The DOI matches its Crossref record; 50 of the 51 references it cites were found in a registry or an index.
METransformer: Radiology Report Generation by Transformer with Multiple Learnable Expert Tokens
Open accessAbstract
In clinical scenarios, multi-specialist consultation could significantly benefit the diagnosis, especially for intricate cases. This inspires us to explore a “multi-expert joint diagnosis” mechanism to upgrade the existing “single expert” framework commonly seen in the current literature. To this end, we propose METransformer, a method to realize this idea with a transformer-based backbone. The key design of our method is the introduction of multiple learnable “expert” tokens into both the transformer encoder and decoder. In the encoder, each expert token interacts with both vision tokens and other expert tokens to learn to attend different image regions for image representation. These expert tokens are encouraged to capture complementary information by an orthogonal loss that minimizes their over-lap. In the decoder, each attended expert token guides the cross-attention between input words and visual tokens, thus influencing the generated report. A metrics-based expert voting strategy is further developed to generate the final report. By the multi-experts concept, our model enjoys the merits of an ensemble-based approach but through a manner that is computationally more efficient and supports more sophisticated interactions among experts. Experimental results demonstrate the promising performance of our proposed model on two widely used benchmarks. Last but not least, the framework-level innovation makes our work ready to incorporate advances on existing “single-expert” models to further improve its performance.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“Such references can be removed to reduce hallucinations when prior studies are not available, 9,14,18 or used in conjunction with prior images to enable descriptions of change.17,19,20”
From MAIRA-2: Grounded Radiology Report Generation (2024) · cites this paper as [14]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
Jakubův, J., Chvalovský, K., Goertzel, Z., et al. (2023) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.itp.2023.19 · Cited by 76,648
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325 · cited 14×
“Specifically, we compare METransformer with 5 state-of-the-art (SOTA) image captioning methods, including Show-tell [32], AdaAtt [24], Att2in [1], Transformer [7] and M2transformer [6].”
MethodInfluentialAn Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 47,472 · cited 3×
“1, our encoder comprises a vision transformer (ViT) [9] encoder and an expert bilinear transformer encoder.”
MethodInfluentialSqueeze-and-Excitation Networks
Hu, J., Shen, L., & Sun, G. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00745 · Cited by 35,667 · cited 1×
“A squeeze-excitation operation [10] to B mid is performed to obtain channel-wise attention Bilinear Encoder Layer.”
MethodPyTorch: An Imperative Style, High-Performance Deep Learning Library
Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al. (2019) · arXiv · doi:10.48550/arxiv.1912.01703 · Cited by 31,473 · cited 1×
“We implement our model using Pytorch [27] and Pytorch-lightning library 2 with two NVIDIA GeForce RTX 3090 GPU cards.”
MethodLayer NormalizationPreprint
Ba, J., Kiros, J., & Hinton, G. E. (2016) · arXiv.org · doi:10.48550/arxiv.1607.06450 · Cited by 13,762 · cited 2×
“Layer-norm(LN) [2] is applied before every block and residual connections are applied after every block.”
BackgroundMethod(2004) · Cited by 9,718
Transformers: State-of-the-Art Natural Language Processing
Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtowicz, M., et al. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations · doi:10.18653/v1/2020.emnlp-demos.6 · Cited by 8,741
Research works printed but not linked: 1
- su, 2019, Vl-bert: Pre-training of generic visuallinguistic representations, ArXiv Preprint
Cited by
Sorted by citations
Items 101–183 of 183
Integrating Clinical Knowledge into Radiology Report Generation to Enhance Diagnosis
Cao, Z., Yue, X., Chen, Y., Xu, Z., Wei, Z., & Li, Z. (2025) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm66473.2025.11355985 · Cited by 1
“To verify the superiority of our CKRG, we compare with a wide range of representative methods for addressing the MRG task , including R2Gen [9], M2TR [10], MKSG [11], CliBert [12], M2KT [13], METrans [14], KiUT [15], DCL [16], RGRG [17], FMVP [18], and PromptMRG [19].”
Beyond Averaging: All-Rounder Capability Evaluation for Radiology Report Generation
Ji, D. Y., Tang, Z., Zhang, D., Li, S., & Zhou, G. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11461738 · Cited by 1
PeFoMed: Parameter efficient fine-tuning of multimodal large language models for medical CXR
Liu, G., Tang, X., He, J., Li, P., Chen, Z., & Zhong, S. (2026) · Scientific Reports · doi:10.1038/s41598-026-47871-2 · Cited by 81
TiBiX: Leveraging Temporal Information for Bidirectional X-Ray and Report Generation
Sanjeev, S., Maani, F. A., Abzhanov, A., Papineni, V. R. K., Almakky, I., Papież, B. W., & Yaqub, M. K. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72744-3_17 · Cited by 3
A Novel Transformer-Based Framework for Chest X-Ray Captioning with Clinical Sign Detection
Tran, T. T., Nguyen, D. V. M., & Huynh, H. T. (2025) · SN Computer Science · doi:10.1007/s42979-025-04032-y · Cited by 2
Izhar, A., Idris, N. B., & Japar, N. B. (2025) · Image and Vision Computing · doi:10.1016/j.imavis.2025.105603 · Cited by 1
Enhanced group relation learning via aligned attention masking for fashion product captioning
Tang, Y., Ye, D., Tao, F., & Du, G. (2025) · Journal of King Saud University - Computer and Information Sciences · doi:10.1007/s44443-025-00195-z · Cited by 3
“However, its performance is inferior to METransformer.”
InfluentialRamadan, W., & Akay, B. (2025) · International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA) · doi:10.1109/acdsa65407.2025.11166431 · Cited by 1
“Also, Wang et al. [39] followed the transformer-transformer approach and proposed METransformer, a transformer-based model with multiple learnable expert tokens to simulate multi-specialist diagnosis.”
Extractive Radiology Reporting With Memory-Based Cross-Modal Representations
Tian, Y., Yan, Z., Lyu, N., & Song, Y. (2025) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2025.3636868 · Cited by 1
“There are also studies that incorporate cross-modal features to further improve the accuracy and details of reports [30], [36], [37].”
Feng, W., Lu, Y., Li, X., Shi, S., & Qi, Y. (2025) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc58881.2025.11343574 · Cited by 1
DMR$$^2$$G: diffusion model for radiology report generation
Ouyang, H., Chang, Z., Tang, B., & Li, S. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-20206-x · Cited by 1
“PPKED [54] and MET [57] are proposed to integrate medical knowledge with typical MRG backbones.”
Multi-granularity Semantic Guided Transformer for Radiology Report Generation
Song, Y., Hua, X., Zhang, K., Zan, H., & Li, R. (2024) · Lecture notes in computer science · doi:10.1007/978-981-97-9437-9_36
Adaptive Enhanced Feature Fusion to Generate Medical ReportsPreprint
Cao, Y., Ding, H., Li, B., Sun, X., She, Q., & Rahman, M. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5160347
Radiology Report Generation Based on Multi-Scale Feature Fusion and Enhancement
Cao, Y., Ding, H., Li, B., Sun, X., She, Q., & Rahman, M. (2025) · Communications in computer and information science · doi:10.1007/978-981-96-2914-5_3
Chen, K., Zhu, X., Zhang, W., Jiang, S., & Chen, Q. (2025) · Communications in computer and information science · doi:10.1007/978-981-96-3755-3_10
Medical Report Generation Based on Mixed Visual Encoding and Vision-Language Model
Xu, H., Yang, C., Chen, H., Xie, J., & Cao, Y. (2025) · doi:10.1145/3730436.3730508
“As shown in Table 1, we have compared our proposed method with existing mainstream methods, including Show-Tell [14], Att2in [15], AdaAtt [16], Transformer [17], M2Transformer [18], MSAT [19], METransformer [20], and R2GenGPT [21].”
Zhu, J., & Lü, P. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-98688-8_14
Liu, J., Deng, Y., Liu, J., Song, D., Duan, X., & Li, J. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.108553
Tang, Y., & Tao, F. (2025) · Cluster Computing · doi:10.1007/s10586-025-05368-z
“Existing methods [15, 16] primarily adopt standard image captioning frameworks and utilize an encoder–decoder architecture.”
Automated Chest X-ray Captioning and Report Generation Using CheXNet and BioGPT
Mishra, D., Sai, K. P., Harshavardhan, K., Ram, C., & Reddy, K. G. (2025) · IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS) · doi:10.1109/avss65446.2025.11149939
“Reinforcement learning [16], hybrid architectures [8], and domain-tuned models like BioGPT [5] have shown promise in enhancing interpretability and factual accuracy.”
Tang, Q., Liu, L., Yang, X., Liu, L., & Peng, W. (2025) · Journal of Imaging Informatics in Medicine · doi:10.1007/s10278-025-01650-z
“More recently, the Transformer architecture [23] has emerged as the dominant paradigm, providing an efficient mechanism for capturing the complex relationship between image and language features [24–26].”
Self-citationRankRRG: A Rank-Aware Framework for Automated Radiology Report Generation
Qiu, M., Li, Y., Fan, X., Liu, Y., Cao, J., Zhang, B., Wang, R., Ma, W., & Lei, W. (2025) · Lecture notes in computer science · doi:10.1007/978-981-95-3456-2_4
Self-Supervised Anatomical Consistency Learning for Vision-Grounded Medical Report Generation
Yang, L., Ni, Z., Wen, Y., Liu, Y., He, L., & Shen, H. T. (2025) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3746027.3754913
“Recent advances in radiology report generation have explored diverse strategies for integrating medical knowledge [11, 36, 43, 45], employing retrieval-augmented meth-ods [9, 19], and leveraging the transfer capabilities of large language models [16, 26, 40, 44].”
Bo, X., Yang, F., Xu, F., & Zhang, X. (2025) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3746027.3755368
Optimizing Efficiency and Visual-Textual Alignment for LLM-Based Radiology Report Generation
Chen, Z., Gao, P., Lee, Y., Barthélemy, J., Zhou, L., & Wang, L. (2025) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme59968.2025.11210172
“Recent advancements have focused on transformer-based approaches, which can better capture complex dependencies in both the visual and textual modalities [1], [2], [8].”
Self-citationRSTA: A Recurrent Scene Topic-Aware Model for Multi-Scene Endoscopic Report Generation
Yuan, X., Kuang, J., Hua, L., Zhao, G., Zhang, C., & Li, J. (2025) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn64981.2025.11227409
“Transformer-based methods are R2Gen [28], R2GenCMN [29], MeTrans-former [30], MMTN [31], CAMANet [36], PromptMRG [33].”
Label Auxiliary and Cross-modal Auxiliary Memory Network for Medical Report Generation
Yang, R., & Li, F. (2025) · International Symposium on Computer and Information Processing Technology (ISCIPT) · doi:10.1109/iscipt67144.2025.11265610
“Other models enhance visual semantic alignment by using disease labels [26], memory-driven Transformers [27], multi-expert modeling through learnable vocabular [28], and the fusion methods of clinical knowledge and internal visual features [29-30].”
Group-wise Compression and Summarization via LLM-based Ensemble for Chest X-ray Report Generation
Park, S., Heo, K.-S., Kang, B., Lim, M., Choi, W., & Kam, T. (2025) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc58623.2025.11251844
“In our experiment, we follow the dataset partitioning ratio used in [19], [20] to ensure a fair evaluation comparison.”
Hentati, R., Hentati, M., & Abid, A. (2025) · Biomedical Physics & Engineering Express · doi:10.1088/2057-1976/ae2e01
Mwcl: Memory-driven and mapping alignment with weighted contrastive learning for radiology reports
Usman, M., Ma, Z., & Yijia, Z. (2025) · Multimedia Systems · doi:10.1007/s00530-025-02040-9
“● METransformer [ 34 ]: This approach replaces single-ex - pert frameworks with multiple interactive expert tokens that capture complementary image features via orthogo - nal learning and vote on the final report, efficiently mim - icking clinical multi-specialist consensus efficiently.”
Joint attention GAN for medical report generation with clinical style preservation
Tang, Z., Liu, X., Bie, J., Xu, L., & Xu, B. (2025) · Scientific Reports · doi:10.1038/s41598-025-28181-5
“Combining the skip connections and convolution block attention modules, METransformer 21 exacts the visual features and employs memory driven Transformer 22 to output the medical reports, and TSGET 23 utilizes the two-stage global enhancement layers to facilitate Transformer to generate more…”
X-Gen: Enhancing Radiology Report Generation via LLM-Driven Data Augmentation and Decoupled Training
Wang, C., Chen, Q., To, M., Kutaiba, N., Yoo, J.-G., Xie, Y., & Wu, Q. (2025) · International Conference on Digital Image Computing: Techniques and Applications (DICTA) · doi:10.1109/dicta68720.2025.11302459
“By the time of the experiment, KiUT[11] and METransformer[32] have not open-sourced their code.”
Towards Generalizable Clinical Knowledge Discovery for Radiology Report Generation
Chen, X., Chen, Q., Phan, V. M. H., Wu, Q., Chen, J., & Xie, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-09513-8_37 · Cited by 1
Cross-View and Multi-Step Interaction for Change Captioning
Xian, T., Zhou, Z., Zhou, W., Zeng, D., & Li, B. (2026) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2026.3651125
“CC is widely used in surface evolution analysis [1], [2], [3], medical report generation [4], [5], [6], and video surveillance systems [7], [8].”
ALV-Net: Adaptive Language and Visual Feature Fusion for Radiology Report Generation
Guo, L., Xu, P., & Liu, R. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-4445-5_15
Ding, S., Fan, M., & Li, M. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-5631-1_10
Multi-scale based cross-modal semantic alignment network for radiology report generation
Zhang, Z., Zhao, L., Lan, D., Wang, Y., Jiang, L., & Dong, X. (2025) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc58881.2025.11343384
“…[3] 0.461 0.294 0.208 0.155 0.187 0.358 JPG [4] 0.479 0.319 0.222 0.174 0.193 0.377 XPRONET* [5] 0.496 0.320 0.224 0.167 0.206 0.376 METransformer [2] [8] 0.325 0.203 0.136 0.096 0.134 0.276 ADAATT [9] 0.299 0.185 0.124 0.088 0.118 0.266 TopDown [7] 0.317 0.195 0.130 0.092 0.128 0.267 CMCL [17]…”
InfluentialTang, M., Ding, W., Wei, C. A. I., & Jiang, Y. (2026) · Journal of King Saud University - Computer and Information Sciences · doi:10.1007/s44443-026-00475-2
Hou, X., Li, Y., & Wang, S. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00133
“Automating this process has emerged as a critical research focus, with significant advancements driven by artificial intelligence and image captioning techniques [2, 11, 17, 29, 33].”
InfluentialFog/Edge-Aware State Space Models for Multi-Task Chest X-Ray Report Generation and Lesion Detection
Feng, W., Lü, Y., Li, X., Yung, K. L., & Ip, W. H. (2026) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2026.3667969
“Recent models, such as Clinical-BERT [12] and METransformer [13], incorporated domain-specific pretraining for improved semantic alignment, and DCL [1] and R2GenGPT [4] utilized contrastive learning and large language models (LLMs).”
Differential attention for retinal image captioning via cross-modal integration
Hieu, N. H., Thanh, L. V., Lan, L. T. H., & Pham, H. V. (2026) · Signal Image and Video Processing · doi:10.1007/s11760-026-05208-6
Case2Knowledge: Structured Knowledge-Infused Prompting for Faithful Radiology Report Generation
Gao, Q., Liu, T., Sun, Z., Wang, B., Yin, B., & Liu, Z. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11464942
Zhao, H., Zhang, Y., Ma, L., Xu, S., & Sun, D. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11460414
He, P., Wang, J., & Yuan, X. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11461023
Advancing In-Context Learning for Efficient and Stable Medical Report Generation
Li, M., Liu, R., Shi, Z., Han, M., Yao, L., Li, Z., Chang, X., Pohl, K. M., Islam, M. T., & Xing, L. (2026) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2026.3689780
“We report the performances (with respect to the previously defined metrics) of our R2VLM and several state-of-the-art MRG methods for comparison, including R2Gen [40], R2GenCMN [63], PPKED [39], DCL [42], MET [64], R2GPT [10], PromptMRG [13], BiomedGPT [65], [66] and CXP [11].”
Laila, A., Islam, M. D., Haider, M. S., Habib, S. M. M., & Prottoy, A. A. (2025) · International Conference on Computer and Information Technology (ICCIT) · doi:10.1109/iccit68739.2025.11491591
Contrastive Multimodal Learning for Chest Radiograph Diagnosis and Report Generation
Gu, Y., Deng, Y., Cai, C., Wang, T., Zhu, W., & Tao, Y. (2026) · International Conference on Computer, Artificial Intelligence and Control Engineering (CAICE) · doi:10.1145/3804601.3804706
“As shown in Table 1, we compare our proposed model Chestcon with several state-of-the-art baselines, including SA [ 12 ] , MET [ 13 ] , and SEI [ 14 ] , on both natural language generation (NLG) metrics and clinical efficacy (CE) metrics.”
Radgaze-LLM: Anatomical Region-Grounded Radiology Report Generation via Learning From Expert Gaze
Sultana, J., Qin, R., & Yin, Z. (2026) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi61048.2026.11515459
“Thus, unlike previous methods [7, 8] that rely on only image-level features, 1. we detect anatomical regions and learn region-grounded visual representations; 2. to introduce region-aware supervision during training, our approach learns from expert knowledge from eye gaze.”
Expert Model Guided Visual Programming for Radiology Report Generation
Liu, Z., & Xu, M. (2026) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi61048.2026.11515840
Gaze2Report: Radiology Report Generation via Visual-Gaze Prompt Tuning of LLMs
Konwer, A., Bhattacharya, M., & Prasanna, P. (2026) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi61048.2026.11515599
“In Table 1, we compare the proposed Gaze2Report framework with standard image captioning methods (ST [14], Att2in [15], AdaAtt [16]) and medical report generation models (XPRONET [18], EGGCA-Net [10], METransformer [19], R2Gen [17], and R2GenGPT [20]).”
REALM: Reinforcement Learning-Based Cross-Modal Alignment for Fine-Grained Medical Report Generation
Teng, P., Liu, W., Wang, Y., Yang, H., & Xing, M. (2026) · Communications in computer and information science · doi:10.1007/978-981-95-7087-4_12
LACFNet: Locality Aware Cross Modal Fusion Network for Automatic Medical Image Report Generation
Li, Y., Peng, K., Liu, T., & Xiao, H. (2025) · Asian Conference on Artificial Intelligence Technology (ACAIT) · doi:10.1109/acait67930.2025.11522656
“Wang et al. [1] introduces trainable ’expert’ tokens into the encoding-decoding process, where each expert token interacts with visual tokens and other expert tokens to learn to focus on different image regions for image representation.”
InfluentialSemi-supervised medical image captioning via anatomical collaborative evidence network
Zhou, S., Liu, Q., Cai, L., et al. (2026) · Frontiers in Medicine · doi:10.3389/fmed.2026.1816295
“As shown in Table 3, we compared our model with several state-of-the-art (SOTA) methods, including R2Gen, AlignTransformer, EIL, METransformer, and KIUT.”
Image mask-guided cross-modal network for radiology report generation
Liu, Y., Hou, X., Li, X., & Zhang, Y. (2026) · Expert Systems with Applications · doi:10.1016/j.eswa.2026.133027
Liu, X., Jin, Q., Zhang, T.-T., Tang, Z., Chen, Y., & Liu, Z. H. (2026) · IEEE Internet of Things Journal · doi:10.1109/jiot.2026.3699691
“We conduct experiments on two major medical imaging datasets, MIMIC-CXR and IU-Xray, comparing the CSDD with methods, including R2Gen [41], CMN [42], Contrastive Attention [43], M2KT [44], CMM + RL [45], METrans [46], DCL [33], MA [47], EKAGen [48], Bootstrap-pingLLM [37], and CMCRL [49].”
MOREGen: multi-view organ-aware model for medical report generation
Chen, L., Hu, K., Yu, E., & Cao, Q. (2026) · Multimedia Systems · doi:10.1007/s00530-026-02359-x
“METransformer [28] strengthens report generation capabilities by designing a “multi-expert joint diagnosis” mechanism.”
SKIM: Semantic Knowledge Infused Modeling for Medical Report Generation
Lee, H., Jo, Y., Hong, I., & Park, S. (2026) · ACM Symposium on Applied Computing · doi:10.1145/3748522.3779771
MMG-RAG: Multimodal Graph RAG for Medical Report Generation
Liu, G., Tang, X., Gan, J., Ren, T., Liu, T., & Zhong, S. (2026) · International Conference on Multimedia Retrieval · doi:10.1145/3805622.3810644
Chen, L., Jin, R., Luo, J., Chen, H., Strotzer, Q., Yan, R., Xue, Y., Prevedello, L. M. S., & Wu, D. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38059-3_41
TAVR-VLM: Risk-Conditioned Causal Grounding for Hallucination-Resistant Report Generation
Lu, Z., Liu, X., Song, S., Ji, C., Nguyen, A., Su, J., Razzak, I., & Wang, J. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38059-3_57
Zhang, X., Zhao, J., Ruan, W., & Li, X. (2026) · International Conference on Computer Supported Cooperative Work in Design (CSCWD) · doi:10.1109/cscwd68734.2026.11582171
“…compared with representative approaches including cross-modal alignment models R2Gen [4], R2GenCMN [15], and ICON [26]; architecture-enhanced meth-ods METransformer [17], PromptMRG [18], and AHP [27]; and recent large language model–based methods R2GenGPT [25], Med-LMM [21], and R2-LLM [22].”
Yang, Q., Li, W., Guo, X., Harwood, M. C., Woo, P. Y. M., Zhang, J., Chen, Y., Zhang, K., & Yuan, Y. (2026) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2026.3710717 · Cited by 1
“Although these methods [1], [2], [4], [5], [7], [10], [11], [12], [16], [21], [22], [26] demonstrate rapid success on medical report generation, they usually employ 3D encoder to process the whole volumes and directly utilize the LLM as the language decoder to generate reports.”
InfluentialLi, Y., Liu, Y., Chen, Z., Chen, T., Chen, Z., Liu, L., Wang, L., & Zhou, L. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37422-6_12 · Cited by 1
Self-citationREVA-PO: Stabilizing Reinforcement Learning for Chest X-Ray Report Generation
Guo, L., Tahir, A. M., & Wang, Z. J. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37321-2_31
Self-citationXie, S., Shen, K., Zhuang, Q., Huang, G., & Chen, X. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3510-0_34
Boltzmann-driven dynamic annealing network for knowledge-guided radiology report generation
Xiao, W., Liu, M., Zhu, H., & Zhang, J. (2026) · Multimedia Systems · doi:10.1007/s00530-026-02524-2
“We compare with several current existing state-of-the-art RRG methods with different architectures to assess ours per - formance, including R2Gen [47], CMN [48], PPKED [49], CMM + RL [50], METransformer [51], M2KT [26], CAM-ANet [52], Token-Mixer [53], AHP [54], MAN [55].”
OtosRG: A Prompt-Augmented Framework via Diagnostic Guidance for Otoscopic Report Generation
He, P., Wang, J., Zhang, X., Wu, Z., Li, X., & Yuan, X. (2026) · Communications in computer and information science · doi:10.1007/978-981-92-3548-3_36
CLEAR: an auditable foundation model for radiology grounded in clinical concepts
Han, T., Wu, R., Tian, Y., Khader, F., Adams, L. C., Bressem, K. K., Davatzikos, C. A., Kather, J. N., Shen, L., Mankoff, D. A., et al. (2026) · Nature Biomedical Engineering · doi:10.1038/s41551-026-01741-4 · Cited by 3
Diagnostic Report Generation via a ProActive Multimodal Agentic System
Li, X., Hou, X., Allen, G., Guirguis, S., Zhou, Y., Huang, Z., & Gan, Y. (2026) · ACM Transactions on Computing for Healthcare · doi:10.1145/3833391
“We compare the diagnostic report generation performance of PAMA with the following state-of-the-arts (SOTAs): R2GCMN [8], METransformer [59], ChatCAD+ [73], SILC [35], R2GenGPT [72], ATL-CA [41], MPO [62].”
Shivakumar, P. G. A., Mahmood, A., & Khatoon, S. (2026) · Journal of Imaging · doi:10.3390/jimaging12080338
“Recent research [6,15,16] has begun to address these challenges through multimodal Transformer-based strategies tailored to breast ultrasound reporting.”
TRACE: Temporal Radiology with Anatomical Change Explanation for Grounded X-Ray Report Generation
Aranya, O. F. M. R. R., & Desai, K. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31933-3_7 · Cited by 3
Image captioning using a transformer with topic–word semantic modeling and multimodal feature fusion
Yahya, A. A., Al-Qatf, M., Hawbani, A., et al. (2026) · Journal of King Saud University - Computer and Information Sciences · doi:10.1007/s44443-026-01085-8
Multi-guided and pixel enhancement model for radiology report generation
Guo, Y., Hou, X., Liu, Z., Zhou, Y., & Zhang, Y. (2026) · Multimedia Systems · doi:10.1007/s00530-026-02628-9
Rahman, H., Mahmood, T., & Raza, A. (2026) · Journal of Artificial Intelligence for Healthcare Systems · doi:10.68159/u145014357
Interpretable Radiology Report Generation via Prototype-Guided Contrastive Learning
张, 颜. (2026) · Hans Journal of Biomedicine · doi:10.12677/hjbm.2026.165093
Wavelet-Enhanced Coarse-to-Fine Radiology Report Generation with Large Language Models
Yang, S., Deng, Y., Qin, Q., Zeng, P., Zhang, H., Hu, W., & Zhang, W. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38062-3_62
TraceCXR: Verifiable Chest X-Ray Reports via Evidence-Addressable Graph Prompting
Dong, H., Sun, C., Yan, H., Hu, W., & Du, B. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38062-3_57
Disease-Aware Visual Token Memory for Chest X-Ray Report Generation with VLMs
Fichadia, R. H. (2026) · IEEE World Conference on Applied Intelligence and Computing (AIC) · doi:10.1109/aic70896.2026.11716964
“METrans-former [29] addresses the single-token bottleneck by proposing multiple learnable “expert” tokens, in which every token focuses on a unique part of the image.”
LlamaRG: A Multi-View Large Language Model for Radiology Report Generation
Jayas, T., Rastogi, A., Raghavan, P., Brugnara, G., Schlamp, K., Foltyn‐Dumitru, M., & Vollmuth, P. (2026) · doi:10.1109/cvprf71909.2026.00552
“…0.181 AdaAtt [24] 0.322 0.239 0.249 Transformer [40] 0.331 0.224 0.228 R2Gen [3] 0.333 0.273 0.276 R2GenCMN [4] 0.334 0.275 0.278 METransformer [46] 0.364 0.309 0.311 R2GenGPT(Deep) [45] 0.392 0.387 0.389 B-LLM (FT) [16] 0.335 0.246 0.284 MedLLM [21] 0.412 0.373 0.395 MLRG* [20] 0.327 0.255…”
InfluentialA Denoising-Enhanced Multimodal Learning Framework for Robust Nasal Endoscopy Report Generation
Yuan, X., Huang, M., Hua, L., Ju, J., Zhao, X., & Wu, L. Y. (2026) · doi:10.1109/cvprf71909.2026.00533
“Following previous works [29, 38], we adopt the standard data split of 70% : 10% : 20% for training, validation, and testing, respectively.”
InfluentialVision-Language Models for Automated 3D PET/CT Report Generation
Jiao, W., Yan, K., Zhang, J., Jin, D., & Xie, Z. (2026) · doi:10.1109/cvprf71909.2026.00509 · Cited by 1
“Some integrate memory modules, such as the disease prototype memory in Dia-LLaMA [7] for class imbalance or the relational memory in CT2Rep [15] for coherence.”
R3-Gen: Enhancing medical report generation with reasoning-reinforced fine-tuning
Xiao, T., Qiu, Y., Yang, H., & Wang, Z. (2026) · Pattern Recognition · doi:10.1016/j.patcog.2026.115111
Self-citationYang, F., Li, G., He, Z., Wu, R., Luo, G., Yue, Y., Ji, D., Wu, J., & Zhang, H. (2026) · Computerized Medical Imaging and Graphics · doi:10.1016/j.compmedimag.2026.102834
Metadata
- Authors
- Wang, Z., Liu, L., Wang, L., & Zhou, L.
- Affiliations
- The University of Sydney (AU)
- The University of Adelaide (AU)
- University of Wollongong (AU)
- Venue
- 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
- Publisher
- IEEE
- Published
- 2023-06
- Pages
- 11558-11567
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W4386076536
- Semantic Scholar
- 4d90382730eefbe8639e64f6855f206657758cc7
- Cited by
- 184
- References
- 51
- Citations by year
- 2026: 58 · 2025: 82 · 2024: 37 · 2023: 7
- FWCI
- 19.62
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by-nc-nd)
- Research Online (University of Wollongong) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
- Keywords
- radiology report generation, vision-language transformers, cross-attention
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
