Pith. sign in

REVIEW 5 major objections 6 minor 52 references

FactCheXcker: Mitigating Measurement Hallucinations in Chest X-ray Report Generation Models

T0 review · 5 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

Pith's one-line read Measurement hallucinations in chest X-ray reports can be corrected post-hoc by a modular pipeline that re-measures the image and rewrites the report.

desk verdict Useful framework, but the 135% improvement headline is miscomputed—standard relative reduction is about 43%. read the letter →

arxiv 2411.18672 v3 pith:IDYF42N5 submitted 2024-11-27 cs.CV

classification cs.CV
keywords measurementhallucinationradiologyreportgenerationchestX-rayendotrachealtubeplacementquery-code-updatetool-augmentedLLMMIMIC-CXRmitigation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

FactCheXcker is a modular pipeline that corrects measurement hallucinations in radiology reports after they are generated, without retraining or modifying the underlying vision-language model. The framework reads a model-written report, turns any quantitative claim into a measurement query, writes and executes Python code against a domain-specific API backed by specialist vision tools, and then rewrites the report with the verified numbers. On the MIMIC-CXR dataset, applying FactCheXcker to endotracheal tube placement improved 10 of 11 report generation models, reducing measurement error by an average of 135% and cutting clinically significant errors (over 1.5 cm) roughly in half, while keeping the original reports' wording and quality intact. The paper positions this as a general prescription: specialized modules can de-hallucinate any report generator, and each new measurable object only requires adding a tool module.

What carries the argument

The load-bearing mechanism is the query-code-update loop over a domain-specific API. A query generator (GPT-4o mini) converts report sentences into structured measurement requests, such as "measure the distance between the endotracheal tube and the carina." A code generator writes Python functions that call API methods—exists, find, distance, width, diameter, dimensions, within—each implemented by swappable specialist tools: a fine-tuned ResNet-50+ for tube presence, a fine-tuned CarinaNet+ that outputs carina and tube-tip coordinates, chest anatomical segmentation models for regions such as heart and lungs, and the image's pixel-spacing metadata to convert pixel distances into centimeters. A report updater then merges the executed results back into the text, deleting hallucinated objects and flagging misplaced tubes using a rule-based interpretation of the 3–7 cm "correct placement" guideline.

What would settle it

Replace CarinaNet+ with any publicly available carina and tube-tip localizer, run FactCheXcker on the MIMIC-CXR test set, and check whether the updated reports still beat the original models on mean absolute error; if the improvement disappears, the 135% reduction is an artifact of the unreleased private module.

Watch

Extended reading notes

Core claim

The paper's central claim is that measurement hallucinations in chest X-ray report generation are not an unavoidable failure of the report model but a separate, correctable output defect. FactCheXcker demonstrates this by decomposing measurement into sub-tasks—deciding whether an object is present, localizing it, locating anatomical landmarks, converting pixels to centimeters, and judging clinical placement—and handing each sub-task to a specialized module. The strongest quantitative evidence is on endotracheal tube placement: across 11 models, the framework raises ETT detection precision from 0.49 to 0.65, lowers measurement mean absolute error from 1.93 cm to 0.82 cm (a 135.0% average improvement), improves the composite MAE/F1 score by 186%, and raises placement-correctness precision from 0.84 to 0.94. The authors also report that this is achieved while retaining the style and readability of the original report.

Load-bearing premise

The entire correction rests on CarinaNet+, a model fine-tuned on a private 1,100-image dataset from 22 hospitals whose weights are not released; if its localization accuracy does not transfer to new hospitals or to the specific images where the baseline models hallucinate, the reported improvement will not replicate.

Editorial extensions

If this is right

  • Any existing report generation model can be de-hallucinated post-hoc without weight access, making the fix applicable to closed and open models alike.
  • The pipeline's measurement accuracy is upper-bounded by its best tool module; improving modules such as CarinaNet+ directly improves all downstream reports.
  • The framework, in principle, extends beyond endotracheal tubes to lesions, catheters, pneumothorax and other measurable findings by adding new tool modules to the same API.
  • Adding a reliable detector for every image (not just reports that mention an ETT) would reduce false negatives, but the paper's ablation shows precision drops, so production use needs a better presence detector or confidence gating.
  • For already-accurate models such as MedVersa and RaDialog the gains are modest (15–32% MAE improvement), so the marginal value of correction is largest for weak generators.

Reading between the lines

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

  • If the unreleased CarinaNet+ weights and its 22-hospital training data were made public, the framework could be independently replicated and stress-tested on new institutions; without them, the 135% figure cannot be separated from the quality of that one private module.
  • The evaluation metric penalizes a missing measurement as a zero (yielding an error equal to the ground-truth value), so a fraction of the measured improvement comes from deleting hallucinated ETT mentions rather than correcting numbers; a per-case decomposition of deletion versus correction would clarify the mechanism.
  • A natural next experiment is to apply the same loop to lesion size measurement or cardiothoracic ratio, where the tool modules are weaker today; the paper predicts the framework would still reduce errors, but only if a reliable segmenter exists.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper proposes FactCheXcker, a modular query-code-update pipeline that corrects measurement hallucinations in chest X-ray report generation. Given a model-generated report and the chest X-ray, FactCheXcker generates measurement queries, writes and executes Python code against an API implemented by fine-tuned tool modules (ResNet-50+ for ETT presence, CarinaNet+ for ETT/carina coordinates, TorchXRayVision for anatomy), and uses a Report Updater to revise the report. The method is evaluated on the MIMIC-CXR test set for endotracheal tube placement with 11 report-generation models, reporting presence precision, measurement MAE, a composite MAE/F1 score, and placement precision. The authors claim that FactCheXcker improves 10 of 11 models and achieves an average MAE improvement of 135.0%.

Significance. If the findings hold, the framework offers a practical, model-agnostic approach to reducing quantitative errors in generated radiology reports, with no retraining of the original model and a modular design that can be extended to other measurement tasks. The evaluation across 11 models is a substantial strength, and the paper ships code. The central direction is credible, but the headline quantitative claim is miscalculated: the reported improvement percentages use an inverted denominator, and the average is a pooled ratio rather than a per-model mean. Several internal inconsistencies in the reported metrics also need to be resolved before the paper's core evidence can be accepted.

major comments (5)
  1. [Abstract and Table 4] The 'Improvement' column in Table 4 is not the standard relative reduction from the original error. For CheXagent, the reported 191.0% equals (1.98−0.68)/0.68, whereas the standard relative reduction is (1.98−0.68)/1.98 = 65.7%. The same inverted denominator is used in every row, and the Abstract's 135.0% is the pooled ratio (1.93−0.82)/0.82 rather than an average of per-model improvements; recomputing each row with the standard formula gives a mean of approximately 43%. The 10-of-11 improvement claim survives, but the headline quantitative claim in the Abstract, Section 5.3, Table 4, Table 8, and Table 10 is overstated by roughly a factor of three. Please state the formula for improvement, report standard relative reductions, and clarify whether pooled or per-model averaging is intended.
  2. [Table 4 and Figure 4] The reported original MAE for RadFM (0.65 cm) is inconsistent with the original failure rate (44.4%) shown in Figure 4. With a failure threshold of 1.5 cm, the MAE must be at least 0.444 × 1.5 = 0.666 cm if all failed cases have errors just above the threshold and all non-failed cases have zero error. Additionally, the updated MAE worsens from 0.65 to 1.04 while the failure rate drops from 44.4% to 11.1%, which is difficult to reconcile. Please specify the exact case populations used for MAE and failure rate (e.g., all studies with ground-truth ETT measurement versus only those with a model-generated measurement) and recompute.
  3. [Section 4.4 and Table 2] The MAE definition sets missing model measurements to 0, but the paper does not state over how many cases each model's MAE is averaged, nor how the 45 radiologist-annotated ground-truth cases are distributed among models. This matters because CarinaNet+, the module that provides the corrected measurement, has a reported MAE of 0.94 cm on the MIMIC-CXR test set (Table 2), yet several updated model MAEs (e.g., GPT4V 0.39, CheXpertPlus 0.66) are lower than this. The reader cannot determine whether these numbers are computed on different subsets or whether the pipeline sometimes retains the original (inaccurate) measurement. Please report the case counts and explain the relationship between CarinaNet+'s MAE and the updated MAE values.
  4. [Abstract and Section 3.2] The claim that FactCheXcker 'maintains the quality of the original reports' is not supported by any evaluation. The paper evaluates ETT presence, measurement, and placement, but does not measure readability, clinical consistency, or preservation of non-ETT findings in the updated reports. The Report Updater prompt instructs the model to retain formatting and tone, but this is not a measurement. Please add a quantitative or human evaluation of updated-report quality, or revise the claim to reflect that only measurement-focused metrics were assessed.
  5. [Section 5.3 and Figure 3] The asterisks in Figure 3 indicate statistical significance with p<0.1, but the test is not described. It is unclear whether the comparisons are paired across the same studies, which test (e.g., Wilcoxon signed-rank) was used, and whether any multiple-comparison correction was applied across the 11 models. A threshold of p<0.1 is also unusually lenient for a claim of 'significant' improvement. Please specify the statistical procedure or remove the significance claims.
minor comments (6)
  1. [Section 5.2] The text states that LLM-CXR shows the most significant precision improvement 'with an increase of 1024%,' but the 1024% figure in Table 4 is the composite-score improvement; the precision improvement is from 0.08 to 0.58 (roughly a six-fold relative increase). Please correct this misattribution.
  2. [Table 4] The header 'Presense' is a typo for 'Presence'.
  3. [Section 4.4] The Composite metric MAE/F1 is unusual because a low F1 amplifies MAE; its interpretation and choice should be justified, and the improvement percentages for it suffer from the same denominator issue as the MAE improvements.
  4. [Section 4.4 and Figure 4] The failure threshold of 1.5 cm is introduced without clinical justification or sensitivity analysis; please cite a source or provide a sensitivity analysis.
  5. [Section 4.4] The assumption that a report with an ETT mention but no placement detail is 'correct' biases placement precision upward; please quantify this effect or use a more neutral handling.
  6. [Section 4.2] The fine-tuned CarinaNet+ uses a private dataset from 22 hospitals that is not released; given that this module is the measurement engine of the pipeline, a public version or detailed protocol would improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the corrected measurements come from independently evaluated tool modules and are compared against independent MIMIC-CXR ground truth; the self-citations are not load-bearing, and the inflated 135% headline is an arithmetic/denominator issue rather than a circular step.

full rationale

The paper's claimed derivation is that applying FactCheXcker to model-generated reports reduces ETT measurement hallucination, and the supporting evidence is an external comparison to MIMIC-CXR test ground truth (MIMIC-CXR reports plus radiologist annotations for 45 reports that lacked numeric values). The corrected values inserted into updated reports come from CarinaNet+, a separately fine-tuned module evaluated on the same test set (Table 2, MAE 0.94 cm), and the presence corrections come from ResNet-50+ (Table 3). No pipeline parameter is fitted to the 11 baseline models' outputs or to the test labels that define the headline MAE improvement, so the '10/11 models improve' claim does not reduce by construction to the input reports. The cited self-work (MedVersa [48], RadFM [42], and the knowledge-gap analysis [47]) provides baselines or motivation, but removing those citations would not change Table 4's comparisons against independent ground truth. The manuscript's own limitation statements (Section 5.5 and Figure 5: 20 failure cases with tube misidentification, obscured carina, and rotated images; Supplementary E: robustness stratification and CheXpert Plus validation) also treat CarinaNet+ and ResNet-50+ accuracy as an empirical ceiling rather than as a definitional identity. Separately, Table 4's 'Improvement' column is arithmetically nonstandard: it uses the updated error as the denominator (e.g., CheXagent, MAE 1.98 to 0.68, is reported as 191% instead of roughly 66% relative reduction), so the Abstract's 135.0% average is inflated by about a factor of three. That is a correctness or reporting concern, not circularity, and it does not affect the circularity score.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim depends on hand-chosen evaluation thresholds (1.5 cm failure cutoff, 3-7 cm optimal range, 0 cm imputation for missing measurements), on the transferability of private fine-tuned vision models (especially CarinaNet+), and on the unverified assumption that the LLM-based update preserves report quality. No new physical or conceptual entities are introduced.

free parameters (3)
  • Failure threshold for ETT measurement error = 1.5 cm
    Defines a failed ETT measurement in Figure 4. The threshold is hand-chosen and no clinical justification is given; reported failure rates depend on it.
  • Optimal ETT placement range = 3-7 cm (5 +/- 2 cm above carina)
    Used by the rule-based Report Updater to classify placement as correct or incorrect. Based on Goodman et al. [10], but applied as a hard threshold in evaluation; different institutions may use different criteria.
  • Missing measurement imputation value = 0 cm
    In MAE computation, if the model report does not include an ETT measurement but the ground truth does, the model value is set to 0 (Section 4.4). This hand-chosen value penalizes missed measurements and influences absolute MAE magnitudes.
assumptions (4)
  • domain assumption Ground-truth ETT measurements extracted from MIMIC-CXR reports by GPT-4o mini are accurate, and the 45 reports lacking measurements were correctly annotated by a radiologist.
    The evaluation metrics (MAE, placement precision) compare against these labels. No human validation of the GPT-4o mini extraction on the full test set is reported.
  • domain assumption CarinaNet+, fine-tuned on a private 1,100-image dataset from 22 hospitals, generalizes to MIMIC-CXR test images.
    The updated measurement in every corrected report is CarinaNet+'s output; the paper reports its MAE as 0.94 cm on MIMIC-CXR test (Table 2) but does not release data or weights.
  • ad hoc to paper The Report Updater preserves the quality and clinical meaning of the original report.
    The paper states quality is maintained but provides no quantitative report-quality evaluation before and after updating, such as BLEU, RadGraph, or clinician review.
  • domain assumption GPT-4o mini can generate correct measurement queries and updated report text.
    The authors note GPT-4o mini's numeric reasoning was insufficient for placement inference and replaced that step with rules (Section 3.2), indicating the LLM components are not fully reliable; failures (20 cases) include tube misidentification and obscured landmarks (Section 5.5).

how reviews work

0 comments
Cite this review

Pith. "Pith review of FactCheXcker: Mitigating Measurement Hallucinations in Chest X-ray Report Generation Models." pith.science (2026). https://pith.science/paper/IDYF42N5

@misc{pith2026241118672,
  author       = {Pith},
  title        = {Pith review of: FactCheXcker: Mitigating Measurement Hallucinations in Chest X-ray Report Generation Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IDYF42N5}},
  note         = {Machine review of arXiv:2411.18672}
}
read the original abstract

Medical vision-language models often struggle with generating accurate quantitative measurements in radiology reports, leading to hallucinations that undermine clinical reliability. We introduce FactCheXcker, a modular framework that de-hallucinates radiology report measurements by leveraging an improved query-code-update paradigm. Specifically, FactCheXcker employs specialized modules and the code generation capabilities of large language models to solve measurement queries generated based on the original report. After extracting measurable findings, the results are incorporated into an updated report. We evaluate FactCheXcker on endotracheal tube placement, which accounts for an average of 78% of report measurements, using the MIMIC-CXR dataset and 11 medical report-generation models. Our results show that FactCheXcker significantly reduces hallucinations, improves measurement precision, and maintains the quality of the original reports. Specifically, FactCheXcker improves the performance of 10/11 models and achieves an average improvement of 135.0% in reducing measurement hallucinations measured by mean absolute error. Code is available at https://github.com/rajpurkarlab/FactCheXcker.

Figures

Figures reproduced from arXiv: 2411.18672 by the authors.

Figure 1
Figure 1. An illustration of FactCheXcker pipeline, which uses a query-code-update approach to alleviate measurement hallucina￾tions in chest X-ray report generation models. ments, including determining the size of a lung nodule or measuring the distance from an endotracheal tube (ETT) to the carina [47]. Incorrect or missing measurements can lead to adverse clinical outcomes since many reporting guide￾lines rely on precise t… view at source ↗
Figure 2
Figure 2. Distribution of measurable findings across different [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Distribution of absolute measurement errors for ETT placement across different models before and after using FactCheXcker. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Comparison of failure rates for ETT placement across different models before and after using FactCheXcker. A case is considered [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: FactCheXcker Case Studies. (Left) Given an input image and an initial model-generated report, FactCheXcker generates queries for the code generator, which synthesizes a program. This program is then executed using a Python interpreter to produce the output. The update …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 21 canonical work pages

  1. [1]

    Hallucination of multimodal large language models: A survey

    Zechen Bai, Pichao Wang, Tianjun Xiao, Tong He, Zongbo Han, Zheng Zhang, and Mike Zheng Shou. Hallucination of multimodal large language models: A survey. arXiv preprint arXiv:2404.18930, 2024. 3

  2. [2]

    Maira-2: Grounded radiology report gen- eration

    Shruthi Bannur, Kenza Bouzid, Daniel C Castro, Anton Schwaighofer, Sam Bond-Taylor, Maximilian Ilse, Fernando P´erez-Garc´ıa, Valentina Salvatelli, Harshita Sharma, Felix Meissen, et al. Maira-2: Grounded radiology report gen- eration. arXiv preprint arXiv:2406.04449, 2024. 1, 3, 5, 7

  3. [3]

    Chexpert plus: Hundreds of thousands of aligned radiology texts, im- ages and patients

    Pierre Chambon, Jean-Benoit Delbrouck, Thomas Sounack, Shih-Cheng Huang, Zhihong Chen, Maya Varma, Steven QH Truong, Chu The Chuong, and Curtis P Langlotz. Chexpert plus: Hundreds of thousands of aligned radiology texts, im- ages and patients. arXiv preprint arXiv:2405.19538, 2024. 3, 5, 7

  4. [4]

    Cross-modal causal intervention for medical report generation

    Weixing Chen, Yang Liu, Ce Wang, Jiarui Zhu, Shen Zhao, Guanbin Li, Cheng-Lin Liu, and Liang Lin. Cross-modal causal intervention for medical report generation. arXiv preprint arXiv:2303.09117, 2023. 3, 5, 7

  5. [5]

    Chexagent: Towards a foun- dation model for chest x-ray interpretation

    Zhihong Chen, Maya Varma, Jean-Benoit Delbrouck, Mag- dalini Paschali, Louis Blankemeier, Dave Van Veen, Jeya Maria Jose Valanarasu, Alaa Youssef, Joseph Paul Cohen, Eduardo Pontes Reis, et al. Chexagent: Towards a foun- dation model for chest x-ray interpretation. arXiv preprint arXiv:2401.12208, 2024. 3, 5, 7

  6. [6]

    Factool: Factuality detection in generative ai–a tool augmented framework for multi-task and multi-domain scenarios

    I Chern, Steffi Chern, Shiqi Chen, Weizhe Yuan, Kehua Feng, Chunting Zhou, Junxian He, Graham Neubig, Pengfei Liu, et al. Factool: Factuality detection in generative ai–a tool augmented framework for multi-task and multi-domain scenarios. arXiv preprint arXiv:2307.13528, 2023. 3

  7. [7]

    Viviano, Paul Bertin, Paul Morrison, Parsa Torabian, Matteo Guarrera, Matthew P

    Joseph Paul Cohen, Joseph D. Viviano, Paul Bertin, Paul Morrison, Parsa Torabian, Matteo Guarrera, Matthew P. Lungren, Akshay Chaudhari, Rupert Brooks, Moham- mad Hashir, and Hadrien Bertrand. TorchXRayVision: A library of chest X-ray datasets and models, 2021. arXiv:2111.00595. 5

  8. [8]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009. 4

Show all 52 references
  1. [9]

    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. 3

  2. [10]

    L. R. Goodman, P. A. Conrardy, F. Laing, and M. M. Singer. Radiographic evaluation of endotracheal tube position. AJR. American journal of roentgenology , 127(3):433–434, 1976. 4

  3. [11]

    Detecting and preventing hallucinations in large vision language models

    Anisha Gunjal, Jihan Yin, and Erhan Bas. Detecting and preventing hallucinations in large vision language models. In Proceedings of the AAAI Conference on Artificial Intelli- gence, pages 18135–18143, 2024. 3

  4. [12]

    Compli- cations of managing the airway

    Carin Hagberg, Rainer Georgi, and Claude Krier. Compli- cations of managing the airway. Best Practice & Research Clinical Anaesthesiology, 19(4):641–659, 2005. 2

  5. [13]

    Deep Residual Learning for Image Recognition, 2015

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep Residual Learning for Image Recognition, 2015. arXiv:1512.03385. 4

  6. [14]

    Gpt-4o system card

    Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perel- man, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Weli- hinda, Alan Hayes, Alec Radford, et al. Gpt-4o system card. arXiv preprint arXiv:2410.21276, 2024. 4

  7. [15]

    Maira-1: A specialised large multi- modal model for radiology report generation

    Stephanie L Hyland, Shruthi Bannur, Kenza Bouzid, Daniel C Castro, Mercy Ranjit, Anton Schwaighofer, Fer- nando P ´erez-Garc´ıa, Valentina Salvatelli, Shaury Srivastav, Anja Thieme, et al. Maira-1: A specialised large multi- modal model for radiology report generation. arXiv p...

  8. [16]

    Survey of hallucination in natural language generation

    Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. Survey of hallucination in natural language generation. ACM Computing Surveys, 55(12):1–38, 2023. 3

  9. [17]

    Mimic-cxr, a de- identified publicly available database of chest radiographs with free-text reports

    Alistair EW Johnson, Tom J Pollard, Seth J Berkowitz, Nathaniel R Greenbaum, Matthew P Lungren, Chih-ying Deng, Roger G Mark, and Steven Horng. Mimic-cxr, a de- identified publicly available database of chest radiographs with free-text reports. Scientific data , 6(1):317, 2019. 2, 4

  10. [18]

    Bert: Pre-training of deep bidirectional trans- formers for language understanding

    Jacob Devlin Ming-Wei Chang Kenton and Lee Kristina Toutanova. Bert: Pre-training of deep bidirectional trans- formers for language understanding. In Proceedings of naacL-HLT, page 2, 2019. 3, 7

  11. [19]

    Llm-cxr: Instruction-finetuned llm for cxr image under- standing and generation

    Suhyeon Lee, Won Jun Kim, Jinho Chang, and Jong Chul Ye. Llm-cxr: Instruction-finetuned llm for cxr image under- standing and generation. arXiv preprint arXiv:2305.11490,

  12. [20]

    Api-bank: A comprehensive benchmark for tool-augmented llms

    Minghao Li, Yingxiu Zhao, Bowen Yu, Feifan Song, Hangyu Li, Haiyang Yu, Zhoujun Li, Fei Huang, and Yongbin Li. Api-bank: A comprehensive benchmark for tool-augmented llms. arXiv preprint arXiv:2304.08244, 2023. 3

  13. [21]

    Evaluating object hallucina- tion in large vision-language models

    Yifan Li, Yifan Du, Kun Zhou, Jinpeng Wang, Wayne Xin Zhao, and Ji-Rong Wen. Evaluating object hallucina- tion in large vision-language models. arXiv preprint arXiv:2305.10355, 2023. 3, 5

  14. [22]

    Taskmatrix

    Yaobo Liang, Chenfei Wu, Ting Song, Wenshan Wu, Yan Xia, Yu Liu, Yang Ou, Shuai Lu, Lei Ji, Shaoguang Mao, et al. Taskmatrix. ai: Completing tasks by connecting foun- dation models with millions of apis. Intelligent Computing, 3:0063, 2024. 3

  15. [23]

    Mitigating hallucination in large multi-modal models via robust instruction tuning

    Fuxiao Liu, Kevin Lin, Linjie Li, Jianfeng Wang, Yaser Ya- coob, and Lijuan Wang. Mitigating hallucination in large multi-modal models via robust instruction tuning. In The Twelfth International Conference on Learning Representa- tions, 2023. 3 9

  16. [24]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF international conference on computer vision, pages 10012–10022, 2021. 3

  17. [25]

    Swin transformer v2: Scaling up capacity and resolution

    Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, et al. Swin transformer v2: Scaling up capacity and resolution. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 12009–12019,...

  18. [26]

    Decoupled weight decay regularization

    I Loshchilov. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017. 7

  19. [27]

    Negative object presence evaluation (nope) to measure object hallucination in vision-language models

    Holy Lovenia, Wenliang Dai, Samuel Cahyawijaya, Ziwei Ji, and Pascale Fung. Negative object presence evaluation (nope) to measure object hallucination in vision-language models. arXiv preprint arXiv:2310.05338, 2023. 3

  20. [28]

    Factscore: Fine-grained atomic evalu- ation of factual precision in long form text generation

    Sewon Min, Kalpesh Krishna, Xinxi Lyu, Mike Lewis, Wen- tau Yih, Pang Wei Koh, Mohit Iyyer, Luke Zettlemoyer, and Hannaneh Hajishirzi. Factscore: Fine-grained atomic evalu- ation of factual precision in long form text generation. arXiv preprint arXiv:2305.14251, 2023. 3

  21. [29]

    Matthieu Oliver, Am ´elie Renou, Nicolas Allou, Lucas Moscatelli, Cyril Ferdynus, and Jer ˆome Allyn. Image augmentation and automated measurement of endotracheal- tube-to-carina distance on chest radiographs in intensive care unit using a deep learning model with external val...

  22. [30]

    Fact-checking complex claims with program-guided reason- ing

    Liangming Pan, Xiaobao Wu, Xinyuan Lu, Anh Tuan Luu, William Yang Wang, Min-Yen Kan, and Preslav Nakov. Fact-checking complex claims with program-guided reason- ing. arXiv preprint arXiv:2305.12744, 2023. 3

  23. [31]

    Radialog: A large vision- language model for radiology report generation and conver- sational assistance

    Chantal Pellegrini, Ege ¨Ozsoy, Benjamin Busam, Nassir Navab, and Matthias Keicher. Radialog: A large vision- language model for radiology report generation and conver- sational assistance. arXiv preprint arXiv:2311.18681, 2023. 5, 7

  24. [32]

    Object hallucination in image cap- tioning

    Anna Rohrbach, Lisa Anne Hendricks, Kaylee Burns, Trevor Darrell, and Kate Saenko. Object hallucination in image cap- tioning. arXiv preprint arXiv:1809.02156, 2018. 5

  25. [33]

    Toolformer: Lan- guage models can teach themselves to use tools

    Timo Schick, Jane Dwivedi-Yu, Roberto Dess `ı, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. Toolformer: Lan- guage models can teach themselves to use tools. Advances in Neural Information Processing Systems, 36, 2024. 3

  26. [34]

    Hugginggpt: Solving ai tasks with chatgpt and its friends in hugging face

    Yongliang Shen, Kaitao Song, Xu Tan, Dongsheng Li, Weiming Lu, and Yueting Zhuang. Hugginggpt: Solving ai tasks with chatgpt and its friends in hugging face. Advances in Neural Information Processing Systems, 36, 2024. 3

  27. [35]

    Attention based automated radiology report generation using cnn and lstm

    Mehreen Sirshar, Muhammad Faheem Khalil Paracha, Muhammad Usman Akram, Norah Saleh Alghamdi, Syeda Zainab Yousuf Zaidi, and Tatheer Fatima. Attention based automated radiology report generation using cnn and lstm. Plos one, 17(1):e0262209, 2022. 3

  28. [36]

    Automated radiology report generation: A review of recent advances

    Phillip Sloan, Philip Clatworthy, Edwin Simpson, and Majid Mirmehdi. Automated radiology report generation: A review of recent advances. IEEE Reviews in Biomedical Engineer- ing, 2024. 1

  29. [37]

    Smith and Nicholay Topin

    Leslie N. Smith and Nicholay Topin. Super-Convergence: Very Fast Training of Neural Networks Using Large Learn- ing Rates, 2018. arXiv:1708.07120. 5, 7

  30. [38]

    Vipergpt: Visual inference via python execution for reasoning

    D ´ıdac Sur´ıs, Sachit Menon, and Carl V ondrick. Vipergpt: Visual inference via python execution for reasoning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 11888–11898, 2023. 3

  31. [39]

    Interactive and explainable region-guided radiol- ogy report generation

    Tim Tanida, Philip M ¨uller, Georgios Kaissis, and Daniel Rueckert. Interactive and explainable region-guided radiol- ogy report generation. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 7433–7442, 2023. 3, 5, 7

  32. [40]

    Towards gen- eralist biomedical ai

    Tao Tu, Shekoofeh Azizi, Danny Driess, Mike Schaeker- mann, Mohamed Amin, Pi-Chuan Chang, Andrew Carroll, Charles Lau, Ryutaro Tanno, Ira Ktena, et al. Towards gen- eralist biomedical ai. NEJM AI, 1(3):AIoa2300138, 2024. 3

  33. [41]

    Mitigating hallucinations in large vision-language models with instruction contrastive decoding

    Xintong Wang, Jingheng Pan, Liang Ding, and Chris Bie- mann. Mitigating hallucinations in large vision-language models with instruction contrastive decoding. arXiv preprint arXiv:2403.18715, 2024. 3

  34. [42]

    Towards generalist foundation model for radiol- ogy

    Chaoyi Wu, Xiaoman Zhang, Ya Zhang, Yanfeng Wang, and Weidi Xie. Towards generalist foundation model for radiol- ogy. arXiv preprint arXiv:2308.02463, 2023. 3, 5, 7

  35. [43]

    Chest imagenome dataset for clinical reasoning

    Joy T Wu, Nkechinyere N Agu, Ismini Lourentzou, Arjun Sharma, Joseph A Paguio, Jasper S Yao, Edward C Dee, William Mitchell, Satyananda Kashyap, Andrea Giovannini, et al. Chest imagenome dataset for clinical reasoning. arXiv preprint arXiv:2108.00316, 2021. 7

  36. [44]

    Multimodal recurrent model with attention for automated radiology re- port generation

    Yuan Xue, Tao Xu, L Rodney Long, Zhiyun Xue, Sameer Antani, George R Thoma, and Xiaolei Huang. Multimodal recurrent model with attention for automated radiology re- port generation. In Medical Image Computing and Com- puter Assisted Intervention–MICCAI 2018: 21st Interna- tion...

  37. [45]

    The dawn of lmms: Preliminary explorations with gpt-4v (ision)

    Zhengyuan Yang, Linjie Li, Kevin Lin, Jianfeng Wang, Chung-Ching Lin, Zicheng Liu, and Lijuan Wang. The dawn of lmms: Preliminary explorations with gpt-4v (ision). arXiv preprint arXiv:2309.17421, 9(1):1, 2023. 5, 7

  38. [46]

    Multimodal healthcare ai: identifying and designing clinically relevant vision-language applications for radiology

    Nur Yildirim, Hannah Richardson, Maria Teodora Wetscherek, Junaid Bajwa, Joseph Jacob, Mark Ames Pinnock, Stephen Harris, Daniel Coelho De Castro, Shruthi Bannur, Stephanie Hyland, et al. Multimodal healthcare ai: identifying and designing clinically relevant vision-language a...

  39. [47]

    Uncovering knowledge gaps in radiology report generation models through knowledge graphs

    Xiaoman Zhang, Juli ´an N Acosta, Hong-Yu Zhou, and Pranav Rajpurkar. Uncovering knowledge gaps in radiology report generation models through knowledge graphs. arXiv preprint arXiv:2408.14397, 2024. 1

  40. [48]

    Measure the distance between the carina and the endotracheal tube

    Hong-Yu Zhou, Subathra Adithan, Juli ´an Nicol ´as Acosta, Eric J Topol, and Pranav Rajpurkar. A generalist learner for multifaceted medical image interpretation. arXiv preprint arXiv:2405.07988, 2024. 1, 3, 5, 7 10 FactCheXcker: Mitigating Measurement Hallucinations in Chest ...

  41. [49]

    Note that mentions of ET tube removal or patient extubation indicate that the ET tube is no longer present

    Identify ET Tube Present: Determine if the report explicitly states that an ET tube is present. Note that mentions of ET tube removal or patient extubation indicate that the ET tube is no longer present

  42. [50]

    Positive values indicate placement above the carina, while negative values indicate placement below the carina

    Extract ET Tube Measurement: If an ET tube is present, extract its relative distance to the carina in centimeters (cm) if specified. Positive values indicate placement above the carina, while negative values indicate placement below the carina

  43. [51]

    too low" or

    Determine ET Tube Placement: If an ET tube is present, determine if the report deems the placement correct or incorrect. If incorrect, categorize the placement as "too low" or "too high," if possible. If you cannot extract a specific category, use "null". There is no need to g...

  44. [52]

    incorrect position

    Identify measured object(s): list all objects that the reports include measurements for using concrete measurements in centimeters (cm) and/or millimeters (mm). Adhere to the following rules: - Do not include objects specified in qualitative descriptions or anatomical landmark...

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.