pith. machine review for the scientific record. sign in

arxiv: 2604.04078 · v1 · submitted 2026-04-05 · 📡 eess.IV · cs.AI· cs.CV

Recognition: no theorem link

BAAI Cardiac Agent: An intelligent multimodal agent for automated reasoning and diagnosis of cardiovascular diseases from cardiac magnetic resonance imaging

Taiping Qu , Hongkai Zhang , Lantian Zhang , Can Zhao , Nan Zhang , Hui Wang , Zhen Zhou , Mingye Zou
show 13 more authors
Kairui Bo Pengfei Zhao Xingxing Jin Zixian Su Kun Jiang Huan Liu Yu Du Maozhou Wang Ruifang Yan Zhongyuan Wang Tiejun Huang Lei Xu Henggui Zhang
Authors on Pith no claims yet

Pith reviewed 2026-05-14 22:12 UTC · model grok-4.3

classification 📡 eess.IV cs.AIcs.CV
keywords cardiac magnetic resonancemultimodal agentautomated diagnosiscardiovascular diseaseimage segmentationfunctional quantificationclinical reporting
0
0 comments X

The pith

The BAAI Cardiac Agent orchestrates specialized expert models to automate full cardiac MRI interpretation and diagnosis.

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

The paper introduces a multimodal agent that combines multiple specialized cardiac models into one workflow for segmenting structures, measuring function, characterizing tissue, diagnosing disease, and producing structured reports from cardiac magnetic resonance scans. The system was tested on data from 2413 patients at two hospitals covering seven major cardiovascular conditions, reaching an internal AUC above 0.93 and external AUC above 0.81 while showing Pearson correlations over 0.90 for key left-ventricular measures such as ejection fraction. A sympathetic reader would care because CMR remains underused due to its complexity and need for scarce expertise; an automated end-to-end system could make reliable interpretation faster and more widely available.

Core claim

By dynamically orchestrating expert models for coordinated multimodal analysis, the BAAI Cardiac Agent enables accurate, efficient CMR interpretation, achieving an area under the ROC curve exceeding 0.93 internally and 0.81 externally across seven disease types, Pearson correlations exceeding 0.90 for ejection fraction, stroke volume, and left ventricular mass, and clinical reports with high concordance to expert radiologists.

What carries the argument

A multimodal intelligent agent that dynamically orchestrates specialized cardiac expert models to perform automated segmentation, functional quantification, tissue characterization, disease diagnosis, and structured report generation within a single workflow.

If this is right

  • Outperforms state-of-the-art models in both segmentation accuracy and diagnostic performance.
  • Delivers left-ventricular function estimates that match clinical reports with Pearson correlations above 0.90 for ejection fraction, stroke volume, and mass.
  • Produces structured clinical reports that show high agreement with radiologists across three experience levels.
  • Demonstrates feasibility of coordinated multimodal analysis for complex clinical imaging workflows.

Where Pith is reading between the lines

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

  • The orchestration approach could transfer to other multi-sequence imaging tasks such as brain or abdominal MRI.
  • Wider deployment might reduce interpretation time and inter-reader variability in routine cardiology practice.
  • Adding uncertainty estimates or active learning loops could further improve robustness on out-of-distribution scans.

Load-bearing premise

The specialized expert models remain accurate when dynamically orchestrated and performance on the two-hospital dataset generalizes to broader real-world clinical populations and imaging variations.

What would settle it

Independent evaluation on a larger multi-center dataset drawn from different hospitals and scanner vendors, checking whether external AUC stays above 0.80 and report concordance with radiologists remains high.

Figures

Figures reproduced from arXiv: 2604.04078 by Can Zhao, Henggui Zhang, Hongkai Zhang, Huan Liu, Hui Wang, Kairui Bo, Kun Jiang, Lantian Zhang, Lei Xu, Maozhou Wang, Mingye Zou, Nan Zhang, Pengfei Zhao, Ruifang Yan, Taiping Qu, Tiejun Huang, Xingxing Jin, Yu Du, Zhen Zhou, Zhongyuan Wang, Zixian Su.

Figure 1
Figure 1. Figure 1: A schematic overview of the BAAI Cardiac Agent illustrating the process of radiology report generation [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Performance of the expert CVDs diagnostic model in internal and external testing. a–c: Receiver operating [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Evaluation results of BAAI Cardiac Agent report consistency and expert model invocation success rate. (a) [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Schematic diagram of the BAAI Cardiac Agent workflow, in which assigned tasks are completed through [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Task-modality-tool mapping of the BAAI Cardiac Agent: the left panel presents the tasks integrated in the [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The architecture of the proposed expert model for cardiac segmentation and CVDs diagnosis is described as [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
read the original abstract

Cardiac magnetic resonance (CMR) is a cornerstone for diagnosing cardiovascular disease. However, it remains underutilized due to complex, time-consuming interpretation across multi-sequences, phases, quantitative measures that heavily reliant on specialized expertise. Here, we present BAAI Cardiac Agent, a multimodal intelligent system designed for end-to-end CMR interpretation. The agent integrates specialized cardiac expert models to perform automated segmentation of cardiac structures, functional quantification, tissue characterization and disease diagnosis, and generates structured clinical reports within a unified workflow. Evaluated on CMR datasets from two hospitals (2413 patients) spanning 7-types of major cardiovascular diseases, the agent achieved an area under the receiver-operating-characteristic curve exceeding 0.93 internally and 0.81 externally. In the task of estimating left ventricular function indices, the results generated by this system for core parameters such as ejection fraction, stroke volume, and left ventricular mass are highly consistent with clinical reports, with Pearson correlation coefficients all exceeding 0.90. The agent outperformed state-of-the-art models in segmentation and diagnostic tasks, and generated clinical reports showing high concordance with expert radiologists (six readers across three experience levels). By dynamically orchestrating expert models for coordinated multimodal analysis, this agent framework enables accurate, efficient CMR interpretation and highlights its potentials for complex clinical imaging workflows. Code is available at https://github.com/plantain-herb/Cardiac-Agent.

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.

Referee Report

2 major / 1 minor

Summary. The manuscript presents BAAI Cardiac Agent, a multimodal intelligent system that integrates specialized cardiac expert models for automated segmentation of cardiac structures, functional quantification, tissue characterization, disease diagnosis across 7 major cardiovascular diseases, and generation of structured clinical reports from CMR scans. Evaluated on 2413 patients from two hospitals, it reports AUC exceeding 0.93 internally and 0.81 externally, Pearson correlations exceeding 0.90 for left ventricular indices (ejection fraction, stroke volume, mass), outperformance of state-of-the-art models, and high concordance with expert radiologists (six readers across experience levels). Code is publicly available.

Significance. If the performance claims hold under scrutiny, the work offers a meaningful step toward automating complex, multi-sequence CMR interpretation, which is currently limited by time and expertise demands. The dynamic orchestration of expert models and public code release are strengths that support reproducibility and potential clinical utility in cardiovascular workflows.

major comments (2)
  1. [Abstract] Abstract: The headline metrics (internal AUC >0.93, external AUC >0.81; Pearson r>0.90 on LV indices) are presented without any description of model architectures, training procedures, exclusion criteria, or statistical testing, which prevents verification of the central performance and generalization claims.
  2. [Evaluation] Evaluation section: External validation uses only a single additional hospital with no reported breakdown of inter-site differences in field strength, sequence parameters, vendor, demographics, or label distribution; the observed drop from 0.93 to 0.81 AUC therefore remains unexamined and undermines the claim of reliable real-world deployment.
minor comments (1)
  1. [Abstract] Abstract: The statement that results are 'highly consistent with clinical reports' should specify the exact concordance metrics (beyond Pearson r) and report inter-reader variability for the six radiologists.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address the major comments point by point below, focusing on improving clarity around performance metrics and external validation details. Revisions will be incorporated where feasible to strengthen the presentation.

read point-by-point responses

  1. Referee: [Abstract] The headline metrics (internal AUC >0.93, external AUC >0.81; Pearson r>0.90 on LV indices) are presented without any description of model architectures, training procedures, exclusion criteria, or statistical testing, which prevents verification of the central performance and generalization claims.

    Authors: We agree the abstract is highly condensed due to length limits and omits these specifics. Full details on model architectures (dynamic orchestration of expert models for segmentation, quantification, and diagnosis), training procedures, exclusion criteria, and statistical methods (AUC computation, Pearson correlations) are provided in the Methods and Results sections. In revision, we will add concise phrasing to the abstract referencing the multimodal agent framework and multi-center evaluation to aid quick verification, while retaining comprehensive descriptions in the main text. revision: partial

  2. Referee: [Evaluation] External validation uses only a single additional hospital with no reported breakdown of inter-site differences in field strength, sequence parameters, vendor, demographics, or label distribution; the observed drop from 0.93 to 0.81 AUC therefore remains unexamined and undermines the claim of reliable real-world deployment.

    Authors: We concur that explicit characterization of inter-site differences would better contextualize the performance drop and support deployment claims. In the revised manuscript, we will add a table in the Evaluation section (or supplement) detailing differences in field strength, sequence parameters, vendors, demographics, and label distributions between the internal and external sites. This will help examine the domain shift contributing to the AUC reduction from 0.93 to 0.81 while preserving the claim of meaningful generalization across hospitals. revision: yes

Circularity Check

0 steps flagged

No circularity: performance metrics derived from held-out evaluation, not by construction

full rationale

The paper presents an empirical system whose core claims (AUC >0.93 internal / 0.81 external; Pearson r >0.90 on LV indices) are obtained by running the agent on separate patient cohorts from two hospitals and comparing outputs to clinical reports and expert readers. No equation or parameter is fitted to the final reported numbers and then re-labeled as a prediction; the external set is not used in training or hyper-parameter selection for the headline metrics. Self-citations, if present, are not load-bearing for the performance numbers themselves. The derivation chain therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central performance claims rest on trained neural network parameters learned from the hospital datasets and standard machine learning assumptions about data representativeness and model generalization.

free parameters (1)
  • model hyperparameters and training settings
    Deep learning models require numerous hyperparameters and optimization choices that are fitted or selected during development.
axioms (1)
  • domain assumption Training and test data distributions are sufficiently similar for the reported generalization to hold.
    Standard assumption invoked for internal-to-external performance drop from 0.93 to 0.81 AUC.

pith-pipeline@v0.9.0 · 5633 in / 1451 out tokens · 61183 ms · 2026-05-14T22:12:23.825410+00:00 · methodology

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

60 extracted references · 60 canonical work pages · 3 internal anchors

  1. [1]

    Mc Namara, H

    K. Mc Namara, H. Alzubaidi, and J. K. Jackson. Cardiovascular disease as a leading cause of death: how are pharmacists getting involved?Integrated Pharmacy Research and Practice, 8:1–11, 2019

    work page 2019

  2. [2]

    Alhabeeb, M

    H. Alhabeeb, M. H. Sohouli, A. Lari, S. Fatahi, F. Shidfar, O. Alomar, and A. Abu-Zaid. Impact of orange juice consumption on cardiovascular disease risk factors: a systematic review and meta-analysis of randomized- controlled trials.Critical Reviews in F ood Science and Nutrition, 62(12):3389–3402, 2020

    work page 2020

  3. [3]

    V Mcmurray, Piotr Ponikowski, Anna Strömberg, Dirk J Veldhuisen, Dan Atar, Arno W Hoes, and Andre Keren

    Kenneth Dickstein, Alain Cohen-Solal, Gerasimos Filippatos, John J. V Mcmurray, Piotr Ponikowski, Anna Strömberg, Dirk J Veldhuisen, Dan Atar, Arno W Hoes, and Andre Keren. Esc guidelines for the diagnosis and treatment of acute and chronic heart failure 2008;.European Journal of Heart Failure, 11(1):110–110, 2014

    work page 2008

  4. [4]

    Screening and diagnosis of cardiovascular disease using artificial intelligence-enabled cardiac magnetic resonance imaging.Nature Medicine, 30(5):1471–1480, 2024

    Yan-Ran Wang, Kai Yang, Yi Wen, Pengcheng Wang, Yuepeng Hu, Yongfan Lai, Yufeng Wang, Kankan Zhao, Siyi Tang, Angela Zhang, et al. Screening and diagnosis of cardiovascular disease using artificial intelligence-enabled cardiac magnetic resonance imaging.Nature Medicine, 30(5):1471–1480, 2024. 16 Cardiac Agent

    work page 2024

  5. [5]

    Michael Salerno and Christopher M. Kramer. Advances in parametric mapping with cmr imaging.JACC: Cardiovascular Imaging, 6(7):806–822, 2013

    work page 2013

  6. [6]

    Quantification of myocardial perfusion by cardiovascular magnetic resonance.Journal of Cardiovascular Magnetic Resonance (BioMed Central), 12(1):1–16, 2010

    Michael Jerosch-Herold. Quantification of myocardial perfusion by cardiovascular magnetic resonance.Journal of Cardiovascular Magnetic Resonance (BioMed Central), 12(1):1–16, 2010

    work page 2010

  7. [7]

    N. I. Bouwer, C. Liesting, Mjm Kofflard, J. J. Brugts, and E. Boersma. 2d-echocardiography vs cardiac mri strain using deep learning: a prospective cohort study in patients with her2-positive breast cancer undergoing trastuzumab.European Heart Journal Cardiovascular Imaging, 22, 2021

    work page 2021

  8. [8]

    Ibrahim, Luba Frank, Dhiraj Baruah, Jason C

    El Sayed H. Ibrahim, Luba Frank, Dhiraj Baruah, Jason C. Rubenstein, V . Emre Arpinar, Andrew S. Nencka, Kevin M. Koch, L Tugan Muftuler, Orhan Unal, and Jadranka Stojanovska. Value cmr: Towards a comprehensive, rapid, cost-effective cardiovascular magnetic resonance imaging.Cold Spring Harbor Laboratory Press, 2020

    work page 2020

  9. [9]

    Guidelines for training in cardiovascular magnetic resonance (cmr).J Cardiovasc Magn Reson, 9(1):3–4, 2007

    Raymond Kim, Albert De Roos, Eckart Fleck, Charles Higgins, Gerald Pohost, Martin Prince, and Warren Manning. Guidelines for training in cardiovascular magnetic resonance (cmr).J Cardiovasc Magn Reson, 9(1):3–4, 2007

    work page 2007

  10. [10]

    Magnetic resonance angiography: current status and future directions.Journal of Cardiovascular Magnetic Resonance, 13(1):19, 2011

    Michael P Hartung, Thomas M Grist, and Christopher J François. Magnetic resonance angiography: current status and future directions.Journal of Cardiovascular Magnetic Resonance, 13(1):19, 2011

    work page 2011

  11. [11]

    Jeanette Schulz-Menger, David A Bluemke, Jens Bremerich, Scott D Flamm, Mark A Fogel, Matthias G Friedrich, Raymond J Kim, Florian von Knobelsdorff-Brenkenhoff, Christopher M Kramer, Dudley J Pennell, et al. Standardized image interpretation and post processing in cardiovascular magnetic resonance: Society for cardiovascular magnetic resonance (scmr) boar...

    work page 2013

  12. [12]

    Ai-based detection and classification of anomalous aortic origin of coronary arteries using coronary ct angiography images.Nature Communications, 16(1):3095, 2025

    Isaac Shiri, Giovanni Baj, Pooya Mohammadi Kazaj, Marius R Bigler, Anselm W Stark, Waldo Valenzuela, Ryota Kakizaki, Matthias Siepe, Stephan Windecker, Lorenz Räber, et al. Ai-based detection and classification of anomalous aortic origin of coronary arteries using coronary ct angiography images.Nature Communications, 16(1):3095, 2025

    work page 2025

  13. [13]

    Myops-net: Myocardial pathology segmentation with flexible combination of multi-sequence cmr images.Medical Image Analysis, 84:102694, 2023

    Junyi Qiu, Lei Li, Sihan Wang, Ke Zhang, Yinyin Chen, Shan Yang, and Xiahai Zhuang. Myops-net: Myocardial pathology segmentation with flexible combination of multi-sequence cmr images.Medical Image Analysis, 84:102694, 2023

    work page 2023

  14. [14]

    Fully automated, quality-controlled cardiac analysis from cmr: validation and large-scale application to characterize cardiac function.Cardiovascular Imaging, 13(3):684–695, 2020

    Bram Ruijsink, Esther Puyol-Antón, Ilkay Oksuz, Matthew Sinclair, Wenjia Bai, Julia A Schnabel, Reza Razavi, and Andrew P King. Fully automated, quality-controlled cardiac analysis from cmr: validation and large-scale application to characterize cardiac function.Cardiovascular Imaging, 13(3):684–695, 2020

    work page 2020

  15. [15]

    Wu, and Shihua Zhao

    Yan-Ran (Joyce) Wang, Kai Yang, Yi Wen, Pengcheng Wang, Yuepeng Hu, Yongfan Lai, Yufeng Wang, Kankan Zhao, Siyi Tang, Angela Zhang, Huayi Zhan, Minjie Lu, Xiuyu Chen, Shujuan Yang, Zhixiang Dong, Yining Wang, Hui Liu, Lei Zhao, Lu Huang, Yunling Li, Lianming Wu, Zixian Chen, Yi Luo, Dongbo Liu, Pengbo Zhao, Keldon Lin, Joseph C. Wu, and Shihua Zhao. Scree...

    work page 2024

  16. [16]

    Clinical application of cmr in cardiomyopathies: evolving concepts and techniques.Heart Failure Reviews, 28(1):77–95, 2023

    Marco Merlo, Kyle Lam, Giulia Gagno, Anna Baritussio, Barbara Bauce, Elena Biagini, Marco Canepa, Alberto Cipriani, Silvia Castelletti, Santo Dellegrottaglie, Andrea Igoren Guaricci, Massimo Imazio, Giuseppe Limongelli, Maria Beatrice Musumeci, Vanda Parisi, Silvia Pica, Gianluca Pontone, Giancarlo Todiere, Camilla Torlasco, Cristina Basso, Gianfranco Sin...

    work page 2023

  17. [17]

    Cardiac magnetic resonance to predict cardiac mass malignancy: The cmr mass score

    Pasquale Paolisso, Luca Bergamaschi, Francesco Angeli, Marta Belmonte, Alberto Foà, Lisa Canton, Damiano Fedele, Matteo Armillotta, Angelo Sansonetti, Francesca Bodega, Sara Amicone, Nicole Suma, Emanuele Gallinoro, Domenico Attinà, Fabio Niro, Paola Rucci, Elisa Gherbesi, Stefano Carugo, Saima Mushtaq, Andrea Baggiano, Anna Giulia Pavon, Marco Guglielmo,...

    work page 2024

  18. [18]

    Towards a holistic framework for multimodal llm in 3d brain ct radiology report generation.Nature Communications, 16(1):2258, 2025

    Cheng-Yi Li, Kao-Jung Chang, Cheng-Fu Yang, Hsin-Yu Wu, Wenting Chen, Hritik Bansal, Ling Chen, Yi-Ping Yang, Yu-Chun Chen, Shih-Pin Chen, et al. Towards a holistic framework for multimodal llm in 3d brain ct radiology report generation.Nature Communications, 16(1):2258, 2025

    work page 2025

  19. [19]

    Vila-m3: Enhancing vision-language models with medical expert knowledge

    Vishwesh Nath, Wenqi Li, Dong Yang, Andriy Myronenko, Mingxin Zheng, Yao Lu, Zhijian Liu, Hongxu Yin, Yee Man Law, Yucheng Tang, Pengfei Guo, Can Zhao, Ziyue Xu, Yufan He, Stephanie Harmon, Benjamin Simon, Greg Heinrich, Stephen Aylward, Marc Edgar, Michael Zephyr, Pavlo Molchanov, Baris Turkbey, Holger Roth, and Daguang Xu. Vila-m3: Enhancing vision-lang...

    work page 2025

  20. [20]

    Med-flamingo: a multimodal medical few-shot learner

    Michael Moor, Qian Huang, Shirley Wu, Michihiro Yasunaga, Yash Dalmia, Jure Leskovec, Cyril Zakka, Eduardo Pontes Reis, and Pranav Rajpurkar. Med-flamingo: a multimodal medical few-shot learner. volume 225 ofProceedings of Machine Learning Research, pages 353–367, 2023

    work page 2023

  21. [21]

    Llava-med: Training a large language-and-vision assistant for biomedicine in one day, 2023

    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, 2023

    work page 2023

  22. [22]

    Summers, and Yingying Zhu

    Xinyue Hu, Lin Gu, Kazuma Kobayashi, Liangchen Liu, Mengliang Zhang, Tatsuya Harada, Ronald M. Summers, and Yingying Zhu. Interpretable medical image visual question answering via multi-modal relationship graph learning.Medical Image Analysis, 97:103279, 2024

    work page 2024

  23. [23]

    Candidate-heuristic in-context learning: A new framework for enhancing medical visual question answering with llms.Information Processing & Management, 61(5):103805, 2024

    Xiao Liang, Di Wang, Haodi Zhong, Quan Wang, Ronghan Li, Rui Jia, and Bo Wan. Candidate-heuristic in-context learning: A new framework for enhancing medical visual question answering with llms.Information Processing & Management, 61(5):103805, 2024

    work page 2024

  24. [24]

    Advanced vision- language pipelines: Contextual learning and interactive segmentation for medical imaging

    Sachin M Sabariram, Sanjay Vikram C B, Sharon Deborah E, Kavin M, and Saravanan G. Advanced vision- language pipelines: Contextual learning and interactive segmentation for medical imaging. In2025 International Conference on Machine Learning and Autonomous Systems (ICMLAS), pages 608–615, 2025

    work page 2025

  25. [25]

    Mmedagent: Learning to use medical tools with multi-modal agent, 2024

    Binxu Li, Tiankai Yan, Yuanting Pan, Jie Luo, Ruiyang Ji, Jiayuan Ding, Zhe Xu, Shilong Liu, Haoyu Dong, Zihao Lin, and Yixin Wang. Mmedagent: Learning to use medical tools with multi-modal agent, 2024

    work page 2024

  26. [26]

    Mdagents: An adaptive collaboration of llms for medical decision-making

    Yubin Kim, Chanwoo Park, Hyewon Jeong, Yik Siu Chan, Xuhai Xu, Daniel McDuff, Hyeonhoon Lee, Marzyeh Ghassemi, Cynthia Breazeal, and Hae Won Park. Mdagents: An adaptive collaboration of llms for medical decision-making. In A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, and C. Zhang, editors, Advances in Neural Information Processing ...

    work page 2024

  27. [27]

    Medsearch: A conversational agent for real-time, evidence-based medical question- answering.Intelligence-Based Medicine, page 100274, 2025

    Maria Camila Villa, Isabella Llano, Natalia Castano-Villegas, Julian Martinez, Maria Fernanda Guevara, Jose Zea, and Laura Velásquez. Medsearch: A conversational agent for real-time, evidence-based medical question- answering.Intelligence-Based Medicine, page 100274, 2025

    work page 2025

  28. [28]

    Jianing Qiu, Kyle Lam, Guohao Li, Amish Acharya, Tien Yin Wong, Ara Darzi, Wu Yuan, and Eric J. Topol. Llm-based agentic systems in medicine and healthcare.Nature Machine Intelligence, 6(12):1418–1420, 2024

    work page 2024

  29. [29]

    Miao, Eduardo Rodriguez Almaraz, Madhumita Sushil, Atul J

    Nikita Mehandru, Brenda Y . Miao, Eduardo Rodriguez Almaraz, Madhumita Sushil, Atul J. Butte, and Ahmed Alaa. Evaluating large language models as agents in the clinic.npj Digital Medicine, 7(1):84, 2024

    work page 2024

  30. [30]

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

    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(2):203–211, 2021

    work page 2021

  31. [31]

    Medsam2: Segment anything in 3d medical images and videos.arXiv preprint arXiv:2504.03600, 2025

    Jun Ma, Zongxin Yang, Sumin Kim, Bihui Chen, Mohammed Baharoon, Adibvafa Fallahpour, Reza Asakereh, Hongwei Lyu, and Bo Wang. Medsam2: Segment anything in 3d medical images and videos.arXiv preprint arXiv:2504.03600, 2025

    work page arXiv 2025

  32. [32]

    Resunet++: An advanced architecture for medical image segmentation

    Debesh Jha, Pia H Smedsrud, Michael A Riegler, Dag Johansen, Thomas De Lange, Pål Halvorsen, and Håvard D Johansen. Resunet++: An advanced architecture for medical image segmentation. In2019 IEEE international symposium on multimedia (ISM), pages 225–2255. IEEE, 2019

    work page 2019

  33. [33]

    Diff-unet: A diffusion embedded network for robust 3d medical image segmentation.Medical Image Analysis, page 103654, 2025

    Zhaohu Xing, Liang Wan, Huazhu Fu, Guang Yang, Yijun Yang, Lequan Yu, Baiying Lei, and Lei Zhu. Diff-unet: A diffusion embedded network for robust 3d medical image segmentation.Medical Image Analysis, page 103654, 2025

    work page 2025

  34. [34]

    An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

    Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale.arXiv preprint arXiv:2010.11929, 2020

    work page internal anchor Pith review Pith/arXiv arXiv 2010

  35. [35]

    Learning spatio-temporal representation with pseudo-3d residual networks

    Zhaofan Qiu, Ting Yao, and Tao Mei. Learning spatio-temporal representation with pseudo-3d residual networks. Inproceedings of the IEEE International Conference on Computer Vision, pages 5533–5541, 2017

    work page 2017

  36. [36]

    Qwen3 Technical Report

    An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, et al. Qwen3 technical report.arXiv preprint arXiv:2505.09388, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  37. [37]

    Medm-vl: What makes a good medical lvlm? InInternational Workshop on Agentic AI for Medicine, pages 290–299

    Yiming Shi, Shaoshuai Yang, Xun Zhu, Haoyu Wang, Xiangling Fu, Miao Li, and Ji Wu. Medm-vl: What makes a good medical lvlm? InInternational Workshop on Agentic AI for Medicine, pages 290–299. Springer, 2025

    work page 2025

  38. [38]

    Llava-med: Training a large language-and-vision assistant for biomedicine in one day

    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, 2023

    work page 2023

  39. [39]

    Use of cardiac imaging in chronic coronary syndromes: the eureca imaging registry.European heart journal, 44(2):142–158, 2023

    Danilo Neglia, Riccardo Liga, Alessia Gimelli, Tomaž Podlesnikar, Marta Cviji´c, Gianluca Pontone, Marcelo Haer- tel Miglioranza, Andrea Igoren Guaricci, Sara Seitun, Alberto Clemente, et al. Use of cardiac imaging in chronic coronary syndromes: the eureca imaging registry.European heart journal, 44(2):142–158, 2023. 18 Cardiac Agent

    work page 2023

  40. [40]

    Steve R Ommen, Seema Mital, Michael A Burke, Sharlene M Day, Anita Deswal, Perry Elliott, Lauren L Evanovich, Judy Hung, Jose A Joglar, Paul Kantor, et al. 2020 aha/acc guideline for the diagnosis and treatment of patients with hypertrophic cardiomyopathy: a report of the american college of cardiology/american heart association joint committee on clinica...

    work page 2020

  41. [41]

    Llava-plus: Learning to use tools for creating multimodal agents

    Shilong Liu, Hao Cheng, Haotian Liu, Hao Zhang, Feng Li, Tianhe Ren, Xueyan Zou, Jianwei Yang, Hang Su, Jun Zhu, et al. Llava-plus: Learning to use tools for creating multimodal agents. InEuropean conference on computer vision, pages 126–142. Springer, 2024

    work page 2024

  42. [42]

    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

    work page 2023

  43. [43]

    Lora: Low-rank adaptation of large language models.ICLR, 1(2):3, 2022

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, et al. Lora: Low-rank adaptation of large language models.ICLR, 1(2):3, 2022

    work page 2022

  44. [44]

    {Zero-offload}: Democratizing {billion-scale} model training

    Jie Ren, Samyam Rajbhandari, Reza Yazdani Aminabadi, Olatunji Ruwase, Shuangyan Yang, Minjia Zhang, Dong Li, and Yuxiong He. {Zero-offload}: Democratizing {billion-scale} model training. In2021 USENIX Annual Technical Conference (USENIX ATC 21), pages 551–564, 2021

    work page 2021

  45. [45]

    Mixed Precision Training

    Paulius Micikevicius, Sharan Narang, Jonah Alben, Gregory Diamos, Erich Elsen, David Garcia, Boris Gins- burg, Michael Houston, Oleksii Kuchaiev, Ganesh Venkatesh, et al. Mixed precision training.arXiv preprint arXiv:1710.03740, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  46. [46]

    Chatcad+: Toward a universal and reliable interactive cad using llms.IEEE Transactions on Medical Imaging, 43(11):3755–3766, 2024

    Zihao Zhao, Sheng Wang, Jinchen Gu, Yitao Zhu, Lanzhuju Mei, Zixu Zhuang, Zhiming Cui, Qian Wang, and Dinggang Shen. Chatcad+: Toward a universal and reliable interactive cad using llms.IEEE Transactions on Medical Imaging, 43(11):3755–3766, 2024

    work page 2024

  47. [47]

    Salim S Virani, L Kristin Newby, Suzanne V Arnold, Vera Bittner, LaPrincess C Brewer, Susan Halli Demeter, Dave L Dixon, William F Fearon, Beverly Hess, Heather M Johnson, et al. 2023 aha/acc/accp/aspc/nla/pcna guideline for the management of patients with chronic coronary disease: a report of the american heart associ- ation/american college of cardiolog...

    work page 2023

  48. [48]

    Guy De Backer, Ettore Ambrosioni, Knut Borch-Johnsen, Carlos Brotons, Renata Cifkova, Jean Dallongeville, Shah Ebrahim, Ole Faergeman, Ian Graham, Giuseppe Mancia, et al. European guidelines on cardiovascular disease prevention in clinical practice; third joint task force of european and other societies on cardiovascular disease prevention in clinical pra...

    work page 2003

  49. [49]

    Elena Arbelo, Alexandros Protonotarios, Juan R Gimeno, Eloisa Arbustini, Roberto Barriales-Villa, Cristina Basso, Connie R Bezzina, Elena Biagini, Nico A Blom, Rudolf A De Boer, et al. 2023 esc guidelines for the management of cardiomyopathies: developed by the task force on the management of cardiomyopathies of the european society of cardiology (esc).Eu...

    work page 2023

  50. [50]

    Resunet-a: A deep learning framework for semantic segmentation of remotely sensed data.ISPRS Journal of Photogrammetry and Remote Sensing, 162:94–114, 2020

    Foivos I Diakogiannis, François Waldner, Peter Caccetta, and Chen Wu. Resunet-a: A deep learning framework for semantic segmentation of remotely sensed data.ISPRS Journal of Photogrammetry and Remote Sensing, 162:94–114, 2020

    work page 2020

  51. [51]

    Cardiovascular magnetic resonance imaging: emerging techniques and applications.Heart, 107(9):697–704, 2021

    Amrit Chowdhary, Pankaj Garg, Arka Das, Muhummad Sohaib Nazir, and Sven Plein. Cardiovascular magnetic resonance imaging: emerging techniques and applications.Heart, 107(9):697–704, 2021

    work page 2021

  52. [52]

    Advancing cardiovascular diagnostics: the expanding role of cmr in heart failure and cardiomyopathies.Journal of Clinical Medicine, 14(3):865, 2025

    Antonio Luca Maria Parlati, Ermanno Nardi, Federica Marzano, Cristina Madaudo, Mariafrancesca Di Santo, Ciro Cotticelli, Simone Agizza, Giuseppe Maria Abbellito, Fabrizio Perrone Filardi, Mario Del Giudice, et al. Advancing cardiovascular diagnostics: the expanding role of cmr in heart failure and cardiomyopathies.Journal of Clinical Medicine, 14(3):865, 2025

    work page 2025

  53. [53]

    Identification of cardiac wall motion abnormalities in diverse populations by deep learning of the electrocardiogram.npj Digital Medicine, 8(1):21, 2025

    Albert J Rogers, Neal K Bhatia, Sabyasachi Bandyopadhyay, James Tooley, Rayan Ansari, Vyom Thakkar, Justin Xu, Jessica Torres Soto, Jagteshwar S Tung, Mahmood I Alhusseini, et al. Identification of cardiac wall motion abnormalities in diverse populations by deep learning of the electrocardiogram.npj Digital Medicine, 8(1):21, 2025

    work page 2025

  54. [54]

    Non-ischemic dilated cardiomyopathy and cardiac fibrosis.Heart Failure Reviews, 26(5):1081–1101, 2021

    Bianca Olivia Cojan-Minzat, Alexandru Zlibut, and Lucia Agoston-Coldea. Non-ischemic dilated cardiomyopathy and cardiac fibrosis.Heart Failure Reviews, 26(5):1081–1101, 2021

    work page 2021

  55. [55]

    Myocardial fibrosis: morphologic patterns and role of imaging in diagnosis and prognostication.Cardiovascular pathology, 56:107391, 2022

    Carla Giordano, Marco Francone, Giulia Cundari, Annalinda Pisano, and Giulia d’Amati. Myocardial fibrosis: morphologic patterns and role of imaging in diagnosis and prognostication.Cardiovascular pathology, 56:107391, 2022. 19 Cardiac Agent

    work page 2022

  56. [56]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016

    work page 2016

  57. [57]

    Learning spatio-temporal features with 3d residual networks for action recognition

    Kensho Hara, Hirokatsu Kataoka, and Yutaka Satoh. Learning spatio-temporal features with 3d residual networks for action recognition. InProceedings of the IEEE international conference on computer vision workshops, pages 3154–3160, 2017

    work page 2017

  58. [58]

    Attention is all you need.Advances in neural information processing systems, 30, 2017

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need.Advances in neural information processing systems, 30, 2017

    work page 2017

  59. [59]

    Multilayer perceptron and neural networks.WSEAS Transactions on Circuits and Systems, 8(7):579–588, 2009

    Marius-Constantin Popescu, Valentina E Balas, Liliana Perescu-Popescu, and Nikos Mastorakis. Multilayer perceptron and neural networks.WSEAS Transactions on Circuits and Systems, 8(7):579–588, 2009

    work page 2009

  60. [60]

    short - axis

    Christopher M Kramer, Jörg Barkhausen, Chiara Bucciarelli-Ducci, Scott D Flamm, Raymond J Kim, and Eike Nagel. Standardized cardiovascular magnetic resonance imaging (cmr) protocols: 2020 update.Journal of Cardiovascular Magnetic Resonance, 22(1):17, 2020. Data availability IRB approval was obtained from all participating institutions for imaging and data...

    work page 2020