Pith. sign in

REVIEW 4 major objections 5 minor 55 references

FG-CXR: A Radiologist-Aligned Gaze Dataset for Enhancing Interpretability in Chest X-Ray Report Generation

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper introduces FG-CXR, a chest X-ray gaze dataset with per-anatomy alignment between report sentences and radiologist attention heatmaps, and Gen-XAI, a generator that it reports surpasses all baselines on language, clinical, and…

desk verdict A useful per-region gaze-report dataset, but the accuracy gains over baselines are confounded by template-generated report ground truth. read the letter →

arxiv 2411.15413 v1 pith:CT2VKRQB submitted 2024-11-23 cs.CV cs.AI

Trong Thang Pham , Ngoc-Vuong Ho , Nhat-Tan Bui , Thinh Phan , Patel Brijesh , Donald Adjeroh , Gianfranco Doretto , Anh Nguyen
show 3 more authors
Carol C. Wu Hien Nguyen Ngan Le
This is my paper · ORCID
classification cs.CVcs.AI
keywords chestX-rayreportgenerationgazeattentioninterpretabilityeye-trackingdatasetradiologyanatomicalheatmapsexplainableAImedicalimaging
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

This paper is trying to establish that radiologists' eye movements can be turned into structured supervision that makes automated chest X-ray report generation both more accurate and more interpretable. It introduces FG-CXR, a curated dataset of 2,951 chest X-rays with 20,657 anatomy-level pairs of gaze attention heatmaps and report sentences, aligned by using sentence timing to trim gaze streams and anatomical masks to discard off-target gaze. It then proposes Gen-XAI, a generator with a gaze attention predictor that outputs seven anatomy heatmaps and a decoder that conditions on those heatmaps. In experiments on FG-CXR, Gen-XAI reportedly outperforms five strong captioning baselines on all natural-language, clinical-efficacy, and attention-similarity metrics. If correct, the paper shows that explicit gaze supervision can replace implicit attention in radiology report generation, giving clinicians both a better report and a visual record of what the model looked at.

What carries the argument

The load-bearing mechanism is the anatomical-aware gaze attention pipeline used to build FG-CXR and the Gaze Attention Predictor (GAP) used to train on it. The curation pipeline takes a raw eye-tracking stream plus a report, finds the latest end time of sentences mentioning each anatomy via keyword matching, removes gaze points after that time, then drops points outside the automatically generated anatomical mask; the surviving points are rendered into a Gaussian-blurred heatmap. During training, the GAP predicts seven such heatmaps from the image and seven anatomical intention tokens, a spatial-aware attended encoder multiplies latent visual features by the predicted maps, and an autoregressive report decoder cross-attends to the reweighted features. This forces the model to ground each generated sentence in a region a radiologist actually looked at, and the two dynamic penalty terms push it to look at the correct anatomy and mention the correct direction.

What would settle it

Re-record eye tracking and transcripts under a protocol that marks each anatomy's examination interval explicitly, or have a radiologist manually segment existing gaze streams in a random subset; if a model trained on those manually aligned heatmaps does not reproduce Gen-XAI's margin over baselines, the temporal trimming in Section 3.2 is the source of the reported advantage.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that a report-generation model explicitly constrained to predict and attend to radiologist gaze heatmaps produces reports that are linguistically closer to radiologist transcripts and clinically more correct than black-box captioning models, while remaining interpretable because its attention is anchored to expert behavior. Anchoring this claim is the FG-CXR dataset, which re-pairs gaze with reports at the level of individual anatomies: each of seven regions (heart, left, right, upper and lower left and right lungs) has its own gaze heatmap and its own sentence or sentences. On this benchmark, the paper reports that Gen-XAI surpasses all compared methods on every reported metric, with its largest margin in attention similarity. The paper interprets this as direct evidence that mimicking radiologists' viewing focus is better than letting the model discover its own attention for both accuracy and explainability.

Load-bearing premise

The load-bearing premise is that a radiologist never returns to an anatomy after dictating the sentence about it, so trimming each gaze stream at the latest sentence end produces clean per-anatomy attention; if radiologists do circle back, the dataset's gaze labels are noisy and the reported gains may come from the curation pipeline rather than from gaze itself.

Editorial extensions

If this is right

  • Radiologist gaze can serve as a direct supervision signal for report generation, not just an evaluation tool, so future systems can be trained to look where experts look.
  • Per-anatomy heatmaps give clinicians a visual audit trail: each generated sentence is tied to a region the model was looking at, which makes errors in reports traceable to errors in attention.
  • The FG-CXR benchmark allows interpretability to be quantified as agreement with expert gaze, alongside traditional language and clinical metrics.
  • The model's gaze attention module can be repurposed for anatomical segmentation, since the same predictor trained on segmentation masks performs competitively.

Reading between the lines

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

  • Inference: A direct extension would be to apply the same sentence-timing and mask-filtering curation to other eye-tracking corpora, producing per-anatomy heatmaps without new gaze recordings; the value of that extension depends entirely on the revisit assumption named below.
  • Inference: The architecture suggests a testable variant where predicted heatmaps are optimized jointly with the report through a differentiable attention-similarity reward rather than fixed penalty coefficients, which could tighten the coupling between gaze and text.
  • Inference: Because missing anatomies are filled with templated sentences from external label sets, the report distribution of FG-CXR is partly synthetic; comparing against a subset containing only original dictated sentences would clarify how much of the reported report-quality gain comes from gaze supervision versus the templates.
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

4 major / 5 minor

Summary. The paper introduces FG-CXR, a curated dataset that aligns radiologist eye-tracking data from EGD and REFLACX with per-anatomy reports and attention heatmaps for seven chest X-ray regions. It also proposes Gen-XAI, a three-module system consisting of a gaze attention predictor built on BiomedCLIP, a spatial-aware attended encoder, and a GPT-2 decoder, trained with heatmap and report losses plus dynamic penalty coefficients. Experiments on a 70/10/20 split of 2,951 images report state-of-the-art results on natural language generation, clinical efficacy, and attention similarity metrics. The dataset and a checkpoint are publicly released.

Significance. If the curation pipeline is sound, FG-CXR is a valuable resource for gaze-interpretable report generation and for benchmarking anatomical segmentation and report generation in chest X-ray analysis, and Gen-XAI is a reasonable first architecture for this new task. The paper ships a public dataset and checkpoint, and the ablations in Tables 7-9 are informative, particularly the comparison between pixel-level and feature-level attention and the penalty-term study. However, the synthetic template reports and the uncontrolled attention comparison currently leave the central claims of improved accuracy and interpretability conditional on additional validation.

major comments (4)
  1. [Section 3.3, Tables 4-5] The report ground truth is partly synthetic. Section 3.3 states that for any anatomy not mentioned in the original REFLACX or EGD report, the authors generate a template sentence from MIMIC-CXR labels, e.g., 'the {area} is possibly normal' or 'the patient is possibly suffering from {findings} in the {area}'. The paper does not report the proportion of the 20,657 region-report pairs that are synthetic, nor does it validate the MIMIC-CXR labels for the specific EGD/REFLACX images. Since the proposed decoder is explicitly conditioned on anatomical intention tokens that mirror the template structure, the NLG and CE gains in Tables 4 and 5 may reflect template memorization rather than the value of radiologist gaze. This is load-bearing for the central claim of more accurate report generation, and the internal ablation in Table 9 does not address it because all settings are trained and evaluated on the same synthetic ground truth.
  2. [Section 5.2, Table 6] The attention similarity comparison is not controlled. Gen-XAI is explicitly trained to minimize an L2 loss against radiologist gaze heatmaps, while the baseline models are never trained to predict gaze and are evaluated on gaze metrics using their internal attention scores or bounding boxes. The large fwIoU advantage (+20.47 over the runner-up) is therefore structurally expected and does not by itself establish that Gen-XAI is more interpretable. The claim that the method 'outperforms other methods across all criteria' conflates being trained to match gaze with being more interpretable. I recommend either training baselines with the same gaze supervision, or evaluating interpretability through a radiologist study or a task-based evaluation that does not use the training objective as the metric.
  3. [Section 3.2] The temporal alignment assumption is not validated. The curation pipeline removes all gaze points after the latest end time of sentences containing anatomy-related keywords, which assumes that a radiologist finishes examining a region before dictating the corresponding sentence and never returns to that region. It also assumes that the automatically generated anatomical masks from [34] are accurate, since gaze points falling outside a mask are discarded. If either assumption fails, the resulting 'fine-grained' gaze heatmaps are noisy, and the reported gains from gaze attention could be artifacts of the curation pipeline. The paper should provide a sensitivity analysis or validate the alignment on a subset of cases, for example by comparing the automatically aligned heatmaps with manually aligned ones.
  4. [Section 3.4, Section 5.2] All quantitative conclusions rest on a single random 70/10/20 split of 582 test images with point estimates and no confidence intervals or multiple-seed results. Given the relatively small test set and the modest margins on some clinical efficacy metrics (e.g., F1ex 0.497 vs. 0.440 in Table 5), the paper should report standard deviations across at least three runs or use a statistical significance test to show that the improvements are robust. Without this, the headline comparisons in Tables 4-6 are not yet fully supported.
minor comments (5)
  1. [Section 4.4, Eq. (1)] The notation '10.5' in Eq. (1) is confusing; replace it with the indicator function \mathbb{1} and clarify that the threshold is 0.5. Also, in the sentence 'we want the model the model explicitly predict directions' there is a duplicated phrase 'the model'.
  2. [Section 5.1, Table 4] The metric R@4 is reported in Table 4 but is not defined in the metrics list in Section 5.1; please provide a definition or citation. Similarly, the Div@2 definition is cited but not stated.
  3. [Section 5.1, Section 4.3] The paper says GPT2 is initialized with DistillGPT2 and cites [39] (Sanh et al.), but reference [39] is the DistilBERT paper; please cite the correct DistilGPT2 source.
  4. [Table 1] The last column of Table 1, 'Method', contains entries such as 'Semi-automated' and 'Automated' that appear to describe how annotations were created, but the column header is unclear. Please rename the column to 'Annotation Method' and ensure the row entries are aligned with the correct columns.
  5. [Section 3.1] The paper filters out images with 'extreme brightness' but does not define the threshold; since this is a free parameter in the curation pipeline, please state the criterion or cite the procedure.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Gen-XAI's gains are measured on a held-out test split under standard supervised objectives, not forced by construction.

full rationale

The paper's central claim, that gaze-conditioned report generation improves accuracy and interpretability, is empirically grounded rather than circular. The Gaze Attention Predictor is trained with an L2 loss on radiologist gaze heatmaps and the Report Decoder with cross-entropy on reports; both are conventional supervised objectives evaluated on a held-out 20% test split (Table 3). The attention-similarity metrics compare predicted heatmaps to the same type of ground truth used for training, but this is a legitimate regression evaluation on unseen data, not a derivation that equals its input by construction; importantly, the baselines are not trained on gaze labels, so the head-to-head comparison is meaningful. The GAP architecture and mask generation borrow from the authors' prior WACV paper [34], but this is transparent component reuse with independent published support, not a load-bearing self-citation chain or an imported uniqueness claim. The template-filled reports in Section 3.3, which use MIMIC-CXR labels to create default sentences for missing anatomies, are a benchmark-validity concern because the fraction of synthetic sentences is unreported and NLG/CE metrics are computed against these references; however, this does not make the model's predictions equal to its inputs by construction, and all compared methods are trained on the same dataset. No equation in the paper reduces to its own input, and no fitted parameter is renamed as a prediction. Hence no significant circularity is present.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claims depend on assumptions about gaze-report alignment, the sufficiency of seven anatomical regions, the reliability of reused masks, and the validity of template-filled reports. No new physical entities are introduced, and the only interactive free parameters are the IoU threshold, the keyword set, and the unspecified brightness filter.

free parameters (3)
  • IoU threshold for penalty coefficient = 0.5
    Used in Equation 1 to decide when a predicted heatmap is considered incorrect, increasing lambda_h. The choice of 0.5 is not justified and affects training dynamics.
  • Keyword set for anatomical regions = Table 2
    Manually chosen by inspecting report sentences; the selection determines which sentences are assigned to which anatomy and influences all dataset statistics.
  • Extreme brightness filter threshold = not specified
    Images with extreme brightness are removed before dataset construction, but the threshold is nowhere defined, so the filtering step is not reproducible.
assumptions (5)
  • domain assumption Radiologist gaze attention is a reliable proxy for diagnostic relevance.
    The entire pipeline is motivated by the hypothesis in Section 1 that understanding gaze patterns can improve AI explanation and accuracy; no independent evidence is given that eye position causally determines the report content.
  • domain assumption Seven anatomical regions fully partition the radiologist's diagnostic focus.
    Section 3.1 states that radiologists focus on seven key areas, but no study is cited to support this exact partition, and the lower-left and lower-right masks are acknowledged to be inaccurate due to heart occlusion and diaphragm fixation.
  • ad hoc to paper Latest end time of an anatomy-related sentence bounds the gaze fixation for that anatomy.
    The core alignment rule in Section 3.2 is introduced specifically for this dataset and has no published validation.
  • ad hoc to paper Template sentences generated from MIMIC-CXR labels are valid ground truth for missing anatomy reports.
    Section 3.3 fills absent diagnoses with 'possibly normal' or 'possibly suffering from {findings}' based on MIMIC-CXR labels; the clinical validity of these synthetic sentences is not evaluated.
  • domain assumption The anatomical mask generation of [34] transfers correctly to EGD and REFLACX images.
    Section 3.1 reuses masks from [34] without retraining or validation on the new image distributions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FG-CXR: A Radiologist-Aligned Gaze Dataset for Enhancing Interpretability in Chest X-Ray Report Generation." pith.science (2026). https://pith.science/paper/CT2VKRQB

@misc{pith2026241115413,
  author       = {Pith},
  title        = {Pith review of: FG-CXR: A Radiologist-Aligned Gaze Dataset for Enhancing Interpretability in Chest X-Ray Report Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CT2VKRQB}},
  note         = {Machine review of arXiv:2411.15413}
}
read the original abstract

Developing an interpretable system for generating reports in chest X-ray (CXR) analysis is becoming increasingly crucial in Computer-aided Diagnosis (CAD) systems, enabling radiologists to comprehend the decisions made by these systems. Despite the growth of diverse datasets and methods focusing on report generation, there remains a notable gap in how closely these models' generated reports align with the interpretations of real radiologists. In this study, we tackle this challenge by initially introducing Fine-Grained CXR (FG-CXR) dataset, which provides fine-grained paired information between the captions generated by radiologists and the corresponding gaze attention heatmaps for each anatomy. Unlike existing datasets that include a raw sequence of gaze alongside a report, with significant misalignment between gaze location and report content, our FG-CXR dataset offers a more grained alignment between gaze attention and diagnosis transcript. Furthermore, our analysis reveals that simply applying black-box image captioning methods to generate reports cannot adequately explain which information in CXR is utilized and how long needs to attend to accurately generate reports. Consequently, we propose a novel explainable radiologist's attention generator network (Gen-XAI) that mimics the diagnosis process of radiologists, explicitly constraining its output to closely align with both radiologist's gaze attention and transcript. Finally, we perform extensive experiments to illustrate the effectiveness of our method. Our datasets and checkpoint is available at https://github.com/UARK-AICV/FG-CXR.

Figures

Figures reproduced from arXiv: 2411.15413 by the authors.

Figure 1
Figure 1. An overview of our interpretable Gen-XAI framework, generating a diagnosis report and its corresponding visual attention for each diagnosis in the report. 2 Related works 2.1 Interpretable Deep Learning. In high-stakes medical settings, understanding the decision-making process is crucial [38]. A direction to enhance interpretability is to design an architecture that can learn concepts [20, 34]. In our paper, we fol… view at source ↗
Figure 2
Figure 2. Annotation comparison between prior gaze-CXR datasets (left) which face challenges in aligning gaze location with textual description and our FG-CXR (right), given a CXR (middle) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Dataset distributions for FG-CXR. create a default sentence based on MIMIC-CXR annotations: “the {area} is possibly normal” for no findings, or “the patient is possibly suffering from {findings} in the {area}” for specific findings. For example, if a pa￾tient’s current report lacks information for the left lung area, we refer to MIMIC￾CXR and find that the label for this patient is “no finding”. We then generate the… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The detailed architecture of our framework consisting of three key modules: Gaze Attention Predictor, Spatial-Aware Attended Encoder, and Report Decoder. and generate a report R of the attended region A. To ensure interpretability, the gaze attention A has to be closel…
Figure 5
Figure 5. Figure 5: Qualitative comparison. The blue text highlights consistent content with ground truth. The red text indicates incorrect content. Metrics. Follow [34, 41, 43], we evaluate our method based on three criteria: – Natural Language Generation (NLG) metrics: BLEU (B), METEOR …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 50 canonical work pages

  1. [34]

    In: WACV (2024) 2, 3, 5, 8, 9, 10, 11

    Pham, T.T., Brecheisen, J., Nguyen, A., Nguyen, H., Le, N.: I-ai: A controllable & interpretable ai system for decoding radiologists’ intense focus for accurate cxr diagnoses. In: WACV (2024) 2, 3, 5, 8, 9, 10, 11

    work page 2024

  2. [1]

    Scientific data (2022) 2, 4, 5, 7

    Bigolin Lanfredi, R., Zhang, M., et al.: Reflacx, a dataset of reports and eye- tracking data for localization of abnormalities in chest x-rays. Scientific data (2022) 2, 4, 5, 7

    work page 2022

  3. [2]

    Medical image anal- ysis (2020) 4

    Bustos, A., Pertusa, A., Salinas, J.M., de la Iglesia-Vayá, M.: Padchest: A large chest x-ray image dataset with multi-label annotated reports. Medical image anal- ysis (2020) 4

    work page 2020

  4. [3]

    arXiv preprint arXiv:2204.13258 (2022) 10, 11, 12

    Chen, Z., Shen, Y., Song, Y., Wan, X.: Cross-modal memory networks for radiology report generation. arXiv preprint arXiv:2204.13258 (2022) 10, 11, 12

    arXiv 2022

  5. [4]

    arXiv preprint arXiv:2010.16056 (2020) 10, 11

    Chen, Z., Song, Y., Chang, T.H., Wan, X.: Generating radiology reports via memory-driven transformer. arXiv preprint arXiv:2010.16056 (2020) 10, 11

    arXiv 2020

  6. [5]

    CattleFace-RGBT: RGB-T Cattle Facial Landmark Benchmark

    Coffman, E., Clark, R., Bui, N.T., Pham, T.T., Kegley, B., Powell, J.G., Zhao, J., Le, N.: Cattleface-rgbt: Rgb-t cattle facial landmark benchmark. arXiv preprint arXiv:2406.03431 (2024) 2

    work page Pith review arXiv 2024

  7. [6]

    In: CVPR (2020) 10, 11, 12

    Cornia, M., Stefanini, M., Baraldi, L., Cucchiara, R.: Meshed-Memory Transformer for Image Captioning. In: CVPR (2020) 10, 11, 12

    work page 2020

  8. [7]

    Data in Brief (2020) 4

    Datta, S., Roberts, K.: A dataset of chest x-ray reports annotated with spatial role labeling annotations. Data in Brief (2020) 4

    work page 2020

Show all 55 references
  1. [8]

    Journal of the American Medical Informatics Association (2016) 4

    Demner-Fushman, D., Kohli, M.D., Rosenman, M.B., Shooshan, S.E., Rodriguez, L., Antani, S., Thoma, G.R., McDonald, C.J.: Preparing a collection of radiol- ogy examinations for distribution and retrieval. Journal of the American Medical Informatics Association (2016) 4

    2016

  2. [9]

    In: CVPR (2009) 10

    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: A large-scale hierarchical image database. In: CVPR (2009) 10

    2009

  3. [10]

    Journal of digital imaging (2020) 3, 4

    Filice, R.W., Stein, A., et al.: Crowdsourcing pneumothorax annotations using machine learning annotations on the nih chest x-ray dataset. Journal of digital imaging (2020) 3, 4

    2020

  4. [11]

    Radiology293(2), 436–440 (2019) 2

    Geis, J.R., Brady, A.P., Wu, C.C., Spencer, J., Ranschaert, E., Jaremko, J.L., Langer, S.G., Borondy Kitts, A., Birch, J., Shields, W.F., et al.: Ethics of artifi- cial intelligence in radiology: summary of the joint european and north american multisociety statement. Radiolog...

    2019

  5. [12]

    ACM computing surveys (CSUR) 51(5), 1–42 (2018) 2

    Guidotti, R., Monreale, A., Ruggieri, S., Turini, F., Giannotti, F., Pedreschi, D.: A survey of methods for explaining black box models. ACM computing surveys (CSUR) 51(5), 1–42 (2018) 2

    2018

  6. [13]

    In: AAAI (2019) 4, 11

    Irvin, J., Rajpurkar, P., Ko, M., Yu, Y., Ciurea-Ilcus, S., Chute, C., Marklund, H., Haghgoo, B., Ball, R., Shpanskaya, K., et al.: Chexpert: A large chest radiograph dataset with uncertainty labels and expert comparison. In: AAAI (2019) 4, 11

    2019

  7. [14]

    Quantitative imaging in medicine and surgery (2014) 4

    Jaeger, S., Candemir, S., Antani, S., Wáng, Y.X.J., Lu, P.X., Thoma, G.: Two public chest x-ray datasets for computer-aided screening of pulmonary diseases. Quantitative imaging in medicine and surgery (2014) 4

    2014

  8. [15]

    arXiv preprint arXiv:2004.12274 (2020) 3

    Jing, B., Wang, Z., Xing, E.: Show, describe and conclude: On exploiting the struc- ture information of chest x-ray reports. arXiv preprint arXiv:2004.12274 (2020) 3

    2020 arXiv

  9. [16]

    Scientific data (2019) 4

    Johnson, A.E., Pollard, T.J., Berkowitz, S.J., et al.: Mimic-cxr, a de-identified publicly available database of chest radiographs with free-text reports. Scientific data (2019) 4

    2019

  10. [17]

    PhysioNet (2020) 2, 4, 5, 7 16 Pham et al

    Karargyris, A., Kashyap, S., Lourentzou, I., Wu, J., Tong, M., Sharma, A., Abedin, S., Beymer, D., Mukherjee, V., Krupinski, E., et al.: Eye gaze data for chest x-rays. PhysioNet (2020) 2, 4, 5, 7 16 Pham et al

    2020

  11. [18]

    In: ISBI (2020) 9

    Kashyap, S., Karargyris, A., Wu, J., Gur, Y., Sharma, A., Wong, K.C., Moradi, M., Syeda-Mahmood, T.: Looking in the right place for anomalies: Explainable ai through automatic location learning. In: ISBI (2020) 9

    2020

  12. [19]

    ACM computing surveys (CSUR)54(10s), 1–41 (2022) 14

    Khan, S., Naseer, M., Hayat, M., Zamir, S.W., Khan, F.S., Shah, M.: Transformers in vision: A survey. ACM computing surveys (CSUR)54(10s), 1–41 (2022) 14

    2022

  13. [20]

    In: ICML (2018) 3

    Kim, B., Wattenberg, M., et al.: Interpretability beyond feature attribution: Quan- titative testing with concept activation vectors (tcav). In: ICML (2018) 3

    2018

  14. [21]

    arXiv preprint arXiv:2407.14709 (2024) 2, 9

    Le, M.Q., Graikos, A., Yellapragada, S., Gupta, R., Saltz, J., Samaras, D.:∞- brush: Controllable large image synthesis with diffusion models in infinite dimen- sions. arXiv preprint arXiv:2407.14709 (2024) 2, 9

    2024 arXiv

  15. [22]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Le, N., Pham, T., Do, T., Tjiputra, E., Tran, Q.D., Nguyen, A.: Music-driven group choreography. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 8673–8682 (2023) 2

    2023

  16. [23]

    Medical image analysis (2020) 7

    Lei, B., Huang, S., et al.: Self-co-attention neural network for anatomy segmenta- tion in whole breast ultrasound. Medical image analysis (2020) 7

    2020

  17. [24]

    Advances in neural information processing systems (2018) 3

    Li, Y., Liang, X., Hu, Z., Xing, E.P.: Hybrid retrieval-generation reinforced agent for medical image report generation. Advances in neural information processing systems (2018) 3

    2018

  18. [25]

    In: CVPR (2021) 4, 7

    Liu, F., Wu, X., Ge, S., Fan, W., Zou, Y.: Exploring and distilling posterior and prior knowledge for radiology report generation. In: CVPR (2021) 4, 7

    2021

  19. [26]

    In: ICLR (2019) 10

    Loshchilov, I., Hutter, F.: Decoupled weight decay regularization. In: ICLR (2019) 10

    2019

  20. [27]

    Artificial intelligence 267, 1–38 (2019) 2

    Miller, T.: Explanation in artificial intelligence: Insights from the social sciences. Artificial intelligence 267, 1–38 (2019) 2

    2019

  21. [28]

    In: CVPR (2023) 3

    Nauta, M., Schlötterer, J., van Keulen, M., Seifert, C.: Pip-net: Patch-based intu- itive prototypes for interpretable image classification. In: CVPR (2023) 3

    2023

  22. [29]

    In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision

    Nguyen, T.P., Pham, T.T., Nguyen, T., Le, H., Nguyen, D., Lam, H., Nguyen, P., Fowler, J., Tran, M.T., Le, N.: Embryosformer: Deformable transformer and collaborative encoding-decoding for embryos stage development classification. In: Proceedings of the IEEE/CVF Winter Confere...

    2023

  23. [30]

    In: Proceed- ings of the IEEE/CVF Winter Conference on Applications of Computer Vision

    Nguyen, V.D., Khaldi, K., Nguyen, D., Mantini, P., Shah, S.: Contrastive viewpoint-aware shape learning for long-term person re-identification. In: Proceed- ings of the IEEE/CVF Winter Conference on Applications of Computer Vision. pp. 1041–1049 (2024) 2

    2024

  24. [31]

    In: 2024 IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)

    Nguyen, V.D., Mantini, P., Shah, S.K.: Occluded cloth-changing person re- identification via occlusion-aware appearance and shape reasoning. In: 2024 IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS). pp. 1–8. IEEE (2024) 2

    2024

  25. [32]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Nguyen, V.D., Mirza, S., Zakeri, A., Gupta, A., Khaldi, K., Aloui, R., Mantini, P., Shah, S.K., Merchant, F.: Tackling domain shifts in person re-identification: A survey and analysis. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 4...

    2024

  26. [33]

    Artificial Intelligence in Medicine (2023) 10, 11

    Nicolson, A., Dowling, J., Koopman, B.: Improving chest X-ray report generation by leveraging warm starting. Artificial Intelligence in Medicine (2023) 10, 11

    2023

  27. [35]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Pham, T.T., Do, T., Le, N., Le, N., Nguyen, H., Tjiputra, E., Tran, Q., Nguyen, A.: Style transfer for 2d talking head generation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 7500–7509 (2024) 2 FG-CXR 17

    2024

  28. [36]

    OpenAI blog (2019) 8, 9, 10

    Radford, A., Wu, J., et al.: Language models are unsupervised multitask learners. OpenAI blog (2019) 8, 9, 10

    2019

  29. [37]

    Nature machine intelligence (2019) 2, 13

    Rudin, C.: Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nature machine intelligence (2019) 2, 13

    2019

  30. [38]

    Statistics Sur- veys (2022) 3

    Rudin, C., Chen, C., Chen, Z., Huang, H., Semenova, L., Zhong, C.: Interpretable machine learning: Fundamental principles and 10 grand challenges. Statistics Sur- veys (2022) 3

    2022

  31. [39]

    arXiv preprint arXiv:1910.01108 (2019) 10

    Sanh, V., Debut, L., Chaumond, J., Wolf, T.: Distilbert, a distilled version of bert: smaller, faster, cheaper and lighter. arXiv preprint arXiv:1910.01108 (2019) 10

    2019 arXiv

  32. [40]

    In: CVPR (2017) 3

    Selvaraju, R.R., Cogswell, M., et al.: Grad-cam: Visual explanations from deep networks via gradient-based localization. In: CVPR (2017) 3

    2017

  33. [41]

    In: ICCV (2017) 11

    Shetty, R., Rohrbach, M., Anne Hendricks, L., Fritz, M., Schiele, B.: Speaking the same language: Matching machine to human captions by adversarial training. In: ICCV (2017) 11

    2017

  34. [42]

    Radiology: Artificial Intelligence (2019) 3, 4

    Shih, G., Wu, C.C., et al.: Augmenting the national institutes of health chest radio- graph dataset with expert annotations of possible pneumonia. Radiology: Artificial Intelligence (2019) 3, 4

    2019

  35. [43]

    In: CVPR (2023) 2, 3, 4, 11

    Tanida,T.,Müller,P.,Kaissis,G.,Rueckert,D.:Interactiveandexplainableregion- guided radiology report generation. In: CVPR (2023) 2, 3, 4, 11

    2023

  36. [44]

    In: CVPR (2023) 10, 11

    Tanida,T.,Müller,P.,Kaissis,G.,Rueckert,D.:Interactiveandexplainableregion- guided radiology report generation. In: CVPR (2023) 10, 11

    2023

  37. [45]

    Controlled clinical trials (2000) 4

    Team, P.P., Gohagan, J.K., Prorok, P.C., Hayes, R.B., Kramer, B.S.: The prostate, lung, colorectal and ovarian (plco) cancer screening trial of the national cancer institute: history, organization, and status. Controlled clinical trials (2000) 4

    2000

  38. [46]

    In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition Workshops

    Tran, M.T., Nguyen, T.V., Hoang, T.H., Le, T.N., Nguyen, K.T., Dinh, D.T., Nguyen, T.A., Nguyen, H.D., Hoang, X.N., Nguyen, T.T., et al.: itask-intelligent traffic analysis software kit. In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition Wo...

    2020

  39. [47]

    Scientific Reports (2023) 7

    Ullah, I., Ali, F., Shah, B., El-Sappagh, S., Abuhmed, T., Park, S.H.: A deep learn- ing based dual encoder–decoder framework for anatomical structure segmentation in chest x-ray images. Scientific Reports (2023) 7

    2023

  40. [48]

    Vo, K., Pham, T.T., Yamazaki, K., Tran, M., Le, N.: Dna: Deformable neural articulations network for template-free dynamic 3d human reconstruction from monocularrgb-dvideo.In:ProceedingsoftheIEEE/CVFConferenceonComputer Vision and Pattern Recognition. pp. 3676–3685 (2023) 2

    2023

  41. [49]

    In: CVPR (2017) 4

    Wang, X., Peng, Y., Lu, L., Lu, Z., Bagheri, M., Summers, R.M.: Chestx-ray8: Hospital-scale chest x-ray database and benchmarks on weakly-supervised classifi- cation and localization of common thorax diseases. In: CVPR (2017) 4

    2017

  42. [50]

    In: ICCV (2021) 9, 10

    Wu, H., Xiao, B., Codella, N., Liu, M., Dai, X., Yuan, L., Zhang, L.: Cvt: Intro- ducing convolutions to vision transformers. In: ICCV (2021) 9, 10

    2021

  43. [51]

    Wu, J.T., Agu, N.N., Lourentzou, I., Sharma, A., Paguio, J.A., Yao, J.S., Dee, E.C., Mitchell, W., Kashyap, S., Giovannini, A., et al.: Chest imagenome dataset (version 1.0. 0). PhysioNet (2021) 2, 3, 4

    2021

  44. [52]

    In: ECCV (2018) 11

    Xiong, Y., Dai, B., Lin, D.: Move forward and tell: A progressive generator of video descriptions. In: ECCV (2018) 11

    2018

  45. [53]

    In: MICCAI (2021) 3 18 Pham et al

    You, D., Liu, F., Ge, S., Xie, X., Zhang, J., Wu, X.: Aligntransformer: Hierarchical alignment of visual regions and disease tags for medical report generation. In: MICCAI (2021) 3 18 Pham et al

    2021

  46. [54]

    arXiv preprint arXiv:2303.00915 (2023) 10

    Zhang, S., Xu, Y., et al.: Biomedclip: a multimodal biomedical foundation model pretrained from fifteen million scientific image-text pairs. arXiv preprint arXiv:2303.00915 (2023) 10

    2023 arXiv

  47. [55]

    In: AAAI (2020) 7

    Zhang, Y., Wang, X., Xu, Z., Yu, Q., Yuille, A., Xu, D.: When radiology report generation meets knowledge graph. In: AAAI (2020) 7

    2020

Pith tools

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