Pith. sign in

REVIEW 4 major objections 6 minor 50 references

SegHeD+: Segmentation of Heterogeneous Data for Multiple Sclerosis Lesions with Anatomical Constraints and Lesion-aware Augmentation

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read SegHeD+ claims that a single multi-task 3D V-Net can simultaneously segment all, new, and vanishing MS lesions from heterogeneous datasets, and that this unified model outperforms task-specific state-of-the-art methods on all-lesion…

desk verdict Solid multi-task extension of SegHeD for MS lesion segmentation, but the vanishing-lesion benchmark is a simulated time-reversal task and is oversold; the public all/new lesion results are credible. read the letter →

arxiv 2412.10946 v1 pith:DAK4UXF2 submitted 2024-12-14 cs.CV cs.LG

Berke Doga Basaran , Paul M. Matthews , Wenjia Bai This is my paper · ORCID
classification cs.CVcs.LG
keywords multiplesclerosislesionsegmentationheterogeneousdatamulti-tasklearninglongitudinalMRIvanishinglesionsnewMixaugmentation
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

The paper tries to show that one segmentation model can learn from MS brain MRI datasets that differ in format (single scan vs. longitudinal) and in what is annotated (all lesions, new lesions, or vanishing lesions), and can simultaneously produce all three lesion maps. The intended payoff is that clinical and research data no longer need to be harmonized into one annotation protocol before they are useful. SegHeD+ combines a multi-head 3D V-Net with three anatomical loss terms — longitudinal, volumetric, and spatial constraints — and a lesion-level augmentation method called LesionMix that synthesizes extra training examples, including rare vanishing lesions. On five datasets it reports higher Dice and F1 scores than task-specific state-of-the-art methods for all-lesion segmentation, comparable performance for new-lesion segmentation, and a first benchmark for vanishing lesions.

What carries the argument

The load-bearing mechanism is a four-headed 3D V-Net trained with a composite loss. Three of the loss terms encode anatomical knowledge: a longitudinal constraint enforces that new lesions are absent at baseline and present at follow-up while vanishing lesions satisfy the reverse; a volumetric constraint penalizes total lesion volume changes beyond thresholds $\alpha_{\text{high}}$ and $\alpha_{\text{low}}$ (1.2 and 0.8 for annual change); and a spatial constraint penalizes predictions outside the white matter mask, which is also fed as an input channel. Around this, LesionMix performs lesion populating and inpainting to synthesize new and vanishing lesions, rebalancing the scarce vanishing-lesion labels and enlarging the training set from 144 to 400 images.

What would settle it

Train SegHeD+ on a real longitudinal MS dataset with expert-annotated disappearing lesions and compare VAN-trained vs real-trained performance on those real cases; also directly measure whether time-reversed new-lesion masks overlap with true vanishing-lesion masks. If the overlap is low or performance drops, the simulation assumption fails.

Watch

Extended reading notes

Core claim

SegHeD+ is a single multi-task segmentation model that takes up to four inputs — baseline and follow-up FLAIR scans, an all-lesion label for the baseline scan when available, and a white matter mask — and outputs four segmentations: all lesions at each timepoint, new lesions at follow-up, and vanishing lesions at follow-up. The method's key claim is that combining heterogeneous training data with constraints derived from MS domain knowledge makes the model generalize better than task-specific models trained on homogeneous data. The paper reports that SegHeD+ outperforms the compared state-of-the-art methods on all-lesion segmentation for both longitudinal (MS2015: 78.57 Dice) and cross-sectional (MS2016: 85.18 Dice) benchmarks, achieves the best new-lesion Dice among heterogeneous methods (MSSEG-2: 50.52), and sets a vanishing-lesion benchmark at 43.84 Dice on the synthesized VAN dataset. It also shows temporal consistency: predicted lesion volumes track ground truth across four timepoints with higher Pearson correlation than baselines.

Load-bearing premise

The load-bearing assumption is that a vanishing lesion can be faithfully simulated by reversing the timepoints of a new-lesion dataset, so that lesions appearing at the second scan become lesions disappearing from the first scan.

Editorial extensions

If this is right

  • A single SegHeD+ model can replace separate models for cross-sectional, longitudinal, new-lesion, and vanishing-lesion segmentation, so heterogeneous multi-site data can be pooled without re-annotation.
  • The anatomical constraints improve temporal consistency of lesion volume estimates, which matters for tracking disease progression across follow-up scans.
  • LesionMix augmentation is what mainly lifts vanishing-lesion segmentation (from 35.23 to 43.84 Dice) and could be applied to other rare lesion types.
  • The method sets a reference benchmark for vanishing-lesion segmentation, a task previously unaddressed by public challenges.
  • The baseline all-lesion label is an optional input channel, so the model still works when such labels are missing from a dataset.

Reading between the lines

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

  • The VAN dataset treats 'vanishing' as exact time-reversal of 'new,' an assumption the paper itself flags as questionable; a real longitudinal dataset with actual disappearing-lesion annotations would be the decisive test and would likely change both training signal and benchmark numbers.
  • Because LesionMix is the main driver of vanishing-lesion gains, increasing the fraction of inpainted (vanishing) examples or using real vanishing-lesion patches could close the remaining gap between new and vanishing Dice scores.
  • The multi-task, heterogeneous-data recipe is not MS-specific; the same input-channel and constraint pattern could be carried over to other diseases where lesions appear, disappear, or change volume across timepoints (e.g., stroke or tumor follow-up).
  • The reported 'superior' claim is task-dependent: for new-lesion segmentation on MSSEG-2, some dedicated methods still score higher; the paper's own framing is that a unified model trades a small new-lesion gap for gains across all other tasks.
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 / 6 minor

Summary. The manuscript introduces SegHeD+, a multi-task 3D V-Net that jointly segments all, new, and vanishing MS lesions from heterogeneous cross-sectional and longitudinal FLAIR data. The method takes up to four inputs (baseline/follow-up images, a baseline all-lesion label when available, and a white-matter mask) and produces four segmentation heads. Training combines a Dice loss with longitudinal, volumetric, and spatial constraint losses plus LesionMix lesion-level augmentation. The model is evaluated on five datasets (MS2015, MS2016, MSSEG-2, MSSEG-2+, and the simulated VAN) with Dice and lesion-wise F1 scores; ablations study each loss component, augmentation, and dataset composition. The paper claims superior performance across all three segmentation tasks and positions SegHeD+ as a unified heterogeneous-data model.

Significance. If the claims are fully supported, SegHeD+ would be a valuable step toward a single model that exploits diverse, partially annotated MS datasets, and the public MSSEG-2 evaluation with challenge-standard metrics is a notable strength. The paper also includes a careful ablation of loss terms and augmentation, and it honestly acknowledges several limitations, including the simulated nature of the vanishing-lesion dataset. However, load-bearing parts of the method are currently not implementable as written (the logical operations in the constraint losses are undefined for continuous predictions), and the vanishing-lesion claims are based on a time-reversal simulation that the authors themselves note is not biologically symmetric to real lesion disappearance. These issues must be resolved or substantially qualified before the empirical contribution is fully established.

major comments (4)
  1. [Section 2.2, Eqs. (2) and (4)] The longitudinal and spatial losses apply element-wise logical AND (∧) and XOR (⊕) to network predictions (denoted p and s) and labels/white-matter masks. Predictions are continuous outputs, typically probabilities or softmax values, and no thresholding or differentiable Boolean surrogate is defined anywhere in the manuscript. As written, these losses are not computable, which blocks reproduction and makes the Table 5 ablation of the anatomical constraints impossible to interpret. Please state explicitly how ∧ and ⊕ act on continuous values (e.g., hard thresholding, min/max fuzzy logic, or a differentiable approximation) and justify that the resulting loss is differentiable.
  2. [Section 3.1, VAN dataset; Tables 3 and 5; Discussion] The vanishing-lesion results are evaluated on VAN, a private dataset created by inverting the timepoints of MSSEG-2 so that new lesions become disappearing lesions. The authors state in the Discussion that 'vanishing and newly forming lesions show different lesion dynamics and characteristics,' which directly undermines the validity of this time-reversal simulation as a clinical vanishing-lesion benchmark. Consequently, the VAN Dice/F1 numbers, the comparison to SegHeD on VAN, and the LesionMix ablation column for VAN do not substantiate the abstract's claim of superior performance in segmenting clinical vanishing lesions. This is an acknowledged limitation, but only in the Discussion; the abstract and contributions present vanishing-lesion segmentation as a demonstrated capability. Please temper these claims or explicitly label all VAN results as a synthetic benchmark.
  3. [Section 3.4, Table 3] In the paragraph on new-lesion segmentation, the text states that SegHeD+ achieves a Dice of 50.52% and an F1 of 55.02%, 'outperforming all SOTA and heterogeneous methods,' but the immediately following sentence says it scores 'slightly lower than the top methods tailored specifically for the new-lesion segmentation.' Table 3 confirms that MedICL (50.67) and Basaran et al. (51.06) are higher than SegHeD+ on MSSEG-2. This internal contradiction means the abstract's phrasing of 'superior performance in segmenting all, new, and vanishing lesions' is not supported for new lesions. Please correct the wording in the text and align the abstract and conclusion with the actual results.
  4. [Section 2.2, volumetric constraint] The thresholds α_high=1.2 and α_low=0.8 are said to be determined from the longitudinal MS dataset [2], which is MS2015, and MS2015 is also one of the held-out test sets in Tables 3 and 4. If the thresholds were chosen using the full MS2015 dataset, including subjects later used for testing, the reported MS2015 improvements are circular and optimistic. Please clarify whether the thresholds were set using only the MS2015 training split or a separate validation set, and if not, provide an analysis showing that the reported MS2015 results are not inflated by this choice.
minor comments (6)
  1. [Section 3.4, Tables 3 and 4] The text reports MS2015 all-lesion Dice and F1 scores of 76.39% and 77.45%, but Table 3 shows 78.57 and Table 4 shows 79.35 for SegHeD+. Please reconcile the numbers in the text with the tables.
  2. [Section 2.1, Eq. (1)] The model input includes the first-timepoint all-lesion label y^t1_a, but the text says this input is not used for predicting s^t1_a. Please clarify how the network architecture prevents the label from leaking into the first-head prediction, since the input channel is present in Eq. (1).
  3. [Section 2.2] The notation p is used in Eqs. (2) and (4) but is never explicitly defined as the model prediction. Please define all prediction variables consistently (e.g., s versus p) at their first use.
  4. [Section 2.2] There are minor language issues, including 'segmentated lesions' in the introduction of Section 2.2 and 'we further penalises' in the spatial-constraint paragraph. These should be corrected during revision.
  5. [Section 3.2 and Section 2.2] The thresholds α_high and α_low are described as annual rates of lesion-volume change, but MSSEG-2 follow-up scans are 1 to 3 years apart. Please state explicitly whether the thresholds were adjusted for non-annual intervals in the experiments.
  6. [Global] The paper does not include a code or data availability statement. The MSSEG-2+ annotations and the VAN dataset are private, so the multi-dataset results cannot be independently reproduced; please add an explicit statement about what will be released, or describe the annotation/protocol in sufficient detail for replication.

Circularity Check

1 steps flagged · score 4.0 of 10

Vanishing-lesion benchmark is a time-reversal rename of MSSEG-2 new-lesion labels; all/new results remain independently grounded.

  1. renaming known result [Section 3.1 (VAN dataset definition), Tables 3-5, Section 4 Discussion]
    "VAN is created by simulating vanishing lesions by the inversion of timepoints in the MSSEG-2 dataset. Consequently, new lesions at the second timepoint are transformed into lesions that disappear from the first timepoint."

    The VAN ground-truth is constructed from MSSEG-2 new-lesion annotations by swapping the two timepoints, so vanishing-lesion labels are exactly the new-lesion masks under time reversal. The VAN Dice/F1 results (43.84/46.02) and the LesionMix ablation gain on VAN therefore evaluate a time-reversed new-lesion task, not clinical disappearance. The abstract and contributions present vanishing-lesion segmentation as a demonstrated capability, but the label definition makes the benchmark coextensive with the new-lesion benchmark. The Discussion concedes that 'Vanishing and newly forming lesions show different lesion dynamics and characteristics [35]', undermining the time-reversal symmetry.

full rationale

The all-lesion and new-lesion claims are grounded in public benchmarks (MS2015, MS2016, MSSEG-2) and do not reduce to fitted parameters or to the authors' prior work; the main architectural contribution is open to independent falsification on those tasks. The one significant circular/renaming element is the vanishing-lesion benchmark (VAN), which is generated by time-inverting MSSEG-2 new-lesion annotations, making the reported vanishing-lesion performance a time-reversed restatement of the new-lesion task. This affects one of the three headline tasks and is acknowledged in the Discussion, so the overall central claim retains independent content for all/new lesion segmentation, but the vanishing-lesion result should not be read as clinical validation. No other load-bearing self-citation or fitted-input-as-prediction step was found.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The method depends on several modeling choices that are either fitted to the evaluation data or constructed by the authors. The volumetric thresholds come from one of the test datasets. The vanishing-lesion task is built from a synthetic time-reversal assumption. The loss functions rely on an unspecified implementation of logical operations on continuous outputs. These are the main ledger entries beyond the standard deep learning toolkit.

free parameters (3)
  • α_high = 1.2
    Upper bound for annual lesion volume change in the volumetric constraint LVol. Set after examining the largest lesion volume change in the MS2015 dataset (Section 2.2), a dataset also used for evaluation.
  • α_low = 0.8
    Lower bound for annual lesion volume change. Set from the same MS2015 examination and cited literature (Section 2.2).
  • λ_L, λ_V, λ_S = 2, 1, 1
    Weights for the longitudinal, volumetric, and spatial losses, set empirically for training (Section 3.2).
assumptions (5)
  • domain assumption Two timepoints suffice to represent longitudinal MS data; missing second timepoints are replaced by duplicating the first.
    Section 2.1 states the model handles a maximum of two timepoints and sets xt2 = xt1 for cross-sectional data.
  • domain assumption SynthSeg white matter masks are reliable even in the presence of lesions.
    Section 2.1 uses SynthSeg to generate xt2_wm, which feeds the spatial constraint and model input.
  • ad hoc to paper Volume change thresholds α_high=1.2 and α_low=0.8 estimated from MS2015 generalize across all datasets and lesion types.
    Section 2.2 derives thresholds from one dataset and applies them to all datasets; sensitivity study only checks these two settings.
  • ad hoc to paper Disappearance of a lesion is the exact time-reversal of its appearance, validating the timepoint-inversion construction of VAN.
    Section 3.1 builds VAN by inverting MSSEG-2 timepoints, despite the Discussion noting new and vanishing lesions have different dynamics.
  • ad hoc to paper Logical AND/XOR operations in the constraint losses can be applied to continuous network outputs (or a differentiable surrogate exists).
    Eqs. 2 and 4 define losses with logical operators but do not specify binarization or a surrogate, leaving the losses undefined as written.
invented entities (1)
  • VAN dataset (simulated vanishing lesions)
    purpose: Provides vanishing-lesion labels for training and evaluation, enabling the paper's vanishing-lesion benchmark.
    Created by swapping timepoints in MSSEG-2, so vanishing-lesion labels are synthetic. No real vanishing-lesion dataset is used; the authors acknowledge this limitation in the Discussion.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SegHeD+: Segmentation of Heterogeneous Data for Multiple Sclerosis Lesions with Anatomical Constraints and Lesion-aware Augmentation." pith.science (2026). https://pith.science/paper/DAK4UXF2

@misc{pith2026241210946,
  author       = {Pith},
  title        = {Pith review of: SegHeD+: Segmentation of Heterogeneous Data for Multiple Sclerosis Lesions with Anatomical Constraints and Lesion-aware Augmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DAK4UXF2}},
  note         = {Machine review of arXiv:2412.10946}
}
read the original abstract

Assessing lesions and tracking their progression over time in brain magnetic resonance (MR) images is essential for diagnosing and monitoring multiple sclerosis (MS). Machine learning models have shown promise in automating the segmentation of MS lesions. However, training these models typically requires large, well-annotated datasets. Unfortunately, MS imaging datasets are often limited in size, spread across multiple hospital sites, and exhibit different formats (such as cross-sectional or longitudinal) and annotation styles. This data diversity presents a significant obstacle to developing a unified model for MS lesion segmentation. To address this issue, we introduce SegHeD+, a novel segmentation model that can handle multiple datasets and tasks, accommodating heterogeneous input data and performing segmentation for all lesions, new lesions, and vanishing lesions. We integrate domain knowledge about MS lesions by incorporating longitudinal, anatomical, and volumetric constraints into the segmentation model. Additionally, we perform lesion-level data augmentation to enlarge the training set and further improve segmentation performance. SegHeD+ is evaluated on five MS datasets and demonstrates superior performance in segmenting all, new, and vanishing lesions, surpassing several state-of-the-art methods in the field.

Figures

Figures reproduced from arXiv: 2412.10946 by the authors.

Figure 1
Figure 1. Visualisation of the proposed framework. SegHeD+ learns from heterogeneous datasets varying in image and label formats. SegHeD+ takes up to four inputs, and can analyse cross-sectional and longitudinal data. Four segmentation heads provide binary-class segmentation to all lesions (red) in two timepoints, and new (green) and vanishing (dashed blue) lesions in a second timepoint. where F represents the SegHeD+ model. … view at source ↗
Figure 2
Figure 2. Example outputs of LesionMix augmentation. Red labels denote lesions which are present in the original image; green labels denote new-lesion generated in the augmented image; blue labels denote vanishing-lesions which have been inpainted from the original image. Images best viewed online. Y’ = Y − M, (9) where f(X, M) denotes the inpainting function using the fast marching algorithm [37], G denotes the Gaussian blur… view at source ↗
Figure 3
Figure 3. Visual illustration of the size of original training data and LesionMix-augmented training data. During training, we implement on-the-fly traditional data augmentation methods, including flip￾ping and rotation in three dimensions, elastic deformation, brightness adjustments (additive and multi￾plicative), and additive Gaussian noise using the batchgenerators framework7 . Five-fold cross-validation is performed, crea… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison of all-lesion (top row) and new-lesion (bottom row) segmentation performance. Yellow regions denote false positive segmentations, whereas cyan regions denote false negative segmentations. SegHeD+ exhibits fewer false segmentations. Best viewed on…
Figure 5
Figure 5. Figure 5: SegHeD+ is capable of simultaneous multi-task segmentation (Rows 3 to 6). Some tasks do not show new/vanishing-lesions predictions as they are not present at the given slice. “Not available" denotes no ground truth annotation for comparison. A: Dataset where all-lesion…
Figure 6
Figure 6. Figure 6: Estimated lesion volumes across four time points for two test subjects. SegHeD+ (blue) results in predictions that are temporally more consistent with the ground truth (black), compared to competing methods. The ρ value for each method indicates its Pearson’s correlati…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 48 canonical work pages

  1. [2]

    Longitudinal multiple sclerosis lesion segmentation: Resource and challenge,

    A. Carass, S. Roy, A. Jog, et al., “Longitudinal multiple sclerosis lesion segmentation: Resource and challenge,” NeuroImage, 2017

    work page 2017

  2. [1]

    Multiple sclerosis pathology,

    H. Lassmann, “Multiple sclerosis pathology,” Cold Spring Harbor Perspectives in Medicine, vol. 8, no. 3, 2018

    work page 2018

  3. [3]

    Objective evaluation of multiple sclerosis lesion segmentation using a data management and processing infrastructure,

    O. Commowick, A. Istace, M. Kain, et al., “Objective evaluation of multiple sclerosis lesion segmentation using a data management and processing infrastructure,” Scientific Reports, 2018

    work page 2018

  4. [4]

    MICCAI 2021 MSSEG-2 challenge quantitative results,

    O. Commowick, A. Masson, B. Combes, et al., “MICCAI 2021 MSSEG-2 challenge quantitative results,” 2021

    work page 2021

  5. [5]

    Disappearing brainstem mri lesions in multiple sclerosis (p3.353),

    S. Eckert, B. Weinstock-Guttman, C. Kolb, and D. Hojnacki, “Disappearing brainstem mri lesions in multiple sclerosis (p3.353),” Neurology, vol. 90, no. 15_supplement, p. P3.353, 2018

    work page 2018

  6. [6]

    Atro- phied brain lesion volume: A new imaging biomarker in multiple sclerosis,

    M. G. Dwyer, N. Bergsland, D. P . Ramasamy, D. Jakimovski, B. Weinstock-Guttman, and R. Zivadinov, “Atro- phied brain lesion volume: A new imaging biomarker in multiple sclerosis,” Journal of Neuroimaging , vol. 28, no. 5, pp. 490–495, 2018

    work page 2018

  7. [7]

    Multiple sclerosis: pathology of recurrent lesions,

    J. W. Prineas, R. O. Barnard, T. Revesz, et al., “Multiple sclerosis: pathology of recurrent lesions,”Brain, vol. 116, no. 3, 1993

    work page 1993

  8. [8]

    Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation,

    K. Kamnitsas, C. Ledig, V . F. Newcombe, et al., “Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation,” Medical Image Analysis, vol. 36, 2017

    work page 2017

Show all 50 references
  1. [9]

    LesionMix: A lesion-level data augmentation method for medical image segmentation,

    B. D. Basaran, W. Zhang, M. Qiao, et al., “LesionMix: A lesion-level data augmentation method for medical image segmentation,” in MICCAI DALI Workshop, 2023. SegHeD+ 19

    2023

  2. [10]

    Temporally consistent probabilistic detection of new mul- tiple sclerosis lesions in brain MRI,

    C. Elliott, D. L. Arnold, D. L. Collins, and T. Arbel, “Temporally consistent probabilistic detection of new mul- tiple sclerosis lesions in brain MRI,” IEEE Transactions on Medical Imaging, vol. 32, no. 8, 2013

    2013

  3. [11]

    Two time point MS lesion segmentation in brain MRI: An expectation- maximization framework,

    S. Jain, A. Ribbens, D. M. Sima, et al., “Two time point MS lesion segmentation in brain MRI: An expectation- maximization framework,” Frontiers in Neuroscience, vol. 10, 2016

    2016

  4. [12]

    Spatio-temporal learning from longitudinal data for multiple sclerosis lesion segmentation,

    S. Denner, A. Khakzar, M. Sajid, et al., “Spatio-temporal learning from longitudinal data for multiple sclerosis lesion segmentation,” in Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries (A. Crimi and S. Bakas, eds.), 2021

    2021

  5. [13]

    nnU-Net: a self-configuring method for deep learning-based biomed- ical image segmentation,

    F. Isensee, P . F. Jaeger, S. A. A. Kohl,et al., “nnU-Net: a self-configuring method for deep learning-based biomed- ical image segmentation,” Nature Methods, 2021

    2021

  6. [14]

    New lesion segmentation for multiple sclerosis brain images with imaging and lesion-aware augmentation,

    B. D. Basaran, P . M. Matthews, and W. Bai, “New lesion segmentation for multiple sclerosis brain images with imaging and lesion-aware augmentation,” Frontiers in Neuroscience, vol. 16, 2022

    2022

  7. [15]

    CoactSeg: Learning from heterogeneous data for new multiple sclerosis lesion segmentation,

    Y. Wu, Z. Wu, H. Shi, et al., “CoactSeg: Learning from heterogeneous data for new multiple sclerosis lesion segmentation,” in Medical Image Computing and Computer Assisted Intervention, 2023

    2023

  8. [16]

    CLIP-driven universal model for organ segmentation and tumor detection,

    J. Liu, Y. Zhang, J. Chen, et al., “CLIP-driven universal model for organ segmentation and tumor detection,” International Conference on Computer Vision, 2023

    2023

  9. [17]

    Marginal loss and exclusion loss for partially supervised multi-organ segmentation,

    G. Shi, L. Xiao, Y. Chen, and S. K. Zhou, “Marginal loss and exclusion loss for partially supervised multi-organ segmentation,” Medical Image Analysis, vol. 70, 2021

    2021

  10. [18]

    Universeg: Universal medical image segmentation,

    V . I. Butoi, J. J. G. Ortiz, T. Ma, M. R. Sabuncu, J. Guttag, and A. V . Dalca, “Universeg: Universal medical image segmentation,” in International Conference on Computer Vision, 2023

    2023

  11. [19]

    Samus: Adapting segment anything model for clinically- friendly and generalizable ultrasound image segmentation,

    X. Lin, Y. Xiang, L. Zhang, X. Yang, Z. Yan, and L. Yu, “Samus: Adapting segment anything model for clinically- friendly and generalizable ultrasound image segmentation,” arXiv: 2309.06824, 2023

    2023 arXiv

  12. [20]

    Anatomical priors in convolutional networks for unsupervised biomedical segmentation,

    A. V . Dalca, J. Guttag, and M. R. Sabuncu, “Anatomical priors in convolutional networks for unsupervised biomedical segmentation,” in Computer Vision and Pattern Recognition, 2018

    2018

  13. [21]

    White matter MS-lesion segmentation using a geometric brain model,

    M. Strumia, F. R. Schmidt, C. Anastasopoulos, et al., “White matter MS-lesion segmentation using a geometric brain model,” IEEE Transactions on Medical Imaging, vol. 35, no. 7, 2016

    2016

  14. [22]

    Segmentation of MRI head anatomy using deep volumetric networks and multiple spatial priors,

    L. Hirsch, Y. Huang, and L. C. Parra, “Segmentation of MRI head anatomy using deep volumetric networks and multiple spatial priors,” Journal of Medical Imaging, vol. 8, no. 3, 2021

    2021

  15. [23]

    CarveMix: A simple data augmentation method for brain lesion segmentation,

    X. Zhang, C. Liu, N. Ou, et al., “CarveMix: A simple data augmentation method for brain lesion segmentation,” in Medical Image Computing and Computer Assisted Intervention, 2021

    2021

  16. [24]

    A structural causal model for MR images of multiple sclerosis,

    J. C. Reinhold, A. Carass, and J. L. Prince, “A structural causal model for MR images of multiple sclerosis,” in Medical Image Computing and Computer Assisted Intervention, 2021

    2021

  17. [25]

    Subject-specific lesion generation and pseudo-healthy synthesis for multiple sclerosis brain images,

    B. D. Basaran, M. Qiao, P . M. Matthews, and W. Bai, “Subject-specific lesion generation and pseudo-healthy synthesis for multiple sclerosis brain images,” in Simulation and Synthesis in Medical Imaging, 2022

    2022

  18. [26]

    SegHeD: Segmentation of heterogeneous data for multi- ple sclerosis lesions with anatomical constraints,

    B. D. Basaran, X. Zhang, P . M. Matthews, and W. Bai, “SegHeD: Segmentation of heterogeneous data for multi- ple sclerosis lesions with anatomical constraints,” arXiv preprint arXiv:2410.01766, 2024

    2024 arXiv

  19. [27]

    SynthSeg: Segmentation of brain MRI scans of any contrast and resolu- tion without retraining,

    B. Billot, D. N. Greve, O. Puonti, et al., “SynthSeg: Segmentation of brain MRI scans of any contrast and resolu- tion without retraining,” Medical Image Analysis, vol. 86, 2023

    2023

  20. [28]

    V-net: Fully convolutional neural networks for volumetric medical image segmentation,

    F. Milletari, N. Navab, and S.-A. Ahmadi, “V-net: Fully convolutional neural networks for volumetric medical image segmentation,” in International Conference on 3D Vision, 2016

    2016

  21. [29]

    Evaluation of the statistical detection of change algorithm for screening patients with MS with new lesion activity on longitudinal brain MRI,

    M. Homssi, E. Sweeney, E. Demmon, et al. , “Evaluation of the statistical detection of change algorithm for screening patients with MS with new lesion activity on longitudinal brain MRI,” American Journal of Neuroradi- ology, vol. 44, no. 6, 2023

    2023

  22. [30]

    Quantitative assessment of MRI lesion load in monitoring the evolution of multiple sclerosis,

    M. Filippi, M. A. Horsfield, P . S. Tofts, et al., “Quantitative assessment of MRI lesion load in monitoring the evolution of multiple sclerosis,” Brain, vol. 118, no. 6, 1995

    1995

  23. [31]

    Imaging biomarkers in multiple sclerosis,

    M. Filippi and F. Agosta, “Imaging biomarkers in multiple sclerosis,” Journal of Magnetic Resonance Imaging , vol. 31, no. 4, pp. 770–788, 2010

    2010

  24. [32]

    Predictive mri biomarkers in ms—a critical review,

    V . E. Tiu, I. Enache, C. A. Panea, C. Tiu, and B. O. Popescu, “Predictive mri biomarkers in ms—a critical review,” Medicina, vol. 58, no. 3, 2022

    2022

  25. [33]

    Slowly eroding lesions in multiple sclerosis,

    V . Sethi, G. Nair, M. Absinta, et al. , “Slowly eroding lesions in multiple sclerosis,” Multiple Sclerosis Journal , vol. 23, no. 3, 2017. 20 B.D. Basaran et al

    2017

  26. [34]

    Atrophied brain T2 lesion volume at MRI is associated with disability progression and conversion to secondary progressive multiple sclerosis,

    A. V . Genovese, J. Hagemeier, N. Bergsland,et al., “Atrophied brain T2 lesion volume at MRI is associated with disability progression and conversion to secondary progressive multiple sclerosis,” Radiology, vol. 293, no. 2, 2019

    2019

  27. [35]

    Correlations between monthly enhanced mri lesion rate and changes in t2 lesion volume in multiple sclerosis,

    P . D. Molyneux, M. Filippi, F. Barkhof, et al. , “Correlations between monthly enhanced mri lesion rate and changes in t2 lesion volume in multiple sclerosis,” Annals of Neurology, vol. 43, no. 3, 1998

    1998

  28. [36]

    Curriculum learning,

    Y. Bengio, J. Louradour, R. Collobert, and J. Weston, “Curriculum learning,” in International Conference on Ma- chine Learning, 2009

    2009

  29. [37]

    An image inpainting technique based on the fast marching method,

    A. Telea, “An image inpainting technique based on the fast marching method,” Journal of Graphics Tools, 2004

    2004

  30. [38]

    An optimized blockwise nonlocal means denoising filter for 3-d magnetic resonance images,

    P . Coupe, P . Yger, S. Prima, P . Hellier, C. Kervrann, and C. Barillot, “An optimized blockwise nonlocal means denoising filter for 3-d magnetic resonance images,” IEEE Transactions on Medical Imaging, vol. 27, no. 4, 2008

    2008

  31. [39]

    Block-matching strategies for rigid registration of multimodal medical images,

    O. Commowick, N. Wiest-Daesslé, and S. Prima, “Block-matching strategies for rigid registration of multimodal medical images,” in IEEE International Symposium on Biomedical Imaging, pp. 700–703, 2012

    2012

  32. [40]

    volbrain: An online mri brain volumetry system,

    J. V . Manjón and P . Coupé, “volbrain: An online mri brain volumetry system,” Frontiers in Neuroinformatics , vol. 10, 2016

    2016

  33. [41]

    N4itk: Improved n3 bias correction,

    N. J. Tustison, B. B. Avants, P . A. Cook, Y. Zheng, A. Egan, P . A. Yushkevich, and J. C. Gee, “N4itk: Improved n3 bias correction,” IEEE Transactions on Medical Imaging, vol. 29, no. 6, 2010

    2010

  34. [42]

    Multi-Atlas Skull-Stripping,

    J. Doshi, G. Erus, Y. Ou, B. Gaonkar, and C. Davatzikos, “Multi-Atlas Skull-Stripping,” Academic Radiology , vol. 20, no. 12, pp. 1566–1576, 2013

    2013

  35. [43]

    Itk-snap: An interactive tool for semi-automatic segmentation of multi- modality biomedical images,

    P . A. Yushkevich, Y. Gao, and G. Gerig, “Itk-snap: An interactive tool for semi-automatic segmentation of multi- modality biomedical images,” in International Conference of the IEEE Engineering in Medicine and Biology Society , 2016

    2016

  36. [44]

    Unbiased nonlinear average age-appropriate brain templates from birth to adulthood,

    V . Fonov, A. Evans, R. McKinstry, et al., “Unbiased nonlinear average age-appropriate brain templates from birth to adulthood,” NeuroImage, 2009

    2009

  37. [45]

    Longitudinal detection of new ms lesions using deep learning,

    R. A. Kamraoui, B. Mansencal, J. V . Manjon, and P . Coupé, “Longitudinal detection of new ms lesions using deep learning,” Frontiers in Neuroimaging, vol. 1, 2022

    2022

  38. [46]

    nnFormer: Volumetric medical image segmentation via a 3D transformer,

    H.-Y. Zhou, J. Guo, Y. Zhang, et al., “nnFormer: Volumetric medical image segmentation via a 3D transformer,” IEEE Transactions on Image Processing, 2023

    2023

  39. [47]

    UNETR: Transformers for 3D medical image segmentation,

    A. Hatamizadeh, Y. Tang, V . Nath, et al., “UNETR: Transformers for 3D medical image segmentation,” in Pro- ceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2022

    2022

  40. [48]

    Segmentation of new MS lesions with tiramisu and 2.5D stacked slices,

    H. Zhang, H. Li, and I. Oguz, “Segmentation of new MS lesions with tiramisu and 2.5D stacked slices,”MSSEG- 2 challenge proceedings: Multiple sclerosis new lesions segmentation challenge using a data management and processing infrastructure, vol. 61, 2021

    2021

  41. [49]

    TransBTS: Multimodal brain tumor segmentation using transformer,

    W. Wang, C. Chen, M. Ding, H. Yu, S. Zha, and J. Li, “TransBTS: Multimodal brain tumor segmentation using transformer,” in Medical Image Computing and Computer-Assisted Intervention, 2021

    2021

  42. [50]

    Transunet: Transformers make strong encoders for medical image segmentation,

    J. Chen, Y. Lu, Q. Yu, X. Luo, E. Adeli, et al., “Transunet: Transformers make strong encoders for medical image segmentation,” in ICML Interpretable Machine Learning in Healthcare Workshop, 2021

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.