RoboSurg-VQA: A Multimodal Benchmark for Surgical Segmentation-Aware Visual Question Answering

Chengyi Zhang; Ziyang Wang; Zi Ye

arxiv: 2605.23068 · v1 · pith:YQRDZPBHnew · submitted 2026-05-21 · 💻 cs.CV

RoboSurg-VQA: A Multimodal Benchmark for Surgical Segmentation-Aware Visual Question Answering

Chengyi Zhang , Zi Ye , Ziyang Wang This is my paper

Pith reviewed 2026-05-25 05:21 UTC · model grok-4.3

classification 💻 cs.CV

keywords surgicalbenchmarkrobosurg-vqaundervisualansweringartefactsconsistency

0 comments

The pith

RoboSurg-VQA is a new segmentation-aware VQA benchmark created by repurposing public surgical datasets with fixed clinically motivated questions and closed answer sets.

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

Surgical videos from robot-assisted procedures often suffer from smoke, blood, reflections, and occlusions that make it hard for AI to understand what is happening. Current work focuses on drawing masks around organs and tools, but doctors also ask questions about context, visibility, and quality. This paper builds a benchmark called RoboSurg-VQA that pairs each video frame with a standard set of questions covering procedure type, anatomy location, imaging view, artifacts like smoke or bleeding, and basic spatial relations. Answers are restricted to closed sets so models can be scored consistently. To create the labels at scale, the authors use AI prompting with automatic checks for validity, then have humans review for plausibility. The result is a shared test set that can evaluate how well multimodal models handle both segmentation and language understanding in surgery.

Core claim

We present RoboSurg-VQA, a segmentation-aware visual question answering (VQA) benchmark built by repurposing public surgical segmentation datasets under a shared schema. Each frame is paired with a fixed set of clinically motivated questions spanning procedure context, anatomy, imaging modality/view, surgical artefacts, image quality, and basic visibility and spatial attributes, with closed answer sets.

Load-bearing premise

That constrained prompting plus automatic validity checks followed by human auditing produces labels that are sufficiently accurate, consistent, and clinically representative without introducing systematic biases from the generation process.

Figures

Figures reproduced from arXiv: 2605.23068 by Chengyi Zhang, Ziyang Wang, Zi Ye.

**Figure 1.** Figure 1: Example dialogue in RoboSurg-VQA (single frame with fixed questions and closed-set answers). Visual Question Answering (VQA) offers a natural interface for linking perception and language [5,9], but surgical VQA differs from natural-image VQA: terminology is specialised, visual degradation is common, and clinically relevant answers often depend on fine-grained cues. Medical VQA benchmarks such as VQA-RAD … view at source ↗

**Figure 2.** Figure 2: Dataset construction workflow of RoboSurg-VQA, including data preparation, model-assisted VQA annotation with human auditing, and quality control prior to release. proposing a scalable, constraint-driven annotation and quality-control pipeline with human auditing; and (iv) providing benchmark statistics, sanity baselines, and evaluation protocols to enable reproducible comparison under realistic degradati… view at source ↗

**Figure 3.** Figure 3: Distribution of question-answer labels in the RoboSurg-VQA benchmark [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** Figure 4: Example of model-assisted VQA annotation in RoboSurg-VQA. site3: 5,848). These statistics are computed from the dataset manifest (to be released upon acceptance) and are used for all subsequent analyses in this paper. Visual inputs and mask semantics. Each query includes two vision inputs: (i) the RGB frame and (ii) a pseudocolour instrument-only segmentation mask with no organ classes. The mask provides a… view at source ↗

read the original abstract

Reliable visual understanding in robot-assisted and minimally invasive surgery (RMIS/MIS) demands more than accurate masks: in clinical practice, clinicians pose language-like questions about procedural context, visibility, artefacts, and the presence of anatomical structures and surgical instruments, often under degraded views caused by occlusion, smoke, bleeding, and specular highlights. We present \textbf{RoboSurg-VQA}, a segmentation-aware visual question answering (VQA) benchmark built by repurposing public surgical segmentation datasets under a shared schema. Each frame is paired with a fixed set of clinically motivated questions spanning procedure context, anatomy (including region), imaging modality/view, surgical artefacts, image quality, and basic visibility and spatial attributes, with closed answer sets to enable consistent evaluation. To scale annotation, we generate candidate answers via constrained prompting with automatic validity and consistency checks, followed by human auditing to improve plausibility and label consistency. We report benchmark statistics, sanity baselines, and common evaluation challenges under challenging surgical conditions. The code will be available on https://github.com/ziyangwang007/Robosurg-VQA.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

RoboSurg-VQA repurposes surgical segmentation data into a VQA benchmark with a fixed question schema, but supplies no numbers on whether the labels are accurate or consistent.

read the letter

The paper's concrete step is to take public surgical segmentation datasets and attach a shared set of closed-answer VQA questions that cover procedure context, anatomy, imaging view, artifacts, quality, and spatial attributes. The schema is applied uniformly, which is the part that looks new relative to the cited prior work. Repurposing existing masks this way avoids new data collection and keeps the questions clinically motivated, which is a reasonable design choice for the domain. The pipeline of constrained LLM prompting plus automatic checks and human auditing is described clearly enough to understand the intent. They also plan to release code and report some benchmark statistics and baselines, which gives a starting point for evaluation. The central limitation is the absence of any quantitative check on the labels themselves. No error rates, inter-rater numbers, or comparison between the audited output and fully manual labels appear in the abstract or the stress-test summary. Without those, the claim that the resulting benchmark is usable rests on an untested assumption about label quality. That gap directly affects whether downstream users can trust the closed-answer sets for model development or comparison. The work is aimed at groups already working on vision models for robot-assisted or minimally invasive surgery who need language-based test sets. A reader building or evaluating VQA systems in this area could extract the question schema and data sources as a useful reference, but would still need to run their own quality audit before relying on the labels. I would send the paper to peer review because the core idea is straightforward to evaluate and the missing validation metrics are the sort of thing referees can request and check. The authors have engaged honestly with the clinical motivation even if the current evidence for reliability is thin.

Referee Report

1 major / 0 minor

Summary. The manuscript presents RoboSurg-VQA, a segmentation-aware VQA benchmark constructed by repurposing existing public surgical segmentation datasets under a shared schema. Each frame receives a fixed set of clinically motivated questions on procedure context, anatomy/region, imaging modality/view, surgical artefacts, image quality, visibility, and spatial attributes, all with closed answer sets. Candidate labels are produced via constrained LLM prompting plus automatic validity/consistency checks, followed by human auditing. The paper reports benchmark statistics, sanity baselines, and common evaluation challenges under degraded surgical conditions.

Significance. If the labels prove reliable, the benchmark would usefully extend surgical scene understanding beyond masks to clinically relevant language queries and could support joint segmentation-VQA models. The shared schema across datasets and emphasis on artefacts/occlusion/smoke are practical strengths. The construction pipeline is a reasonable attempt at scalable annotation, but the lack of any quantitative validation of label quality makes the usability claim provisional rather than demonstrated.

major comments (1)

[Abstract and Methods] Abstract / Methods (annotation pipeline description): The central claim that RoboSurg-VQA constitutes a usable, reliable benchmark rests on the assertion that constrained prompting, automatic checks, and human auditing yield accurate, consistent, and clinically representative closed-answer labels. No quantitative evidence is supplied—error rates, inter-rater agreement, bias from the LLM step, or agreement between audited vs. fully manual labels are absent. This is load-bearing for the benchmark's claimed value.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback and for recognizing the potential utility of the shared schema and artefact-focused questions. We agree that quantitative validation of label quality is necessary to move the usability claim from provisional to demonstrated and will strengthen this aspect in revision.

read point-by-point responses

Referee: [Abstract and Methods] Abstract / Methods (annotation pipeline description): The central claim that RoboSurg-VQA constitutes a usable, reliable benchmark rests on the assertion that constrained prompting, automatic checks, and human auditing yield accurate, consistent, and clinically representative closed-answer labels. No quantitative evidence is supplied—error rates, inter-rater agreement, bias from the LLM step, or agreement between audited vs. fully manual labels are absent. This is load-bearing for the benchmark's claimed value.

Authors: We acknowledge the absence of quantitative label-quality metrics in the current manuscript. The pipeline description emphasizes constrained prompting plus automatic checks followed by human auditing to improve plausibility and consistency, but no error rates, inter-rater statistics, or LLM-vs-audited comparisons are reported. In the revised manuscript we will add a dedicated validation subsection that includes: (i) inter-rater agreement (Cohen’s kappa) computed on a randomly sampled subset of frames re-annotated by two additional clinical experts; (ii) agreement between the initial LLM candidate labels and the final audited labels; and (iii) a small-scale comparison of audited versus fully manual labels on a held-out set. These additions will supply the requested empirical evidence while preserving the scalable construction approach. revision: yes

Circularity Check

0 steps flagged

No circularity: dataset construction with no derivations or self-referential claims

full rationale

The paper presents a benchmark dataset created by repurposing public surgical segmentation datasets, generating labels via constrained LLM prompting plus automatic checks and human auditing. No equations, fitted parameters, predictions, or mathematical derivations are present. No self-citations are invoked as load-bearing for any uniqueness theorem or ansatz. The central claim (a usable VQA benchmark under a shared schema) does not reduce to its inputs by construction; it is a self-contained data curation effort. This matches the default expectation of no significant circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the repurposability of existing public segmentation datasets and the clinical relevance of the chosen question categories; no free parameters or invented entities are introduced.

axioms (2)

domain assumption Public surgical segmentation datasets can be directly repurposed under a shared schema without loss of clinical validity.
Invoked in the abstract when stating the benchmark is built by repurposing public datasets.
domain assumption Constrained prompting with automatic checks plus human auditing yields sufficiently consistent and plausible labels.
Described as the scaling method for annotation.

pith-pipeline@v0.9.0 · 5725 in / 1341 out tokens · 17983 ms · 2026-05-25T05:21:29.145463+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

24 extracted references · 9 canonical work pages · 2 internal anchors

[1]

Scientific Data12(1), 825 (2025)

Alabi, O., et al.: CholecInstanceSeg: a tool instance segmentation dataset for la- paroscopic surgery. Scientific Data12(1), 825 (2025)

2025
[2]

Medical Image Analysis p

Alabi, O., et al.: Multitask learning in minimally invasive surgical vision: a review. Medical Image Analysis p. 103480 (2025)

2025
[3]

arXiv preprint arXiv:2001.11190 (2020)

Allan, M., Kondo, S., Bodenstedt, S., Leger, S., Kadkhodamohammadi, R., Luengo, I., Fuentes, F., Flouty, E., Mohammed, A., Pedersen, M., et al.: 2018 robotic scene segmentation challenge. arXiv preprint arXiv:2001.11190 (2020)

work page arXiv 2018
[4]

2017 Robotic Instrument Segmentation Challenge

Allan, M., et al.: 2017 robotic instrument segmentation challenge. arXiv preprint arXiv:1902.06426 (2019)

work page internal anchor Pith review Pith/arXiv arXiv 2017
[5]

In: Proceedings of the IEEE International Con- ference on Computer Vision (ICCV)

Antol, S., Agrawal, A., Lu, J., Mitchell, M., Batra, D., Zitnick, C.L., Parikh, D.: VQA: Visual question answering. In: Proceedings of the IEEE International Con- ference on Computer Vision (ICCV). pp. 2425–2433 (2015)

2015
[6]

arXiv preprint arXiv:2305.11692 (2023)

Bai, L., Islam, M., Seenivasan, L., Ren, H.: Surgical-VQLA: transformer with gated vision-language embedding for visual question localized-answering in robotic surgery. arXiv preprint arXiv:2305.11692 (2023)

work page arXiv 2023
[7]

Comparative evaluation of instrument segmentation and tracking methods in minimally invasive surgery

Bodenstedt, S., et al.: Comparative evaluation of instrument segmentation and tracking methods in minimally invasive surgery. arXiv preprint arXiv:1805.02475 (2018)

work page internal anchor Pith review Pith/arXiv arXiv 2018
[8]

Scientific Data10(1), 1–8 (2023)

Carstens, M., et al.: The Dresden Surgical Anatomy Dataset for abdominal organ segmentation in surgical data science. Scientific Data10(1), 1–8 (2023)

2023
[9]

In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

Goyal, Y., Khot, T., Summers-Stay, D., Batra, D., Parikh, D.: Making the V in VQA Matter: elevating the role of image understanding in visual question answer- ing. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 6904–6913 (2017)

2017
[10]

In: International Conference on Multimedia Modeling

Jha, D., et al.: Kvasir-Instrument: diagnostic and therapeutic tool segmentation dataset in gastrointestinal endoscopy. In: International Conference on Multimedia Modeling. pp. 218–229. Springer (2021)

2021
[11]

In: Proceedings of the IEEE International Conference on Computer Vision (ICCV)

Kirillov, A., et al.: Segment Anything. In: Proceedings of the IEEE International Conference on Computer Vision (ICCV). pp. 4015–4026 (2023)

2023
[12]

In: Medical Image Computing and Computer Assisted Intervention – MICCAI

Koleilat, T., Asgariandehkordi, H., Rivaz, H., Xiao, Y.: MedCLIP-SAM: bridging text and image towards universal medical image segmentation. In: Medical Image Computing and Computer Assisted Intervention – MICCAI. pp. 643–653. Springer (2024)

2024
[13]

Scientific Data 5(1), 180251 (2018)

Lau, J.J., Gayen, S., Ben Abacha, A., Demner-Fushman, D.: A dataset of clinically generated visual questions and answers about radiology images. Scientific Data 5(1), 180251 (2018)

2018
[14]

In: 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)

Liu, B., Zhan, L.M., Xu, L., Ma, L., Yang, Y., Wu, X.M.: SLAKE: a semantically- labeled knowledge-enhanced dataset for medical visual question answering. In: 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI). pp. 1650–1654. IEEE (2021)

2021
[15]

Nature Communications15(1), 654 (2024)

Ma, J., He, Y., Li, F., Han, L., You, C., Wang, B.: Segment Anything in Medical Images. Nature Communications15(1), 654 (2024)

2024
[16]

arXiv preprint arXiv:2312.12429 (2023)

Murali, A., et al.: The Endoscapes dataset for surgical scene segmentation, object detection, and critical view of safety assessment: official splits and benchmark. arXiv preprint arXiv:2312.12429 (2023)

work page arXiv 2023
[17]

arXiv preprint arXiv:2401.00496 (2023) 10 C

Psychogyios, D., et al.: SAR-RARP50: segmentation of surgical instrumentation and action recognition on robot-assisted radical prostatectomy challenge. arXiv preprint arXiv:2401.00496 (2023) 10 C. Zhang et al

work page arXiv 2023
[18]

arXiv preprint arXiv:2407.19305 (2024)

Schmidgall, S., Cho, J., Zakka, C., Hiesinger, W.: GP-VLS: a general-purpose vision language model for surgery. arXiv preprint arXiv:2407.19305 (2024)

work page arXiv 2024
[19]

In: Medical Image Computing and Computer Assisted Intervention – MICCAI

Seenivasan, L., Islam, M., Kannan, G., Ren, H.: SurgicalGPT: end-to-end language- vision GPT for visual question answering in surgery. In: Medical Image Computing and Computer Assisted Intervention – MICCAI. pp. 281–290. Springer (2023)

2023
[20]

In: Medical Image Computing and Computer Assisted Intervention – MICCAI

Seenivasan, L., Islam, M., Krishna, A.K., Ren, H.: Surgical-VQA: visual question answering in surgical scenes using transformer. In: Medical Image Computing and Computer Assisted Intervention – MICCAI. pp. 33–43. Springer (2022)

2022
[21]

IEEE Transactions on Medical Imaging36(1), 86–97 (2017)

Twinanda, A.P., et al.: EndoNet: a deep architecture for recognition tasks on la- paroscopic videos. IEEE Transactions on Medical Imaging36(1), 86–97 (2017). https://doi.org/10.1109/TMI.2016.2593957

work page doi:10.1109/tmi.2016.2593957 2017
[22]

arXiv preprint arXiv:2501.11347 (2025)

Wang, G., et al.: EndoChat: grounded multimodal large language model for endo- scopic surgery. arXiv preprint arXiv:2501.11347 (2025)

work page arXiv 2025
[23]

IEEE Transactions on Medical Robotics and Bionics (2025)

Wang, Z., et al.: S4RoboFormer: scribble-supervised surgical robotic segmentation transformer via augmented consistency training. IEEE Transactions on Medical Robotics and Bionics (2025)

2025
[24]

npj Digital Medicine8(1), 566 (2025)

Zhao, Z., et al.: Large-vocabulary segmentation for medical images with text prompts. npj Digital Medicine8(1), 566 (2025)

2025

[1] [1]

Scientific Data12(1), 825 (2025)

Alabi, O., et al.: CholecInstanceSeg: a tool instance segmentation dataset for la- paroscopic surgery. Scientific Data12(1), 825 (2025)

2025

[2] [2]

Medical Image Analysis p

Alabi, O., et al.: Multitask learning in minimally invasive surgical vision: a review. Medical Image Analysis p. 103480 (2025)

2025

[3] [3]

arXiv preprint arXiv:2001.11190 (2020)

Allan, M., Kondo, S., Bodenstedt, S., Leger, S., Kadkhodamohammadi, R., Luengo, I., Fuentes, F., Flouty, E., Mohammed, A., Pedersen, M., et al.: 2018 robotic scene segmentation challenge. arXiv preprint arXiv:2001.11190 (2020)

work page arXiv 2018

[4] [4]

2017 Robotic Instrument Segmentation Challenge

Allan, M., et al.: 2017 robotic instrument segmentation challenge. arXiv preprint arXiv:1902.06426 (2019)

work page internal anchor Pith review Pith/arXiv arXiv 2017

[5] [5]

In: Proceedings of the IEEE International Con- ference on Computer Vision (ICCV)

Antol, S., Agrawal, A., Lu, J., Mitchell, M., Batra, D., Zitnick, C.L., Parikh, D.: VQA: Visual question answering. In: Proceedings of the IEEE International Con- ference on Computer Vision (ICCV). pp. 2425–2433 (2015)

2015

[6] [6]

arXiv preprint arXiv:2305.11692 (2023)

Bai, L., Islam, M., Seenivasan, L., Ren, H.: Surgical-VQLA: transformer with gated vision-language embedding for visual question localized-answering in robotic surgery. arXiv preprint arXiv:2305.11692 (2023)

work page arXiv 2023

[7] [7]

Comparative evaluation of instrument segmentation and tracking methods in minimally invasive surgery

Bodenstedt, S., et al.: Comparative evaluation of instrument segmentation and tracking methods in minimally invasive surgery. arXiv preprint arXiv:1805.02475 (2018)

work page internal anchor Pith review Pith/arXiv arXiv 2018

[8] [8]

Scientific Data10(1), 1–8 (2023)

Carstens, M., et al.: The Dresden Surgical Anatomy Dataset for abdominal organ segmentation in surgical data science. Scientific Data10(1), 1–8 (2023)

2023

[9] [9]

In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

Goyal, Y., Khot, T., Summers-Stay, D., Batra, D., Parikh, D.: Making the V in VQA Matter: elevating the role of image understanding in visual question answer- ing. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 6904–6913 (2017)

2017

[10] [10]

In: International Conference on Multimedia Modeling

Jha, D., et al.: Kvasir-Instrument: diagnostic and therapeutic tool segmentation dataset in gastrointestinal endoscopy. In: International Conference on Multimedia Modeling. pp. 218–229. Springer (2021)

2021

[11] [11]

In: Proceedings of the IEEE International Conference on Computer Vision (ICCV)

Kirillov, A., et al.: Segment Anything. In: Proceedings of the IEEE International Conference on Computer Vision (ICCV). pp. 4015–4026 (2023)

2023

[12] [12]

In: Medical Image Computing and Computer Assisted Intervention – MICCAI

Koleilat, T., Asgariandehkordi, H., Rivaz, H., Xiao, Y.: MedCLIP-SAM: bridging text and image towards universal medical image segmentation. In: Medical Image Computing and Computer Assisted Intervention – MICCAI. pp. 643–653. Springer (2024)

2024

[13] [13]

Scientific Data 5(1), 180251 (2018)

Lau, J.J., Gayen, S., Ben Abacha, A., Demner-Fushman, D.: A dataset of clinically generated visual questions and answers about radiology images. Scientific Data 5(1), 180251 (2018)

2018

[14] [14]

In: 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)

Liu, B., Zhan, L.M., Xu, L., Ma, L., Yang, Y., Wu, X.M.: SLAKE: a semantically- labeled knowledge-enhanced dataset for medical visual question answering. In: 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI). pp. 1650–1654. IEEE (2021)

2021

[15] [15]

Nature Communications15(1), 654 (2024)

Ma, J., He, Y., Li, F., Han, L., You, C., Wang, B.: Segment Anything in Medical Images. Nature Communications15(1), 654 (2024)

2024

[16] [16]

arXiv preprint arXiv:2312.12429 (2023)

Murali, A., et al.: The Endoscapes dataset for surgical scene segmentation, object detection, and critical view of safety assessment: official splits and benchmark. arXiv preprint arXiv:2312.12429 (2023)

work page arXiv 2023

[17] [17]

arXiv preprint arXiv:2401.00496 (2023) 10 C

Psychogyios, D., et al.: SAR-RARP50: segmentation of surgical instrumentation and action recognition on robot-assisted radical prostatectomy challenge. arXiv preprint arXiv:2401.00496 (2023) 10 C. Zhang et al

work page arXiv 2023

[18] [18]

arXiv preprint arXiv:2407.19305 (2024)

Schmidgall, S., Cho, J., Zakka, C., Hiesinger, W.: GP-VLS: a general-purpose vision language model for surgery. arXiv preprint arXiv:2407.19305 (2024)

work page arXiv 2024

[19] [19]

In: Medical Image Computing and Computer Assisted Intervention – MICCAI

Seenivasan, L., Islam, M., Kannan, G., Ren, H.: SurgicalGPT: end-to-end language- vision GPT for visual question answering in surgery. In: Medical Image Computing and Computer Assisted Intervention – MICCAI. pp. 281–290. Springer (2023)

2023

[20] [20]

In: Medical Image Computing and Computer Assisted Intervention – MICCAI

Seenivasan, L., Islam, M., Krishna, A.K., Ren, H.: Surgical-VQA: visual question answering in surgical scenes using transformer. In: Medical Image Computing and Computer Assisted Intervention – MICCAI. pp. 33–43. Springer (2022)

2022

[21] [21]

IEEE Transactions on Medical Imaging36(1), 86–97 (2017)

Twinanda, A.P., et al.: EndoNet: a deep architecture for recognition tasks on la- paroscopic videos. IEEE Transactions on Medical Imaging36(1), 86–97 (2017). https://doi.org/10.1109/TMI.2016.2593957

work page doi:10.1109/tmi.2016.2593957 2017

[22] [22]

arXiv preprint arXiv:2501.11347 (2025)

Wang, G., et al.: EndoChat: grounded multimodal large language model for endo- scopic surgery. arXiv preprint arXiv:2501.11347 (2025)

work page arXiv 2025

[23] [23]

IEEE Transactions on Medical Robotics and Bionics (2025)

Wang, Z., et al.: S4RoboFormer: scribble-supervised surgical robotic segmentation transformer via augmented consistency training. IEEE Transactions on Medical Robotics and Bionics (2025)

2025

[24] [24]

npj Digital Medicine8(1), 566 (2025)

Zhao, Z., et al.: Large-vocabulary segmentation for medical images with text prompts. npj Digital Medicine8(1), 566 (2025)

2025