Pith. sign in

REVIEW 4 major objections 5 minor 51 references

MediRound demonstrates that medical image segmentation can be driven by multi-round queries that refer back to previously segmented entities, with an inference-time correction mechanism preventing errors from compounding.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-03 22:05 UTC pith:WSCLEP42

load-bearing objection Useful new task and dataset, but the multi-round evaluation may be compromised by a non-image-disjoint train/test split. the 4 major comments →

arxiv 2511.12110 v5 pith:WSCLEP42 submitted 2025-11-15 cs.CV cs.AI

MediRound: Multi-Round Entity-Level Reasoning Segmentation in Medical Images

Qinyue Tong , Ziqian Lu , Jun Liu , Rui Zuo , Zheming Lu , Yueming Jin This is my paper
classification cs.CV cs.AI
keywords multi-round segmentationmedical image segmentationentity-level reasoningreferring segmentationdialogue historyreasoning segmentationerror correctionmultimodal LLM
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper introduces a new task, Multi-Round Entity-Level Medical Reasoning Segmentation, where each query refers back to an entity segmented in a previous round. To study it, the authors built a 177K-conversation dataset of medical image dialogues with five types of cross-round entity relationships. They propose MediRound, which feeds conversation history, a crop of the previously referred mask, and the bounding box into a multimodal language model that outputs a [SEG] token whose hidden feature is decoded into a mask. The claimed result is that MediRound outperforms prior single-round medical segmentation methods and a general multi-round reasoning segmenter, with the largest gains in later rounds. The practical importance is that interactive, education-style medical image questioning becomes possible if these results hold in real use.

Core claim

The central discovery is that error accumulation in chain-like multi-round segmentation can be mitigated at inference time without retraining the end-to-end model. The paper shows that a lightweight two-MLP Judgment & Correction Mechanism, applied after training, judges the quality of the [SEG] hidden feature each round and replaces low-quality features with corrected ones before they are used as reference for the next round. With the mechanism, the model's cIoU on the MR-MedSeg validation set rises from 55.8 to 58.9 overall, and from 46.1 to 54.8 at round 8, indicating that gains concentrate precisely where error propagation should be worst. This is presented as evidence that end-to-end con

What carries the argument

The load-bearing machinery is the [SEG] hidden feature from the multimodal LLM, combined with a reference-aware input. For the current round, MediRound concatenates the text of the current query and prior dialogue, the full image, and the cropped region from the mask cited in the query, together with its bounding-box coordinates. That joint embedding goes into the LLM; the [SEG] token's last-layer hidden feature is decoded into the mask by a medical segmentation head. The Judgment & Correction Mechanism then acts as an inference-time gate: an MLP scores the feature, and if the score falls below a threshold, a second MLP rewrites the feature before decoding, so later rounds receive a cleaner

Load-bearing premise

The load-bearing premise is that the MR-MedSeg test conversations — generated from 50–80 paraphrased templates per entity relationship on one underlying medical image collection — represent the distribution of real multi-round medical reasoning dialogues; if real users phrase queries outside that template manifold, the reported gains may not transfer.

What would settle it

Collect a small set of naturally occurring multi-round clinical or educational Q&A sequences (e.g., from medical teaching files or clinician notes) and run MediRound on them, comparing against the template-based test set; if the gap over single-round baselines shrinks sharply on such queries, the template distribution explains the result.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • If correct, interactive medical education tools could let learners ask follow-up questions that refer to previous segmentations and get coherent masks.
  • The inference-time correction suggests error propagation in any chain-like referring segmentation can be partly offset without training a new model.
  • The dataset provides a fixed benchmark for comparing multi-round medical reasoners across relationship types such as organ–lesion, anatomical hierarchy, spatial, attribute, and strong inference.
  • MediRound also remains competitive on single-round referring segmentation, so the multi-round capability does not sacrifice standard utility.
  • Per-round results show that with the correction mechanism the model holds up past round six, the regime where naïve pipelines typically collapse.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The test set is generated by the same template-based pipeline as the training set, so the reported scores likely upper-bound performance on unscripted queries; a natural-language paraphrase challenge would be a more stringent test.
  • The five relationship types form a taxonomy of cross-round references that could transfer to other medical imaging settings, such as pathology or 3D volumes, where anatomical constraints differ.
  • Because the correction module is threshold-gated, logging quality scores per round and correlating them with human-perceived query difficulty would be a direct test of whether the gate measures true uncertainty.
  • The 'Human-Thinking + X' baseline protocol lacks a published prompt and human-expertise specification; documenting it would clarify what the 15-point performance gap actually measures.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces MEMR-Seg, a new task for multi-round, entity-level reasoning segmentation in medical images, and constructs MR-MedSeg, a large dataset of 177K conversations derived from SA-Med2D-20M with GPT-5-assisted template-based generation. The authors propose MediRound, an MLLM-based architecture (LLaVA-Med + MedSAM) that conditions on dialogue history, a referred-round crop, and bounding-box information, plus a Judgment & Correction Mechanism (JCM) that refines low-quality [SEG] features at inference. Experiments on MR-MedSeg report large gains over prior single-round methods and hybrid MLLM+segmentation baselines, and a single-round external evaluation on SA-Med2D-20M shows competitive performance. The central claim is that end-to-end multi-round conditioning and JCM mitigate error accumulation and enable accurate cross-round entity segmentation.

Significance. If the MR-MedSeg evaluation is valid, the paper makes a useful contribution by defining a new task, releasing a substantial dataset, and providing a reproducible baseline with a clearly specified architecture and inference procedure. The strengths are the concrete equations (Eqs. 1–2), Algorithm 1, detailed hyperparameters, and the external anchor on SA-Med2D-20M (Table 3), where MediRound reaches 62.1 Dice versus 61.2 for MediSee. However, the significance depends heavily on the integrity of the MR-MedSeg split and the representativeness of the template-generated test set. The current manuscript does not establish image-disjointness between training and test conversations, and the test distribution is generated by the same pipeline as the training distribution. These issues directly threaten the paper's main empirical claims, so the contribution is currently promising but not yet established.

major comments (4)
  1. [§3 Data Statistics] The paper states the split only at conversation level: 174,934/1,270/1,273 conversations, with 177K conversations over 118K images (~1.5 conversations per image). It does not state that the split is image-disjoint. Under a random conversation-level split, the same image and its entity masks will very likely appear in both training and test. Since MediRound receives the full input image and can memorize image–mask associations, the reported test Dice/gIoU/cIoU (Table 1) and the per-round gains (Table 2) may reflect image memorization rather than multi-round reasoning generalization. Please (a) clarify whether any image appears in both training and test, and (b) if not already done, re-run the evaluation with an image-disjoint split or report the overlap statistics and results on a disjoint subset.
  2. [§3 Step 3 / Data Statistics] The test conversations are produced by the same semi-automatic pipeline as the training conversations: 50–80 GPT-5-generated semantically equivalent templates per entity relationship, with entities manually inserted. The test set is therefore drawn from the same template manifold as the training set. The paper's practical-utility claims (§1: medical education, continuous clinical interaction) require generalization to free-form user queries and unseen phrasings. A model could achieve high scores by pattern-matching on template structure rather than by genuinely resolving cross-round entity references. Please provide an out-of-distribution evaluation (e.g., held-out template families, paraphrases written by humans, or real clinical queries) or analyze the diversity of the test set to show that template leakage is not driving the results.
  3. [§5.2 Human-Thinking baselines] The 'Human-Thinking + X' protocol is not specified. No details are given for how human experts phrase multi-round queries, what conversation history is provided, how mistakes or corrections are handled, how many human annotators participated, or how the MLLM-hybrid baselines are prompted. Since these baselines are a central part of the 'outperforms conventional medical referring segmentation methods' claim, the comparison cannot be assessed for fairness. Please report the complete protocol, including prompts, error handling, and annotator expertise, or explicitly reposition these numbers as upper-bound anchors rather than direct comparisons.
  4. [§4 Algorithm 1 / JCM training] The Quality Judgment and Correction Modules are described as lightweight MLPs, but the training objective, loss function, and data labeling for q (the quality score) are not given in the main text and are deferred to an appendix that is not included in the submission. Without this information the JCM ablation (Figure 7) and the final test improvement (58.4 vs 55.8 Dice) cannot be fully interpreted. Please include the JCM training details in the main paper or a clearly accessible appendix, and specify whether β is selected on the validation set and then fixed for the reported test results.
minor comments (5)
  1. [§4 Eq. (1)] In Eq. (1), G_v^enc(x_crop) and G_b^enc(x_box) are written as inputs to G_i, but Figure 4 suggests these are concatenated with text/image embeddings. Clarify the concatenation/embedding order in the equation or caption.
  2. [§5.5 Figure 7] The β sweep is reported on the validation set without error bars or multiple seeds. State whether β=0.6 was chosen on validation and then frozen for the test numbers in Table 1; otherwise the reported test improvements could include selection bias.
  3. [Table 1] The symbol 'T' in the table header is not defined in the main text. Add a footnote explaining what 'independent model' means and why SegLLM and MediRound are marked as such.
  4. [References] The paper cites GPT-4's technical report for GPT-5-based generation. If a specific GPT-5 model or version was used, cite the actual model or API documentation; otherwise clarify the exact version and access date.
  5. [General] Several details, including scenario definitions, data examples, and JCM training, are deferred to 'the Appendix,' but no appendix is present in the submitted version. Please include the appendix in the revised submission so that the claims can be verified.

Circularity Check

0 steps flagged

No significant circularity: the paper's central claims are empirical benchmark results, and no prediction reduces by construction to its own input.

full rationale

The paper's central claim is an empirical system/dataset paper: it builds MR-MedSeg, trains MediRound, and reports held-out segmentation metrics against prior methods. There is no derivation chain in which an output quantity is defined in terms of the quantity it is supposed to predict. Equations (1)-(2) are direct forward-model compositions; JCM (Algorithm 1) is an inference-time MLP correction whose threshold β is swept on the validation set (Figure 7), i.e., standard hyperparameter selection, not a fitted parameter renamed as a prediction. The MR-MedSeg dialogues are generated from five template families, and test conversations come from the same pipeline; this raises external-validity and possible train/test image-overlap concerns (the split is reported only at conversation level, 174,934/1,270/1,273, and no image-disjointness is stated), but that is a benchmark-integrity issue, not circularity. Reference [39] (MediSee) is prior work by overlapping authors and is used as a baseline; the reported comparison is empirical and MediRound outperforms it, so the self-citation is not load-bearing. No uniqueness theorem or ansatz is imported from the authors' own prior work. Overall, no step in the paper reduces by construction to its own input.

Axiom & Free-Parameter Ledger

3 free parameters · 6 axioms · 0 invented entities

The paper assembles existing components (LLaVA-Med, MedSAM, [SEG]-token conditioning, SegLLM-style multi-round design) and contributes a task, a template-derived dataset, and two small MLP modules. There are no derived predictions; the empirically fitted quantity is the JCM threshold β, while dataset design choices (template counts, scenario partitions) control how much of the reported gain could be template memorization. The load-bearing assumptions are domain assumptions about data quality, dialogue representativeness, and transfer of [SEG]-conditioning to 168 medical entity categories; no new physical or conceptual entities are postulated.

free parameters (3)
  • JCM quality threshold β = 0.6 (chosen on MR-MedSeg val set)
    Automated during evaluation via the Quality Judgment Module; swept over {0.4, 0.5, 0.6, 0.7} on the validation set (Figure 7). The headline JCM gains depend on this choice; at round 8 the cIoU drops from 54.8 (β=0.6) to 49.8 (β=0.7).
  • Template count per entity relationship = 50–80 GPT-5 templates per relationship
    A dataset-construction choice in §3 Step 3 that controls dialogue diversity; influences the overlap between train/test distributions and the potential for template-memorization rather than genuine reasoning.
  • Regular vs hard case partition rule = unspecified
    The test set is split into 'regular' and 'hard' cases by reasoning complexity and turn count (§5.2, Table 1), but the exact classification criterion is not given, so the per-subset numbers are not reproducible.
axioms (6)
  • domain assumption SA-Med2D-20M masks and labels are correct and complete for the 168 selected entity categories
    §3 Step 1: all images, masks, and labels originate from SA-Med2D-20M; upstream annotation errors would propagate into both training and evaluation.
  • ad hoc to paper The five template-derived scenario types cover real-world multi-round medical interactions
    Figure 2 claims the scenarios 'encompass nearly all multi-round interaction scenarios encountered in real-world medical applications,' but this is asserted without a user study or corpus of natural multi-round medical dialogues; the scenario taxonomy is invented for the paper.
  • domain assumption GPT-5 phrasing refinement and three expert inspectors yield dialogues representative of natural user queries
    §3 Step 3: dialogues are template instantiations refined by GPT-5 and filtered by three inspectors; no inter-annotator agreement or error-rate numbers are reported, and no free-form dialogue validation is given.
  • domain assumption [SEG]-token hidden features are a sufficient conditioning signal for MedSAM's decoder across 168 medical entity categories
    §4 Eq. 2: the mask is decoded from h_c alone; the LISA-style [SEG] conditioning is assumed to transfer from natural images to fine-grained medical entities.
  • domain assumption Teacher-forced training transfers to autoregressive multi-round inference
    §4 JCM paragraph: the authors explicitly state the train/test discrepancy — 'during multi-round evaluation, the trained model can only rely on its own outputs from previous rounds' — making transfer a load-bearing assumption that JCM only partially mitigates.
  • domain assumption LLaVA-Med provides adequate pretrained biomedical vision-language grounding
    §4/§5.1: the MLLM is LLaVA-Med-v1.5-Mistral-7B; the ablation in Table 4 (Frozen LLaVA-Med drops to 48.0 Dice vs 55.8 with LoRA) confirms the assumption's importance but does not validate the underlying alignment across all 9 modalities.

pith-pipeline@v1.3.0-alltime-deepseek · 14686 in / 20638 out tokens · 194471 ms · 2026-08-03T22:05:27.314803+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of MediRound: Multi-Round Entity-Level Reasoning Segmentation in Medical Images." pith.science (2026). https://pith.science/paper/WSCLEP42

@misc{pith2026251112110,
  author       = {Pith},
  title        = {Pith review of: MediRound: Multi-Round Entity-Level Reasoning Segmentation in Medical Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WSCLEP42}},
  note         = {Machine review of arXiv:2511.12110}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Despite notable progress in text-guided medical image segmentation nowadays, these methods are limited to single-round dialogues and fail to support multi-round reasoning, which is important for medical education scenarios. In this work, we introduce Multi-Round Entity-Level Medical Reasoning Segmentation (MEMR-Seg), a new task that requires generating segmentation masks through multi-round queries with entity-level reasoning, helping learners progressively develop their understanding of medical knowledge. To support this task, we construct MR-MedSeg, a large-scale dataset of 177K multi-round medical segmentation dialogues, featuring entity-based reasoning across rounds. Furthermore, we propose MediRound, an effective baseline model designed for multi-round medical reasoning segmentation. To mitigate the inherent error propagation within the chain-like pipeline of multi-round segmentation, we introduce a lightweight yet effective Judgment & Correction Mechanism during model inference. Experimental results demonstrate that our method effectively addresses the MEMR-Seg task and outperforms conventional medical referring segmentation methods. The project is available at https://github.com/Edisonhimself/MediRound.

Figures

Figures reproduced from arXiv: 2511.12110 by Jun Liu, Qinyue Tong, Rui Zuo, Yueming Jin, Zheming Lu, Ziqian Lu.

Figure 1
Figure 1. Figure 1: A demo dialogue of our proposed MediRound. Our model can comprehend user queries that refer to the mask results from [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of MR-MedSeg. Our dataset comprises five types of medical reasoning dialogues, each characterized by a specific form [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Semi-automatic pipeline for constructing MR-MedSeg dataset. The process includes three stages: entity selection, relationship [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Overview of the MediRound framework. The figure illustrates the model’s workflow in processing the fourth-round conversation, [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Illustration of how the Judgment & Correction Mech [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative comparison with different methods on medical multi-round interactive segmentation. [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Ablation study on the threshold β value in JCM. also report the segmentation results of IMIS-Net [6] and MedSAM [25] using conventional bounding box guidance. 5.4. Qualitative Analysis We present qualitative comparisons in [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

51 extracted references · 15 linked inside Pith

  1. [1]

    Gpt-4 technical report.arXiv preprint arXiv:2303.08774,

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ah- mad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report.arXiv preprint arXiv:2303.08774,

    Pith/arXiv arXiv

  2. [2]

    The medical segmentation decathlon.Nature communications, 13(1):4128, 2022

    Michela Antonelli, Annika Reinke, Spyridon Bakas, Key- van 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(1):4128, 2022. 2

    2022

  3. [3]

    Qwen technical report.arXiv preprint arXiv:2309.16609, 2023

    Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, et al. Qwen technical report.arXiv preprint arXiv:2309.16609, 2023. 6, 7, 8

    Pith/arXiv arXiv 2023

  4. [4]

    The rsna-asnr-miccai brats 2021 benchmark on brain tumor segmentation and radiogenomic classification

    Ujjwal Baid, Satyam Ghodasara, Suyash Mohan, Michel Bilello, Evan Calabrese, Errol Colak, Keyvan Farahani, Jayashree Kalpathy-Cramer, Felipe C Kitamura, Sarthak Pati, et al. The rsna-asnr-miccai brats 2021 benchmark on brain tumor segmentation and radiogenomic classification. arXiv preprint arXiv:2107.02314, 2021. 2

    Pith/arXiv arXiv 2021

  5. [5]

    Sam-med2d.arXiv preprint arXiv:2308.16184, 2023

    Junlong Cheng, Jin Ye, Zhongying Deng, Jianpin Chen, Tianbin Li, Haoyu Wang, Yanzhou Su, Ziyan Huang, Ji- long Chen, Lei Jiang, et al. Sam-med2d.arXiv preprint arXiv:2308.16184, 2023. 7

    Pith/arXiv arXiv 2023

  6. [6]

    Interactive medical image segmentation: A bench- mark dataset and baseline.arXiv preprint arXiv:2411.12814,

    Junlong Cheng, Bin Fu, Jin Ye, Guoan Wang, Tianbin Li, Haoyu Wang, Ruoyu Li, He Yao, Junren Chen, JingWen Li, et al. Interactive medical image segmentation: A bench- mark dataset and baseline.arXiv preprint arXiv:2411.12814,

    Pith/arXiv arXiv

  7. [7]

    The importance of skip connections in biomedical image segmentation

    Michal Drozdzal, Eugene V orontsov, Gabriel Chartrand, Samuel Kadoury, and Chris Pal. The importance of skip connections in biomedical image segmentation. InInterna- tional Workshop on Deep Learning in Medical Image Anal- ysis, pages 179–187. Springer, 2016. 3

    2016

  8. [8]

    Segvol: Universal and interactive volumetric medical image segmen- tation.Advances in Neural Information Processing Systems, 37:110746–110783, 2024

    Yuxin Du, Fan Bai, Tiejun Huang, and Bo Zhao. Segvol: Universal and interactive volumetric medical image segmen- tation.Advances in Neural Information Processing Systems, 37:110746–110783, 2024. 3

    2024

  9. [9]

    The kits21 chal- lenge: Automatic segmentation of kidneys, renal tumors, and renal cysts in corticomedullary-phase ct.arXiv preprint arXiv:2307.01984, 2023

    Nicholas Heller, Fabian Isensee, Dasha Trofimova, Resha Tejpaul, Zhongchen Zhao, Huai Chen, Lisheng Wang, Alex Golts, Daniel Khapun, Daniel Shats, et al. The kits21 chal- lenge: Automatic segmentation of kidneys, renal tumors, and renal cysts in corticomedullary-phase ct.arXiv preprint arXiv:2307.01984, 2023. 2

    Pith/arXiv arXiv 2023

  10. [10]

    Intracranial hemorrhage segmentation using a deep convolutional model.Data, 5(1):14, 2020

    Murtadha D Hssayeni, Muayad S Croock, Aymen D Salman, Hassan Falah Al-Khafaji, Zakaria A Yahya, and Behnaz Ghoraani. Intracranial hemorrhage segmentation using a deep convolutional model.Data, 5(1):14, 2020. 2

    2020

  11. [11]

    Cross-modal conditioned recon- struction for language-guided medical image segmentation

    Xiaoshuang Huang, Hongxiang Li, Meng Cao, Long Chen, Chenyu You, and Dong An. Cross-modal conditioned recon- struction for language-guided medical image segmentation. IEEE Transactions on Medical Imaging, 2024. 2

    2024

  12. [12]

    Towards a multimodal large language model with pixel-level insight for biomedicine

    Xiaoshuang Huang, Lingdong Shen, Jia Liu, Fangxin Shang, Hongxiang Li, Haifeng Huang, and Yehui Yang. Towards a multimodal large language model with pixel-level insight for biomedicine. InProceedings of the AAAI Conference on Artificial Intelligence, pages 3779–3787, 2025. 6, 7

    2025

  13. [13]

    Medseg-r: Reasoning segmentation in medical images with multimodal large language models

    Yu Huang, Zelin Peng, Yichen Zhao, Piao Yang, Xiaokang Yang, and Wei Shen. Medseg-r: Reasoning segmentation in medical images with multimodal large language models. arXiv preprint arXiv:2506.10465, 2025. 2, 3

    Pith/arXiv arXiv 2025

  14. [14]

    nnu-net: a self-configuring method for deep learning-based biomedical image segmen- tation.Nature methods, 18(2):203–211, 2021

    Fabian Isensee, Paul F Jaeger, Simon AA Kohl, Jens Pe- tersen, and Klaus H Maier-Hein. nnu-net: a self-configuring method for deep learning-based biomedical image segmen- tation.Nature methods, 18(2):203–211, 2021. 3

    2021

  15. [15]

    Harnessing progress in radiotherapy for global cancer control.Nature Cancer, 4(9):1228–1238,

    David A Jaffray, Felicia Knaul, Michael Baumann, and Mary Gospodarowicz. Harnessing progress in radiotherapy for global cancer control.Nature Cancer, 4(9):1228–1238,

  16. [16]

    Mmr: A large-scale benchmark dataset for multi-target and multi-granularity reasoning segmentation

    Donggon Jang, Yucheol Cho, Suin Lee, Taehyeon Kim, and Dae-Shik Kim. Mmr: A large-scale benchmark dataset for multi-target and multi-granularity reasoning segmentation. arXiv preprint arXiv:2503.13881, 2025. 3, 5

    Pith/arXiv arXiv 2025

  17. [17]

    Segment any- thing

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C Berg, Wan-Yen Lo, et al. Segment any- thing. InProceedings of the IEEE/CVF international confer- ence on computer vision, pages 4015–4026, 2023. 3

    2023

  18. [18]

    Lisa: Reasoning segmentation via large language model

    Xin Lai, Zhuotao Tian, Yukang Chen, Yanwei Li, Yuhui Yuan, Shu Liu, and Jiaya Jia. Lisa: Reasoning segmentation via large language model. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9579–9589, 2024. 2, 3

    2024

  19. [19]

    Llava-med: Training a large language- and-vision assistant for biomedicine in one day.Advances in Neural Information Processing Systems, 36:28541–28564,

    Chunyuan Li, Cliff Wong, Sheng Zhang, Naoto Usuyama, Haotian Liu, Jianwei Yang, Tristan Naumann, Hoifung Poon, and Jianfeng Gao. Llava-med: Training a large language- and-vision assistant for biomedicine in one day.Advances in Neural Information Processing Systems, 36:28541–28564,

  20. [20]

    Lvit: language meets vision transformer in medical image seg- mentation.IEEE transactions on medical imaging, 43(1): 96–107, 2023

    Zihan Li, Yunxiang Li, Qingde Li, Puyang Wang, Dazhou Guo, Le Lu, Dakai Jin, You Zhang, and Qingqi Hong. Lvit: language meets vision transformer in medical image seg- mentation.IEEE transactions on medical imaging, 43(1): 96–107, 2023. 2

    2023

  21. [21]

    Visual instruction tuning.Advances in neural information processing systems, 36:34892–34916, 2023

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning.Advances in neural information processing systems, 36:34892–34916, 2023. 3, 5

    2023

  22. [22]

    Grounding dino: Marrying dino with grounded pre-training for open-set object detection

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Qing Jiang, Chunyuan Li, Jianwei Yang, Hang Su, et al. Grounding dino: Marrying dino with grounded pre-training for open-set object detection. In European Conference on Computer Vision, pages 38–55. Springer, 2024. 7

    2024

  23. [23]

    Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101, 2017

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101, 2017. 7

    Pith/arXiv arXiv 2017

  24. [24]

    Image segmenta- tion using text and image prompts

    Timo L ¨uddecke and Alexander Ecker. Image segmenta- tion using text and image prompts. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7086–7096, 2022. 3

    2022

  25. [25]

    Segment anything in medical images.Nature Communications, 15(1):654, 2024

    Jun Ma, Yuting He, Feifei Li, Lin Han, Chenyu You, and Bo Wang. Segment anything in medical images.Nature Communications, 15(1):654, 2024. 3, 5, 7, 8 10

    2024

  26. [26]

    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 Health, 6(11):e815–e826, 2024. 2

    2024

  27. [27]

    Noninvasive assessment of organ-specific and shared pathways in multi-organ fibrosis using t1 mapping.Nature Medicine, 30(6):1749–1760, 2024

    Victor Nauffal, Marcus DR Klarqvist, Matthew C Hill, Danielle F Pace, Paolo Di Achille, Seung Hoan Choi, Joel T R¨am¨o, James P Pirruccello, Pulkit Singh, Shinwan Kany, et al. Noninvasive assessment of organ-specific and shared pathways in multi-organ fibrosis using t1 mapping.Nature Medicine, 30(6):1749–1760, 2024. 2

    2024

  28. [28]

    Medical image segmentation methods, algorithms, and applications.IETE Technical Re- view, 31(3):199–213, 2014

    Alireza Norouzi, Mohd Shafry Mohd Rahim, Ayman Al- tameem, Tanzila Saba, Abdolvahab Ehsani Rad, Amjad Rehman, and Mueen Uddin. Medical image segmentation methods, algorithms, and applications.IETE Technical Re- view, 31(3):199–213, 2014. 2

    2014

  29. [29]

    Perceptiongpt: Effectively fusing visual perception into llm

    Renjie Pi, Lewei Yao, Jiahui Gao, Jipeng Zhang, and Tong Zhang. Perceptiongpt: Effectively fusing visual perception into llm. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 27124– 27133, 2024. 3

    2024

  30. [30]

    Reasoning to attend: Try to understand how¡ seg¿ token works.arXiv preprint arXiv:2412.17741, 2024

    Rui Qian, Xin Yin, and Dejing Dou. Reasoning to attend: Try to understand how¡ seg¿ token works.arXiv preprint arXiv:2412.17741, 2024. 6, 7

    arXiv 2024

  31. [31]

    A review of medical image segmen- tation algorithms.EAI Endorsed Transactions on Pervasive Health & Technology, 7(27), 2021

    KKD Ramesh, G Kiran Kumar, K Swapna, Debabrata Datta, and S Suman Rajest. A review of medical image segmen- tation algorithms.EAI Endorsed Transactions on Pervasive Health & Technology, 7(27), 2021. 2

    2021

  32. [32]

    Glamm: Pixel grounding large multimodal model

    Hanoona Rasheed, Muhammad Maaz, Sahal Shaji, Abdel- rahman Shaker, Salman Khan, Hisham Cholakkal, Rao M Anwer, Eric Xing, Ming-Hsuan Yang, and Fahad S Khan. Glamm: Pixel grounding large multimodal model. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13009–13018, 2024. 3

    2024

  33. [33]

    Deepspeed: System optimizations enable train- ing deep learning models with over 100 billion parameters

    Jeff Rasley, Samyam Rajbhandari, Olatunji Ruwase, and Yuxiong He. Deepspeed: System optimizations enable train- ing deep learning models with over 100 billion parameters. InProceedings of the 26th ACM SIGKDD international con- ference on knowledge discovery & data mining, pages 3505– 3506, 2020. 7

    2020

  34. [34]

    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. InMedical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part III 18, pages 234–241. Springer, 2015. 2, 3

    2015

  35. [35]

    Comparing algorithms for automated vessel segmentation in computed tomography scans of the lung: the vessel12 study

    Rina D Rudyanto, Sjoerd Kerkstra, Eva M Van Rikxoort, Catalin Fetita, Pierre-Yves Brillet, Christophe Lefevre, Wen- zhe Xue, Xiangjun Zhu, Jianming Liang, Ilkay ¨Oks¨uz, et al. Comparing algorithms for automated vessel segmentation in computed tomography scans of the lung: the vessel12 study. Medical image analysis, 18(7):1217–1232, 2014. 2

    2014

  36. [36]

    Anindo Saha, Matin Hosseinzadeh, and Henkjan Huisman. End-to-end prostate cancer detection in bpmri via 3d cnns: effects of attention mechanisms, clinical priori and decou- pled false positive reduction.Medical image analysis, 73: 102155, 2021

    2021

  37. [37]

    Arnaud Arindra Adiyoso Setio, Alberto Traverso, Thomas De Bel, Moira SN Berens, Cas Van Den Bogaard, Piergiorgio Cerello, Hao Chen, Qi Dou, Maria Evelina Fantacci, Bram Geurts, et al. Validation, comparison, and combination of algorithms for automatic detection of pulmonary nodules in computed tomography images: the luna16 challenge.Medi- cal image analys...

    2017

  38. [38]

    Gemini: a family of highly capable multimodal models.arXiv preprint arXiv:2312.11805, 2023

    Gemini Team, Rohan Anil, Sebastian Borgeaud, Jean- Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, Katie Millican, et al. Gemini: a family of highly capable multimodal models.arXiv preprint arXiv:2312.11805, 2023. 6, 7, 8

    Pith/arXiv arXiv 2023

  39. [39]

    Medisee: Reasoning-based pixel-level percep- tion in medical images.arXiv preprint arXiv:2504.11008,

    Qinyue Tong, Ziqian Lu, Jun Liu, Yangming Zheng, and Zheming Lu. Medisee: Reasoning-based pixel-level percep- tion in medical images.arXiv preprint arXiv:2504.11008,

    Pith/arXiv arXiv

  40. [40]

    Llm-seg: Bridging image segmen- tation and large language model reasoning

    Junchi Wang and Lei Ke. Llm-seg: Bridging image segmen- tation and large language model reasoning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1765–1774, 2024. 2

    2024

  41. [41]

    Medical image segmentation using deep learning: A survey.IET image processing, 16(5): 1243–1267, 2022

    Risheng Wang, Tao Lei, Ruixia Cui, Bingtao Zhang, Hongy- ing Meng, and Asoke K Nandi. Medical image segmentation using deep learning: A survey.IET image processing, 16(5): 1243–1267, 2022. 2

    2022

  42. [42]

    Segllm: Multi-round reasoning segmenta- tion.arXiv preprint arXiv:2410.18923, 2024

    XuDong Wang, Shaolun Zhang, Shufan Li, Konstantinos Kallidromitis, Kehan Li, Yusuke Kato, Kazuki Kozuka, and Trevor Darrell. Segllm: Multi-round reasoning segmenta- tion.arXiv preprint arXiv:2410.18923, 2024. 2, 3, 5, 6, 7, 8

    Pith/arXiv arXiv 2024

  43. [43]

    Organ at risk segmentation in head and neck ct im- ages using a two-stage segmentation framework based on 3d u-net.IEEE Access, 7:144591–144602, 2019

    Yueyue Wang, Liang Zhao, Manning Wang, and Zhijian Song. Organ at risk segmentation in head and neck ct im- ages using a two-stage segmentation framework based on 3d u-net.IEEE Access, 7:144591–144602, 2019. 2

    2019

  44. [44]

    A learning algorithm for continually running fully recurrent neural networks.Neu- ral computation, 1(2):270–280, 1989

    Ronald J Williams and David Zipser. A learning algorithm for continually running fully recurrent neural networks.Neu- ral computation, 1(2):270–280, 1989. 6, 7

    1989

  45. [45]

    Deeplesion: automated mining of large-scale lesion annota- tions and universal lesion detection with deep learning.Jour- nal of medical imaging, 5(3):036501–036501, 2018

    Ke Yan, Xiaosong Wang, Le Lu, and Ronald M Summers. Deeplesion: automated mining of large-scale lesion annota- tions and universal lesion detection with deep learning.Jour- nal of medical imaging, 5(3):036501–036501, 2018. 2

    2018

  46. [46]

    Medreasoner: Reinforcement learning drives reasoning grounding from clinical thought to pixel-level precision.arXiv preprint arXiv:2508.08177,

    Zhonghao Yan, Muxi Diao, Yuxuan Yang, Jiayuan Xu, Kaizhou Zhang, Ruoyan Jing, Lele Yang, Yanxi Liu, Kong- ming Liang, and Zhanyu Ma. Medreasoner: Reinforcement learning drives reasoning grounding from clinical thought to pixel-level precision.arXiv preprint arXiv:2508.08177,

    arXiv

  47. [47]

    Lisa++: An improved baseline for reasoning segmentation with large language model.arXiv preprint arXiv:2312.17240, 2023

    Senqiao Yang, Tianyuan Qu, Xin Lai, Zhuotao Tian, Bohao Peng, Shu Liu, and Jiaya Jia. Lisa++: An improved baseline for reasoning segmentation with large language model.arXiv preprint arXiv:2312.17240, 2023. 2, 5

    Pith/arXiv arXiv 2023

  48. [48]

    Sa-med2d-20m dataset: Segment any- thing in 2d medical imaging with 20 million masks.arXiv preprint arXiv:2311.11969, 2023

    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 any- thing in 2d medical imaging with 20 million masks.arXiv preprint arXiv:2311.11969, 2023. 2, 3, 4, 8

    Pith/arXiv arXiv 2023

  49. [49]

    Next-chat: An lmm for chat, detection and segmenta- tion.arXiv preprint arXiv:2311.04498, 2023

    Ao Zhang, Yuan Yao, Wei Ji, Zhiyuan Liu, and Tat-Seng Chua. Next-chat: An lmm for chat, detection and segmenta- tion.arXiv preprint arXiv:2311.04498, 2023. 3, 5 11

    Pith/arXiv arXiv 2023

  50. [50]

    A foundation model for joint segmentation, detection and recognition of biomedical objects across nine modalities.Nature methods, pages 1–11,

    Theodore Zhao, Yu Gu, Jianwei Yang, Naoto Usuyama, Ho Hin Lee, Sid Kiblawi, Tristan Naumann, Jianfeng Gao, Angela Crabtree, Jacob Abel, et al. A foundation model for joint segmentation, detection and recognition of biomedical objects across nine modalities.Nature methods, pages 1–11,

  51. [51]

    Unet++: A nested u-net ar- chitecture for medical image segmentation

    Zongwei Zhou, Md Mahfuzur Rahman Siddiquee, Nima Tajbakhsh, and Jianming Liang. Unet++: A nested u-net ar- chitecture for medical image segmentation. InDeep learning in medical image analysis and multimodal learning for clin- ical decision support: 4th international workshop, DLMIA 2018, and 8th international workshop, ML-CDS 2018, held in conjunction wit...

    2018