Pith. sign in

REVIEW 2 major objections 5 minor 51 references

MedSAMix: A Training-Free Model Merging Approach for Medical Image Segmentation

T0 review · 2 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read MedSAMix claims that a search over layer-wise merge recipes, using only calibration images, turns generalist SAM plus its fine-tuned medical variants into a segmenter that beats every single model on 25 medical tasks.

desk verdict A useful, well-defined model-merging recipe for SAM-based medical segmentation, but the headline gains overstate the method because MedSAMix sees labeled calibration data while baselines are zero-shot. read the letter →

arxiv 2508.11032 v1 pith:RAXKZIGX submitted 2025-08-14 cs.CV

Yanwu Yang , Guinan Su , Jiesi Hu , Francesco Sammarco , Jonas Geiping , Thomas Wolfers This is my paper · ORCID
classification cs.CV
keywords medicalimagesegmentationmodelmergingSegmentAnythingtraining-freeBayesianoptimizationlayer-wisemulti-objectivezero-shot
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 sets out to show that the best medical image segmenter may be a blend of existing models rather than any single one of them. MedSAMix is a training-free procedure that searches over layer-wise merge recipes, choosing per layer group among task arithmetic, TIES, linear combination, and spherical interpolation, using only a calibration split of the target data as reward. The paper reports that the resulting merged model beats the generalist SAM where fine-tuned medical specialists underperform, and beats the specialists where they are strong, with average Dice improvements of 6.67% on specialized tasks and 4.37% across a 25-task multi-task evaluation. A reader should care because the gains come without retraining, gradient updates, or data sharing, so separately trained clinical models could be combined into one stronger universal segmenter.

What carries the argument

The carrying mechanism is a layer-wise, searched merge recipe over the SAM architecture. The image encoder, prompt encoder, and mask decoder are partitioned into layer groups (with granularity itself searchable), and for each group the search chooses one of four merging operations—task arithmetic, TIES (trim, elect sign, and merge), linear combination, or spherical linear interpolation (SLERP)—plus method-specific hyperparameters. A Bayesian optimizer with a random-forest surrogate proposes configurations, and the reward is segmentation Dice on a calibration set, either for one task or scalarized across tasks to find a Pareto front. This converts model merging from a manually fixed hyperparameter choice into an automatic, per-task configuration search.

What would settle it

On a representative set of medical scans, compute the loss barrier along linear interpolation between SAM and MedSAM, and between SAM and MedicoSAM; if the barrier is steep, with loss well above both endpoint models, the shared-basin assumption is false and the reported merged-model Dice gains should not reproduce.

Watch

Extended reading notes

Core claim

MedSAMix claims that SAM, MedSAM, and MedicoSAM—all initialized from the same SAM weights—converge into a shared loss basin, so their layer weights can be mixed by arithmetic operations, and that the best mixture is not uniform but varies layer by layer and task by task. The method therefore treats the merge recipe itself as an optimization variable: each layer group can use a different merging operation and different hyperparameters, and a Bayesian search over these recipes is driven by Dice on a calibration set. For a single task, the search finds a recipe that outperforms both the best specialist model and the generalist base. For universal use, a multi-objective version of the same search produces a Pareto-optimal recipe that the paper reports as beating all baselines on 18 of 25 tasks and the best competing universal model by 4.37% average Dice. The authors interpret these results as evidence that fine-tuning bias can be corrected at the model level, without additional data or training.

Load-bearing premise

The load-bearing premise is that after fine-tuning from the same initial SAM weights, SAM, MedSAM, and MedicoSAM end up in compatible regions of parameter space, so mixing their layer weights produces a working model rather than a broken one; if that premise fails, the search has nothing useful to combine.

Editorial extensions

If this is right

  • A clinic can fine-tune a local specialist, merge it with a generalist foundation model using a small calibration set, and obtain a better segmenter without sharing data or retraining.
  • For a specific clinical task, the single-task search yields a merged model that outperforms both the best specialist and the generalist base on that task.
  • For universal segmentation, the multi-task Pareto search yields one merged model that beats the strongest baseline on 18 of 25 tasks and the best universal competitor by 4.37% average Dice.
  • Fixed merging strategies applied uniformly are inconsistent across tasks, so the automatic layer-wise search is what carries the reported gains.
  • The procedure is training-free and inference-only, with 120 search trials completed in roughly 70 minutes on two 8GB GPUs in the reported single-task setting.

Reading between the lines

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

  • The same searched layer-wise mixing could rescue any SAM-family specialist that underperforms its base on a niche, provided the variant keeps the base architecture and shared initialization.
  • A testable extension is to run the search on a held-out dataset absent from the 25 tasks, or with corrupted calibration labels; if gains vanish, the method is mostly fitting the calibration distribution rather than discovering transferable skill combinations.
  • The Pareto front of merge recipes could be used as a per-prediction model-selection mechanism, choosing among several merged checkpoints for different inputs at inference time.
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

2 major / 5 minor

Summary. The manuscript introduces MedSAMix, a training-free layer-wise model merging framework for SAM-based medical image segmentation. Given a base SAM model and two fine-tuned variants (MedSAM and MedicoSAM) that share the same architecture, MedSAMix defines a search space of per-layer-group merging operations (TIES, task arithmetic, linear combination, SLERP) and uses SMAC Bayesian optimization with either a single-task or a multi-task (ParEGO) objective to discover merging configurations. The objective is evaluated on a 20% calibration split of each dataset, and the resulting merged model is evaluated on a held-out 80% test split across 25 segmentation tasks. The authors report improvements of 6.67 percentage points over MedicoSAM for single-task merging and 4.37 percentage points for multi-task merging, and claim consistent improvements over all baselines across the 25 tasks.

Significance. If the evaluation protocol is sound, the contribution is significant: the paper demonstrates a practical, gradient-free way to combine existing public SAM checkpoints that can improve both specialized and general medical segmentation performance, and the 25-task benchmark with a genuinely held-out 80% test split is a strength. The search-space formulation is clear, the use of SMAC is appropriate, and the authors provide code and weights for reproducibility. However, the primary quantitative claim currently rests on an asymmetric evaluation in which MedSAMix uses labeled calibration data for configuration search while all baselines, including the fixed merging methods, are evaluated zero-shot. As a result, the magnitude of the improvement attributable specifically to model merging is not yet established, and the multi-task generalization claim is weakened because the tasks used for Pareto search are included among the evaluation tasks.

major comments (2)
  1. [Section 4, 'Evaluations' paragraph, and Table 2] The comparison is asymmetric. The 20% calibration split is explicitly used to search for MedSAMix's merging configuration, whereas SAM, MedSAM, MedicoSAM, and the fixed merging baselines (TIES, TA, Linear, SLERP) are evaluated zero-shot without access to that data. Consequently, the reported gains—6.67 percentage points over MedicoSAM and the superiority over the fixed merging baselines—conflate the effect of the merging operation with the effect of access to labeled calibration samples and per-task automatic search. To isolate the benefit of the layer-wise search, please add calibration-exposed controls, for example a few-shot fine-tune or prompt-only adaptation of MedSAM on the same 20% split, or a grid-selected fixed merging recipe using the same calibration set. Without such controls, the headline improvement cannot be attributed to model merging itself.
  2. [Section 4, 'Objective and Optimizer', and Section 5.2] The multi-task generalization claim is not fully supported because the eight tasks used for the Pareto-front search (optic disk, tumor, vascular, lateral ventricle, mice-lung, cardiac, nasal pharynx, prostate) are a subset of the 25 evaluation tasks. The MedSAMix-M average in Table 1 is therefore computed partly on tasks that were in-sample for configuration selection, so the reported 4.37% gain over the second-best model does not demonstrate generalization to genuinely unseen tasks. Please report results separately on the 17 tasks not used in the search, or use a leave-some-out protocol over search-task subsets, so that the multi-task regime can be evaluated on truly held-out tasks.
minor comments (5)
  1. [Abstract and Table 2] The numbers '6.67%' and '4.37%' are described as 'improvements' but correspond to absolute percentage-point differences in Dice (79.64 vs. 72.97 and 77.34 vs. 72.97). Please state 'percentage points' or report relative improvements to avoid ambiguity.
  2. [Section 4, 'Evaluations', and Table 1] Table 4 reports three runs of MedSAMix-M with some variation (e.g., 77.14, 77.40, 77.33 average), but Table 1 reports a single run. Please clarify whether Table 1 is one selected run or an average, and consider reporting variance for the main results.
  3. [Table 5] The merged architecture is reported only for MedSAMix-M. Since MedSAMix-S is a per-task configuration for each of the 25 tasks, the authors should either show an example or explicitly state that the per-task configurations are provided in the released code.
  4. [Section 5.4, Figure 4] The sensitivity analysis in Figure 4 does not show error bars or repeated-run variability, even though Section B.1 demonstrates that optimization runs have some variance. Adding error bars would make the claims about insensitivity to granularity and the optimal number of tasks more robust.
  5. [Section 6, Discussion] The statement that MedSAMix works 'without the need for additional data or retraining' should be qualified: the search does use the 20% labeled calibration split. This is distinct from retraining, but it is still additional data access relative to the zero-shot baselines.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: MedSAMix's merge recipes are optimized on calibration data and measured on held-out test splits; the only self-citations are background and non-load-bearing.

full rationale

The paper's reported predictions are held-out evaluations, not fitted values. The text states: 'For each dataset, 80% is reserved for testing... The remaining 20% serves... as a calibration set for MedSAMix during the merging search' (Section 4). The single-task and multi-task objectives (Eqs. 3-5) are computed on the calibration split, while the Dice values in Tables 1, 2, and 4 are reported on the 80% test split. No parameter is fitted to the test split, and no reported accuracy is defined as the optimization objective by construction. The basin-compatibility assumption is attributed to Neyshabur et al. (2020), an external source, not to the authors' own prior work. The self-citations (Su and Geiping 2025; Su et al. 2025; Hu et al. 2024, 2025) motivate the general model-merging and in-context-learning paradigm but are not invoked as a theorem or fitted parameter that forces MedSAMix's configuration or its measured improvement. The calibration-label asymmetry between MedSAMix and zero-shot baselines is a legitimate evaluation-fairness concern, but it is a threat to the comparison's validity, not a circularity by construction. No circular step was found.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The ledger shows the method has no invented physical entities and does not claim parameter-free prediction. All gains are obtained by fitting merge configurations to labeled calibration data, which is the method itself but means the central numbers depend on the calibration protocol and the chosen search hyperparameters.

free parameters (6)
  • Layer granularity g_enc, g_prompt, g_dec = searched in [1,4]
    Defines how layers are grouped for merging; sensitivity analysis shows multi-task performance depends on it (Sec 5.4).
  • Merge method per layer group = one of TIES, TA, SLERP, Linear per group
    The searched choice per layer group; the resulting MedSAMix-M recipe in Table 5 mixes multiple methods.
  • TIES retain ratio and scaling factor = chosen from ranges in Sec A.6
    Hyperparameters of TIES-merging, tuned by SMAC on calibration sets.
  • TA scaling factor = range 0.0 to 0.5 per Sec A.6
    Scaling factor for task-vector addition, tuned per layer group.
  • SLERP interpolation parameter t = values in [0,1], e.g., Table 5
    Interpolation coefficient between two models, tuned per layer group.
  • Number of search tasks for multi-objective mode = 8
    Chosen by hand from 25 tasks; sensitivity analysis shows best at 8, a post-hoc selection (Sec 5.4).
assumptions (5)
  • domain assumption Fine-tuned models initialized from the same pre-trained weights converge to similar loss basins, so merging their weights combines capabilities.
    Invoked in Section 1 and 2.2 to justify model merging; if the basins are incompatible, the merged model could be worse.
  • domain assumption SAM, MedSAM, and MedicoSAM share an identical architecture and can be merged layer-wise.
    Stated in Section 3.2 and A.2; architectural differences would break the parameter averaging.
  • domain assumption The 20% calibration set is representative of the 80% test distribution and provides ground-truth masks for computing Dice rewards.
    Used in the evaluation protocol (Section 4); if calibration and test are misaligned, the selected recipe would not transfer.
  • standard math Random-forest surrogate (SMAC) with 120 or 200 trials finds a near-optimal merge recipe in the defined search space.
    Standard Bayesian optimization assumption; the paper does not compare against random search to validate it.
  • domain assumption Pareto front selection via ParEGO identifies configurations that generalize across the 8 search tasks to the remaining 17.
    Used for MedSAMix-M; the paper assumes the 8 representative tasks cover the diversity of all 25.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MedSAMix: A Training-Free Model Merging Approach for Medical Image Segmentation." pith.science (2026). https://pith.science/paper/RAXKZIGX

@misc{pith2026250811032,
  author       = {Pith},
  title        = {Pith review of: MedSAMix: A Training-Free Model Merging Approach for Medical Image Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RAXKZIGX}},
  note         = {Machine review of arXiv:2508.11032}
}
read the original abstract

Universal medical image segmentation models have emerged as a promising paradigm due to their strong generalizability across diverse tasks, showing great potential for a wide range of clinical applications. This potential has been partly driven by the success of general-purpose vision models such as the Segment Anything Model (SAM), which has inspired the development of various fine-tuned variants for medical segmentation tasks. However, fine-tuned variants like MedSAM are trained on comparatively limited medical imaging data that often suffers from heterogeneity, scarce annotations, and distributional shifts. These challenges limit their ability to generalize across a wide range of medical segmentation tasks. In this regard, we propose MedSAMix, a training-free model merging method that integrates the strengths of both generalist models (e.g., SAM) and specialist models (e.g., MedSAM) for medical image segmentation. In contrast to traditional model merging approaches that rely on manual configuration and often result in suboptimal outcomes, we propose a zero-order optimization method to automatically discover optimal layer-wise merging solutions. Furthermore, for clinical applications, we develop two regimes to meet the demand of domain-specificity and generalizability in different scenarios by single-task optimization and multi-objective optimization respectively. Extensive evaluations on 25 medical segmentation tasks demonstrate that MedSAMix effectively mitigates model bias and consistently improves performance in both domain-specific accuracy and generalization, achieving improvements of 6.67% on specialized tasks and 4.37% on multi-task evaluations.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 27 canonical work pages

  1. [1]

    Evolutionary optimization of model merging recipes

    Takuya Akiba, Makoto Shing, Yujin Tang, Qi Sun, and David Ha. Evolutionary optimization of model merging recipes. arXiv preprint arXiv:2403.13187, 2024

    arXiv 2024

  2. [2]

    Evolutionary optimization of model merging recipes

    Takuya Akiba, Makoto Shing, Yujin Tang, Qi Sun, and David Ha. Evolutionary optimization of model merging recipes. Nature Machine Intelligence, 7 0 (2): 0 195--204, 2025

    work page 2025

  3. [3]

    Catastrophic forgetting in deep learning: A comprehensive taxonomy

    Everton L Aleixo, Juan G Colonna, Marco Cristo, and Everlandio Fernandes. Catastrophic forgetting in deep learning: A comprehensive taxonomy. arXiv preprint arXiv:2312.10549, 2023

    arXiv 2023

  4. [4]

    Medmerge: merging models for effective transfer learning to medical imaging tasks

    Ibrahim Almakky, Santosh Sanjeev, Anees Ur Rehman Hashmi, Mohammad Areeb Qazi, Hu Wang, and Mohammad Yaqub. Medmerge: merging models for effective transfer learning to medical imaging tasks. arXiv preprint arXiv:2403.11646, 2024

    arXiv 2024

  5. [5]

    The medical segmentation decathlon

    Michela Antonelli, Annika Reinke, Spyridon Bakas, Keyvan Farahani, Annette Kopp-Schneider, Bennett A Landman, Geert Litjens, Bjoern Menze, Olaf Ronneberger, Ronald M Summers, et al. The medical segmentation decathlon. Nature communications, 13 0 (1): 0 4128, 2022

    work page 2022

  6. [6]

    Medicosam: Towards foundation models for medical image segmentation

    Anwai Archit, Luca Freckmann, and Constantin Pape. Medicosam: Towards foundation models for medical image segmentation. arXiv preprint arXiv:2501.11734, 2025

    arXiv 2025

  7. [7]

    Random forests

    Leo Breiman. Random forests. Machine learning, 45: 0 5--32, 2001

    2001

  8. [8]

    Universeg: Universal medical image segmentation

    Victor Ion Butoi, Jose Javier Gonzalez Ortiz, Tianyu Ma, Mert R Sabuncu, John Guttag, and Adrian V Dalca. Universeg: Universal medical image segmentation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 21438--21451, 2023

    2023

Show all 51 references
  1. [9]

    Neuralizer: General neuroimage analysis without re-training

    Steffen Czolbe and Adrian V Dalca. Neuralizer: General neuroimage analysis without re-training. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6217--6230, 2023

    2023

  2. [10]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv...

    2010 arXiv

  3. [11]

    Parameter competition balancing for model merging

    Guodong Du, Junlin Lee, Jing Li, Runhua Jiang, Yifei Guo, Shuyang Yu, Hanting Liu, Sim K Goh, Ho-Kin Tang, Daojing He, et al. Parameter competition balancing for model merging. Advances in Neural Information Processing Systems, 37: 0 84746--84776, 2024

    2024

  4. [12]

    Show and segment: Universal medical image segmentation via in-context learning

    Yunhe Gao, Di Liu, Zhuowei Li, Yunsheng Li, Dongdong Chen, Mu Zhou, and Dimitris N Metaxas. Show and segment: Universal medical image segmentation via in-context learning. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 20830--20840, 2025

    2025

  5. [13]

    Icl-sam: Synergizing in-context learning model and sam in medical image segmentation

    Jiesi Hu, Yang Shang, Yanwu Yang, Xutao Guo, Hanyang Peng, and Ting Ma. Icl-sam: Synergizing in-context learning model and sam in medical image segmentation. Medical Imaging with Deep Learning, pages 641--656, 2024

    2024

  6. [14]

    Building 3d in-context learning universal model in neuroimaging

    Jiesi Hu, Hanyang Peng, Yanwu Yang, Xutao Guo, Yang Shang, Pengcheng Shi, Chenfei Ye, and Ting Ma. Building 3d in-context learning universal model in neuroimaging. arXiv preprint arXiv:2503.02410, 2025

    2025 arXiv

  7. [15]

    Editing models with task arithmetic

    Gabriel Ilharco, Marco Tulio Ribeiro, Mitchell Wortsman, Suchin Gururangan, Ludwig Schmidt, Hannaneh Hajishirzi, and Ali Farhadi. Editing models with task arithmetic. arXiv preprint arXiv:2212.04089, 2022

    2022 arXiv

  8. [16]

    nnu-net: a self-configuring method for deep learning-based biomedical image segmentation

    Fabian Isensee, Paul F Jaeger, Simon AA Kohl, Jens Petersen, and Klaus H Maier-Hein. nnu-net: a self-configuring method for deep learning-based biomedical image segmentation. Nature methods, 18 0 (2): 0 203--211, 2021

    2021

  9. [17]

    Whitwell, Chadwick Ward, et al

    Clifford R Jack Jr, Matt A Bernstein, Nick C Fox, Paul Thompson, Gene Alexander, Danielle Harvey, Bret Borowski, Paula J Britson, Jennifer L. Whitwell, Chadwick Ward, et al. The alzheimer's disease neuroimaging initiative (adni): Mri methods. Journal of Magnetic Resonance Imag...

    2008

  10. [18]

    Dataless knowledge fusion by merging weights of language models

    Xisen Jin, Xiang Ren, Daniel Preotiuc-Pietro, and Pengxiang Cheng. Dataless knowledge fusion by merging weights of language models. arXiv preprint arXiv:2212.09849, 2022

    2022 arXiv

  11. [19]

    Measuring catastrophic forgetting in neural networks

    Ronald Kemker, Marc McClure, Angelina Abitino, Tyler Hayes, and Christopher Kanan. Measuring catastrophic forgetting in neural networks. In Proceedings of the AAAI conference on artificial intelligence, volume 32, 2018

    2018

  12. [20]

    Segment anything

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. Segment anything. In Proceedings of the IEEE/CVF international conference on computer vision, pages 4015--4026, 2023

    2023

  13. [21]

    Parego: A hybrid algorithm with on-line landscape approximation for expensive multiobjective optimization problems

    Joshua Knowles. Parego: A hybrid algorithm with on-line landscape approximation for expensive multiobjective optimization problems. IEEE transactions on evolutionary computation, 10 0 (1): 0 50--66, 2006

    2006

  14. [22]

    Domain generalization for medical imaging classification with linear-dependency regularization

    Haoliang Li, YuFei Wang, Renjie Wan, Shiqi Wang, Tie-Qiang Li, and Alex Kot. Domain generalization for medical imaging classification with linear-dependency regularization. Advances in neural information processing systems, 33: 0 3118--3129, 2020

    2020

  15. [23]

    Smac3: A versatile bayesian optimization package for hyperparameter optimization

    Marius Lindauer, Katharina Eggensperger, Matthias Feurer, Andr \'e Biedenkapp, Difan Deng, Carolin Benjamins, Tim Ruhkopf, Ren \'e Sass, and Frank Hutter. Smac3: A versatile bayesian optimization package for hyperparameter optimization. Journal of Machine Learning Research, 23...

    2022

  16. [24]

    Evaluation of prostate segmentation algorithms for mri: the promise12 challenge

    Geert Litjens, Robert Toth, Wendy Van De Ven, Caroline Hoeks, Sjoerd Kerkstra, Bram Van Ginneken, Graham Vincent, Gwenael Guillard, Neil Birbeck, Jindang Zhang, et al. Evaluation of prostate segmentation algorithms for mri: the promise12 challenge. Medical image analysis, 18 0...

    2014

  17. [25]

    Rethinking abdominal organ segmentation (raos) in the clinical scenario: A robustness evaluation benchmark with challenging cases

    Xiangde Luo, Zihan Li, Shaoting Zhang, Wenjun Liao, and Guotai Wang. Rethinking abdominal organ segmentation (raos) in the clinical scenario: A robustness evaluation benchmark with challenging cases. In International Conference on Medical Image Computing and Computer-Assisted ...

    2024

  18. [26]

    Segment anything in medical images

    Jun Ma, Yuting He, Feifei Li, Lin Han, Chenyu You, and Bo Wang. Segment anything in medical images. Nature Communications, 15 0 (1): 0 654, 2024 a

    2024

  19. [27]

    Unleashing the strengths of unlabelled data in deep learning-assisted pan-cancer abdominal organ quantification: the flare22 challenge

    Jun Ma, Yao Zhang, Song Gu, Cheng Ge, Shihao Mae, Adamo Young, Cheng Zhu, Xin Yang, Kangkang Meng, Ziyan Huang, et al. Unleashing the strengths of unlabelled data in deep learning-assisted pan-cancer abdominal organ quantification: the flare22 challenge. The Lancet Digital Hea...

    2024

  20. [28]

    u ller, Christof von Kalle, Jochen S Utikal, Verena M \

    Roman C Maron, Achim Hekler, Sarah Haggenm \"u ller, Christof von Kalle, Jochen S Utikal, Verena M \"u ller, Maria Gaiser, Friedegund Meier, Sarah Hobelsberger, Frank F Gellrich, et al. Model soups improve performance of dermoscopic skin cancer classifiers. European Journal of...

    2022

  21. [29]

    Merging models with fisher-weighted averaging

    Michael S Matena and Colin A Raffel. Merging models with fisher-weighted averaging. Advances in Neural Information Processing Systems, 35: 0 17703--17716, 2022

    2022

  22. [30]

    The multimodal brain tumor image segmentation benchmark (brats)

    Bjoern H Menze, Andras Jakab, Stefan Bauer, Jayashree Kalpathy-Cramer, Keyvan Farahani, Justin Kirby, Yuliya Burren, Nicole Porz, Johannes Slotboom, Roland Wiest, et al. The multimodal brain tumor image segmentation benchmark (brats). IEEE transactions on medical imaging, 34 0...

    1993

  23. [31]

    What is being transferred in transfer learning? Advances in neural information processing systems, 33: 0 512--523, 2020

    Behnam Neyshabur, Hanie Sedghi, and Chiyuan Zhang. What is being transferred in transfer learning? Advances in neural information processing systems, 33: 0 512--523, 2020

    2020

  24. [32]

    Dynammo: Dynamic model merging for efficient class incremental learning for medical images

    Mohammad Areeb Qazi, Ibrahim Almakky, Anees Ur Rehman Hashmi, Santosh Sanjeev, and Mohammad Yaqub. Dynammo: Dynamic model merging for efficient class incremental learning for medical images. In Annual Conference on Medical Image Understanding and Analysis, pages 245--257. Spri...

    2024

  25. [33]

    Tyche: Stochastic in-context learning for medical image segmentation

    Marianne Rakic, Hallee E Wong, Jose Javier Gonzalez Ortiz, Beth A Cimini, John V Guttag, and Adrian V Dalca. Tyche: Stochastic in-context learning for medical image segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 11159-...

    2024

  26. [34]

    A preclinical micro-computed tomography database including 3d whole body organ segmentations

    Stefanie Rosenhain, Zuzanna A Magnuska, Grace G Yamoah, Wa’ Rawashdeh, Fabian Kiessling, Felix Gremse, et al. A preclinical micro-computed tomography database including 3d whole body organ segmentations. Scientific data, 5 0 (1): 0 1--9, 2018

    2018

  27. [35]

    Fissionfusion: fast geometric generation and hierarchical souping for medical image analysis

    Santosh Sanjeev, Nuren Zhaksylyk, Ibrahim Almakky, Anees Ur Rehman Hashmi, Mohammad Areeb Qazi, and Mohammad Yaqub. Fissionfusion: fast geometric generation and hierarchical souping for medical image analysis. In International Conference on Medical Image Computing and Computer...

    2024

  28. [36]

    Ridge-based vessel segmentation in color images of the retina

    Joes Staal, Michael D Abr \`a moff, Meindert Niemeijer, Max A Viergever, and Bram Van Ginneken. Ridge-based vessel segmentation in color images of the retina. IEEE transactions on medical imaging, 23 0 (4): 0 501--509, 2004

    2004

  29. [37]

    Fine, i'll merge it myself: A multi-fidelity framework for automated model merging

    Guinan Su and Jonas Geiping. Fine, i'll merge it myself: A multi-fidelity framework for automated model merging. arXiv preprint arXiv:2502.04030, 2025

    2025 arXiv

  30. [38]

    Gptailor: Large language model pruning through layer cutting and stitching

    Guinan Su, Li Shen, Lu Yin, Shiwei Liu, Yanwu Yang, and Jonas Geiping. Gptailor: Large language model pruning through layer cutting and stitching. arXiv preprint arXiv:2506.20480, 2025

    2025 arXiv

  31. [39]

    In-context learning for medical image segmentation

    Eichi Takaya and Shinnosuke Yamamoto. In-context learning for medical image segmentation. arXiv preprint arXiv:2412.13299, 2024

    2024 arXiv

  32. [40]

    Weight averaging for neural networks and local resampling schemes

    Joachim Utans. Weight averaging for neural networks and local resampling schemes. In Proc. AAAI-96 Workshop on Integrating Multiple Learned Models. AAAI Press, pages 133--138. Citeseer, 1996

    1996

  33. [41]

    Sam-med3d: towards general-purpose segmentation models for volumetric medical images

    Haoyu Wang, Sizheng Guo, Jin Ye, Zhongying Deng, Junlong Cheng, Tianbin Li, Jianpin Chen, Yanzhou Su, Ziyan Huang, Yiqing Shen, et al. Sam-med3d: towards general-purpose segmentation models for volumetric medical images. In European Conference on Computer Vision, pages 51--67....

    2024

  34. [42]

    Seggpt: Towards segmenting everything in context

    Xinlong Wang, Xiaosong Zhang, Yue Cao, Wen Wang, Chunhua Shen, and Tiejun Huang. Seggpt: Towards segmenting everything in context. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 1130--1140, 2023

    2023

  35. [43]

    Sampling generative networks

    Tom White. Sampling generative networks. arXiv preprint arXiv:1609.04468, 2016

    2016 arXiv

  36. [44]

    Medical sam adapter: Adapting segment anything model for medical image segmentation

    Junde Wu, Ziyue Wang, Mingxuan Hong, Wei Ji, Huazhu Fu, Yanwu Xu, Min Xu, and Yueming Jin. Medical sam adapter: Adapting segment anything model for medical image segmentation. Medical image analysis, 102: 0 103547, 2025

    2025

  37. [45]

    Ties-merging: Resolving interference when merging models

    Prateek Yadav, Derek Tam, Leshem Choshen, Colin A Raffel, and Mohit Bansal. Ties-merging: Resolving interference when merging models. Advances in Neural Information Processing Systems, 36, 2024

    2024

  38. [46]

    Adamerging: Adaptive model merging for multi-task learning

    Enneng Yang, Zhenyi Wang, Li Shen, Shiwei Liu, Guibing Guo, Xingwei Wang, and Dacheng Tao. Adamerging: Adaptive model merging for multi-task learning. arXiv preprint arXiv:2310.02575, 2023

    2023 arXiv

  39. [47]

    Benchmarking the cow with the topcow challenge: Topology-aware anatomical segmentation of the circle of willis for cta and mra

    Kaiyuan Yang, Fabio Musio, Yihui Ma, Norman Juchler, Johannes C Paetzold, Rami Al-Maskari, Luciano H \"o her, Hongwei Bran Li, Ibrahim Ethem Hamamci, Anjany Sekuboyina, et al. Benchmarking the cow with the topcow challenge: Topology-aware anatomical segmentation of the circle ...

    2024

  40. [48]

    Sa-med2d-20m dataset: Segment anything in 2d medical imaging with 20 million masks

    Jin Ye, Junlong Cheng, Jianpin Chen, Zhongying Deng, Tianbin Li, Haoyu Wang, Yanzhou Su, Ziyan Huang, Jilong Chen, Lei Jiang, et al. Sa-med2d-20m dataset: Segment anything in 2d medical imaging with 20 million masks. arXiv preprint arXiv:2311.11969, 2023

    2023 arXiv

  41. [49]

    Language models are super mario: Absorbing abilities from homologous models as a free lunch

    Le Yu, Bowen Yu, Haiyang Yu, Fei Huang, and Yongbin Li. Language models are super mario: Absorbing abilities from homologous models as a free lunch. In Forty-first International Conference on Machine Learning, 2024

    2024

  42. [50]

    Nasalseg: A dataset for automatic segmentation of nasal cavity and paranasal sinuses from 3d ct images

    Yichi Zhang, Jing Wang, Tan Pan, Quanling Jiang, Jingjie Ge, Xin Guo, Chen Jiang, Jie Lu, Jianning Zhang, Xueling Liu, et al. Nasalseg: A dataset for automatic segmentation of nasal cavity and paranasal sinuses from 3d ct images. Scientific Data, 11 0 (1): 0 1329, 2024

    2024

  43. [51]

    Segmic: A universal model for medical image segmentation through in-context learning

    Jianwei Zhao, Fan Yang, Xin Li, Zicheng Jiao, Qiang Zhai, Xiaomeng Li, De Wu, Huazhu Fu, and Hong Cheng. Segmic: A universal model for medical image segmentation through in-context learning. Pattern Recognition, page 112179, 2025

    2025

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.