi-task experiments are also a useful negative result about task interference. However, the contribution is modest in scope: all comparisons are internal, no state-of-the-art external baseline is reported, and the central quantitative claim rests on a 0.02 Dice difference without statistical inference. The paper is clearly written and the experimental design is straightforward, but the evidence as presented does not yet support the word 'significantly' in the abstract and conclusion.

major comments (4)
  1. [Section 3.3, Table 2] The central claim that 'multimodal input significantly improves the segmentation performance' is not supported by any statistical test. The concatenated T1+FLAIR model achieves Dice 0.74 ± 0.11 versus FLAIR-only 0.72 ± 0.12 on a 60-subject test set; with overlapping standard deviations, a 0.02 mean difference can easily arise from sampling noise. Please report per-subject paired comparisons (e.g., Wilcoxon signed-rank test or permutation test) with effect sizes or confidence intervals for the differences between configurations. If the difference is not significant, the headline conclusion must be softened accordingly.
  2. [Section 3.1, Table 1] The dataset split description is ambiguous. The text states that the authors 'inverted the original challenge-defined splits, repurposing the original test set for training' and then used the resulting dataset for 5-fold cross-validation, but Table 1 lists Train 110 and Test 60 with no further explanation. It is not clear which subjects constitute the final test set, whether the reported metrics are averages over cross-validation folds or from a single held-out set, or how the inversion interacts with the 5-fold procedure. Please specify the exact subject allocation and evaluation protocol.
  3. [Section 2.2] The anatomical WM region labels are generated by affinely registering the JHU MNI White Matter Atlas to each subject's T1 using the extracted WM region from [19], but no validation of this registration against manual anatomical annotations is provided. Registration errors propagate directly into the region segmentation scores and the region-wise lesion burden estimates, which are central to the localization claims. Please report a registration-quality measure (e.g., overlap with manual labels on a subset, or at least a sensitivity analysis) or explicitly discuss this limitation.
  4. [Section 3.4, Tables 4 and 5] The comparison between single-task and multi-task region segmentation may not be apples-to-apples. The multi-task model is trained on masked regional labels (lesion masks intersected with WM region labels), while the single-task region model is trained on full 34-region anatomical masks. The text in Section 3.4 says region predictions were 'evaluated only for WM subregions present in each scan,' but it is not clear whether the multi-task region Dice is computed over all anatomical regions or only over regions containing lesions. Please clarify the label space used for each metric and, if the multi-task region score is restricted to lesion-bearing regions, state this explicitly and adjust the interpretation.
minor comments (5)
  1. [Abstract and Section 3.3] The word 'significantly' is used without statistical support; consider replacing it with 'numerically higher' unless paired tests are added.
  2. [Figure 1 caption] The caption describes configuration (D) as 'sequential training where FLAIR and T1 are treated as interchangeable modalities,' while the body text says they are 'considered interchangeable variants, effectively augmenting the training set.' Please harmonize the terminology.
  3. [Section 2.3] For configurations (A) and (B), the text says predictions from T1 and FLAIR scans are fused by averaging softmax outputs, but these models were trained on a single modality. Please clarify how the unavailable modality is handled at inference and how the 'T1 & FLAIR' column in Table 2 is generated for the FLAIR-only and T1-only rows.
  4. [Tables 2 and 5] There are typographical artifacts in the tables, such as '0 .73' instead of '0.73'; please correct the formatting.
  5. [Section 4] The conclusion states 'Compared to state-of-the-art WMH segmentation approaches...' but no external state-of-the-art comparison is reported in the paper. Either include such a comparison or rephrase this sentence.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the paper is an empirical benchmarking study whose claims rest on direct test-set measurements, not on inputs reconstructed from outputs.

full rationale

The paper contains no formal derivation, equations, or fitted parameters that are renamed as predictions. The central comparisons (Table 2: concatenated T1+FLAIR DSC 0.74 vs FLAIR-only 0.72 and T1-only 0.59) are direct measurements on the MICCAI WMH test set; the multimodal-input advantage is an empirical result, not a consequence of how the models or labels were defined. Self-citations [16], [17], [18], [19] are used for augmentation code, the Yucca framework, the extraction of WM regions for atlas registration, and a prior native-space region segmentation method; none is invoked as evidence for the headline segmentation claim, and the WMH lesion labels come from the external MICCAI challenge rather than from the authors' own tools. The atlas-derived region labels (Section 2.2) are an input representation, and the reported region Dice measures agreement with those labels, so there is no construction-level equivalence between input and output. The absence of paired significance tests for the 0.02 DSC gap is a statistical-rigor concern, not circularity. No self-definitional, fitted-input-as-prediction, or uniqueness-imported-from-authors pattern is present.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The paper is empirical, so there are no invented physical entities. The main unvalidated inputs are the atlas-derived region labels and the affine registration that produces them. The single listed free parameter is the loss weighting, which is a hand-chosen constant that could affect the multi-task comparison.

free parameters (1)
  • Composite loss weight (CE:Dice) = 1:1
    Equal weighting of cross-entropy and Dice in Section 2.3; no tuning is reported, and the multi-task drop could depend on this ratio.
assumptions (3)
  • domain assumption JHU MNI White Matter Atlas Type II refined labels are a valid ground truth for WM subregions
    Section 2.2 uses these atlas labels as supervision for region segmentation and for constructing regional lesion labels; if the atlas is inaccurate for a subject, all localization metrics are affected.
  • domain assumption Affine registration of the atlas to the subject T1 (using Fast-AI Brain [19]) is accurate in native space
    Section 2.2 generates subject-specific labels via an affine transform of the atlas T1; no registration-quality check is reported.
  • domain assumption Repurposing the original MICCAI test set for training and 5-fold CV yields a valid evaluation
    Section 3.1 inverts the challenge splits; this changes the evaluation protocol from the benchmark's intended use, and no external validation is performed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Scalable and Robust White Matter Lesion Localization via Multimodal Deep Learning." pith.science (2026). https://pith.science/paper/DQCR7QQN

@misc{pith2026250622041,
  author       = {Pith},
  title        = {Pith review of: Towards Scalable and Robust White Matter Lesion Localization via Multimodal Deep Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DQCR7QQN}},
  note         = {Machine review of arXiv:2506.22041}
}
read the original abstract

White matter hyperintensities (WMH) are radiological markers of small vessel disease and neurodegeneration, whose accurate segmentation and spatial localization are crucial for diagnosis and monitoring. While multimodal MRI offers complementary contrasts for detecting and contextualizing WM lesions, existing approaches often lack flexibility in handling missing modalities and fail to integrate anatomical localization efficiently. We propose a deep learning framework for WM lesion segmentation and localization that operates directly in native space using single- and multi-modal MRI inputs. Our study evaluates four input configurations: FLAIR-only, T1-only, concatenated FLAIR and T1, and a modality-interchangeable setup. It further introduces a multi-task model for jointly predicting lesion and anatomical region masks to estimate region-wise lesion burden. Experiments conducted on the MICCAI WMH Segmentation Challenge dataset demonstrate that multimodal input significantly improves the segmentation performance, outperforming unimodal models. While the modality-interchangeable setting trades accuracy for robustness, it enables inference in cases with missing modalities. Joint lesion-region segmentation using multi-task learning was less effective than separate models, suggesting representational conflict between tasks. Our findings highlight the utility of multimodal fusion for accurate and robust WMH analysis, and the potential of joint modeling for integrated predictions.

Figures

Figures reproduced from arXiv: 2506.22041 by the authors.

Figure 1
Figure 1. Overview of the proposed method for WM lesion segmentation. The pipeline illustrates four input configurations used during [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed multi-task framework for multimodal regional WMH segmentation. The pipeline adopts the same [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 24 canonical work pages

  1. [19]

    FAST-AID Brain: Fast and Accurate Segmentation Tool using Artificial Intelligence Developed for Brain

    Mostafa Mehdipour Ghazi and Mads Nielsen. Fast-aid brain: Fast and accurate segmentation tool using artificial intelli- gence developed for brain.arXiv preprint arXiv:2208.14360,

    work page Pith review arXiv

  2. [1]

    Lesion location and cognitive impact of cerebral small vessel disease

    J Matthijs Biesbroek, Nick A Weaver, and Geert Jan Bies- sels. Lesion location and cognitive impact of cerebral small vessel disease. Clinical Science, 131(8):715–728, 2017. 1

    work page 2017

  3. [2]

    Association of stroke lesion pattern and white matter hyperintensity burden with stroke severity and outcome

    Anna K Bonkhoff, Sungmin Hong, Martin Bretzner, Markus D Schirmer, Robert W Regenhardt, E Murat Arsava, Kathleen Donahue, Marco Nardin, Adrian Dalca, Anne- Katrin Giese, et al. Association of stroke lesion pattern and white matter hyperintensity burden with stroke severity and outcome. Neurology, 99(13):e1364–e1379, 2022. 1

    work page 2022

  4. [3]

    Reconsidering harbingers of dementia: progres- sion of parietal lobe white matter hyperintensities predicts alzheimer’s disease incidence

    Adam M Brickman, Laura B Zahodne, Vanessa A Guzman, Atul Narkhede, Irene B Meier, Erica Y Griffith, Frank A Provenzano, Nicole Schupf, Jennifer J Manly, Yaakov Stern, et al. Reconsidering harbingers of dementia: progres- sion of parietal lobe white matter hyperintensities predicts alzheimer’s disease incidence. Neurobiology of aging , 36 (1):27–32, 2015. 1

    work page 2015

  5. [4]

    Strate- gic white matter hyperintensity locations for cognitive im- pairment: a multicenter lesion-symptom mapping study in 3525 memory clinic patients

    Mirthe Coenen, Hugo J Kuijf, Irene MC Huenges Wa- jer, Marco Duering, Frank J Wolters, Evan F Fletcher, Pauline M Maillard, Alzheimer’s Disease Neuroimaging Ini- tiative, Frederik Barkhof, Josephine Barnes, et al. Strate- gic white matter hyperintensity locations for cognitive im- pairment: a multicenter lesion-symptom mapping study in 3525 memory clinic ...

    work page 2023

  6. [5]

    Strategic white matter hyperintensity locations associated with post-stroke cognitive impairment: A multicenter study in 1568 stroke patients

    Mirthe Coenen, Floor AS de Kort, Nick A Weaver, Hugo J 5 Kuijf, Hugo P Aben, Hee-Joon Bae, R ´egis Bordet, Christo- pher PLH Chen, Anna Dewenter, Thomas Doeven, et al. Strategic white matter hyperintensity locations associated with post-stroke cognitive impairment: A multicenter study in 1568 stroke patients. International Journal of Stroke, page 17474930...

    work page 2024

  7. [6]

    The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: systematic review and meta-analysis

    St ´ephanie Debette and HS Markus. The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: systematic review and meta-analysis. Bmj, 341,

    work page

  8. [7]

    White matter hyperintensities are common in midlife and al- ready associated with cognitive decline

    Tracy d’Arbeloff, Maxwell L Elliott, Annchen R Knodt, Tracy R Melzer, Ross Keenan, David Ireland, Sandhya Ram- rakha, Richie Poulton, Tim Anderson, Avshalom Caspi, et al. White matter hyperintensities are common in midlife and al- ready associated with cognitive decline. Brain Communica- tions, 1(1):fcz041, 2019. 1

    work page 2019

Show all 26 references
  1. [8]

    Small white matter lesion detection in cerebral small vessel disease

    Mohsen Ghafoorian, Nico Karssemeijer, Inge van Uden, Frank Erik de Leeuw, Tom Heskes, Elena Marchiori, and Bram Platel. Small white matter lesion detection in cerebral small vessel disease. In Medical Imaging 2015: Computer- Aided Diagnosis, pages 265–270. SPIE, 2015. 2

    2015

  2. [9]

    Location sensitive deep convolutional neural networks for segmentation of white matter hyperintensities

    Mohsen Ghafoorian, Nico Karssemeijer, Tom Heskes, Inge WM van Uden, Clara I Sanchez, Geert Litjens, Frank- Erik de Leeuw, Bram van Ginneken, Elena Marchiori, and Bram Platel. Location sensitive deep convolutional neural networks for segmentation of white matter hyperintensitie...

    2017

  3. [10]

    The impact of strategic white matter hy- perintensity lesion location on language.The American Jour- nal of Geriatric Psychiatry, 29(2):156–165, 2021

    Saima Hilal, J Matthijs Biesbroek, Henri Vrooman, Eddie Chong, Hugo J Kuijf, Narayanaswamy Venketasubramanian, Ching-Yu Cheng, Tien Yin Wong, Geert Jan Biessels, and Christopher Chen. The impact of strategic white matter hy- perintensity lesion location on language.The America...

    2021

  4. [11]

    Im- provement of the image quality of t1-weighted anatomical brain scans

    Clare Howarth, Chloe Hutton, and Ralf Deichmann. Im- provement of the image quality of t1-weighted anatomical brain scans. Neuroimage, 29(3):930–937, 2006. 1

    2006

  5. [12]

    Deep-learning-based segmentation and localization of white matter hyperintensities on mag- netic resonance images

    Wenhao Jiang, Fengyu Lin, Jian Zhang, Taowei Zhan, Peng Cao, and Silun Wang. Deep-learning-based segmentation and localization of white matter hyperintensities on mag- netic resonance images. Interdisciplinary Sciences: Com- putational Life Sciences, 12(4):438–446, 2020. 2

    2020

  6. [13]

    Standardized assessment of automatic segmentation of white matter hyperintensities and results of the wmh segmentation challenge

    Hugo J Kuijf, J Matthijs Biesbroek, Jeroen De Bresser, Rut- ger Heinen, Simon Andermatt, Mariana Bento, Matt Berseth, Mikhail Belyaev, M Jorge Cardoso, Adria Casamitjana, et al. Standardized assessment of automatic segmentation of white matter hyperintensities and results of t...

    2019

  7. [14]

    An anatomical knowledge-based mri deep learning pipeline for white matter hyperintensity quantification associated with cognitive im- pairment

    Li Liang, Pengzheng Zhou, Wanxin Lu, Xutao Guo, Chen- fei Ye, Haiyan Lv, Tong Wang, and Ting Ma. An anatomical knowledge-based mri deep learning pipeline for white matter hyperintensity quantification associated with cognitive im- pairment. Computerized Medical Imaging and Gra...

    2021

  8. [15]

    Examining the identification of age-related atrophy between t1 and t1+ t2-flair cortical thickness measurements.Scientific reports, 9(1):11288, 2019

    Heidi Lindroth, Veena A Nair, Casandra Stanfield, Cameron Casey, Rosaleena Mohanty, Daniel Wayer, Paul Rowley, Roger Brown, Vivek Prabhakaran, and Robert D Sanders. Examining the identification of age-related atrophy between t1 and t1+ t2-flair cortical thickness measurements....

    2019

  9. [16]

    Data augmentation-based unsupervised domain adaptation in medical imaging

    Sebastian Nørgaard Llambias, Mads Nielsen, and Mostafa Mehdipour Ghazi. Data augmentation-based unsupervised domain adaptation in medical imaging. arXiv preprint arXiv:2308.04395, 2023. 2

    2023 arXiv

  10. [17]

    Yucca: A deep learning framework for medical image analysis

    Sebastian Nørgaard Llambias, Julia Machnio, Asbjørn Munk, Jakob Ambsdorf, Mads Nielsen, and Mostafa Mehdipour Ghazi. Yucca: A deep learning framework for medical image analysis. arXiv preprint arXiv:2407.19888 ,

    arXiv

  11. [18]

    Deep learning for localization of white matter lesions in neurological diseases

    Julia Machnio, Mads Nielsen, and Mostafa Mehdipour Ghazi. Deep learning for localization of white matter lesions in neurological diseases. In Northern Lights Deep Learning Conference 2025, 2024. 2

    2025

  12. [20]

    Atlas-based whole brain white matter analysis using large deformation diffeo- morphic metric mapping: application to normal elderly and alzheimer’s disease participants

    Kenichi Oishi, Andreia Faria, Hangyi Jiang, Xin Li, Kazi Akhter, Jiangyang Zhang, John T Hsu, Michael I Miller, Pe- ter CM van Zijl, Marilyn Albert, et al. Atlas-based whole brain white matter analysis using large deformation diffeo- morphic metric mapping: application to norm...

    2009

  13. [21]

    Neuroradiological differentiation of white matter le- sions in patients with multiple sclerosis and fabry disease

    Jakob Rath, Olivia Foesleitner, Lukas Haider, Hubert Bickel, Fritz Leutmezer, Stephan Polanec, Michael A Arnoldner, Gere Sunder-Plassmann, Daniela Prayer, Thomas Berger, et al. Neuroradiological differentiation of white matter le- sions in patients with multiple sclerosis and ...

    2022

  14. [22]

    U- net: Convolutional networks for biomedical image segmen- tation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U- net: Convolutional networks for biomedical image segmen- tation. In Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, pa...

    2015

  15. [23]

    Flair*: a combined mr contrast tech- nique for visualizing white matter lesions and parenchymal veins

    Pascal Sati, Ilena C George, Colin D Shea, Mar ´ıa I Gait ´an, and Daniel S Reich. Flair*: a combined mr contrast tech- nique for visualizing white matter lesions and parenchymal veins. Radiology, 265(3):926–932, 2012. 1

    2012

  16. [24]

    Multi-atlas based detec- tion and localization (madl) for location-dependent quantifi- cation of white matter hyperintensities

    Dan Wu, Marilyn Albert, Anja Soldan, Corinne Pettigrew, Kenichi Oishi, Yusuke Tomogane, Chenfei Ye, Ting Ma, Michael I Miller, and Susumu Mori. Multi-atlas based detec- tion and localization (madl) for location-dependent quantifi- cation of white matter hyperintensities. Neuro...

    2019

  17. [25]

    Significance of white matter high intensity lesions as a predictor of stroke from arteriolosclerosis

    H Yamauchi, H Fukuda, and C Oyanagi. Significance of white matter high intensity lesions as a predictor of stroke from arteriolosclerosis. Journal of Neurology, Neurosurgery & Psychiatry, 72(5):576–582, 2002. 1

    2002

  18. [26]

    Analysis of risk factors for white matter hyperin- tensity in older adults without stroke

    Kai Zheng, Zheng Wang, Xi Chen, Jiajie Chen, Yu Fu, and Qin Chen. Analysis of risk factors for white matter hyperin- tensity in older adults without stroke. Brain Sciences, 13(5): 835, 2023. 1 6

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.