The paper presents a two‑stage multimodal framework for chest X‑ray interpretation that incorporates radiologist gaze data from the MIMIC‑Eye dataset. Stage 1 introduces a gaze‑token classifier that fuses image patches, bounding‑box masks, transcription embeddings, and fixation maps, and a curriculum‑scheduled loss that improves accuracy and spatial alignment, yielding a 4.4% AUC gain and 13.3% F1 improvement. Stage 2 translates classifier predictions into region‑specific diagnostic sentences using confidence‑weighted keywords, an expert dictionary, and a prompted large language model, boosting clinical‑term BERTScore and ROUGE over keyword baselines.
By Tanjim Islam Riju, Shuchismita Anwar, Saman Sarker Joy, Farig Sadeque, Swakkhar Shatabda
The study evaluates the robustness of medical vision‑language models for tuberculosis screening on chest X‑rays by testing them across multiple datasets, prompts, and evaluation settings. Three specialized models (BioMedCLIP, CheXficient, MedSigLIP) and a general OpenCLIP model were audited on 12,200 images, producing 244,000 model–image–prompt scores. Results show that no model consistently outperforms others across all cohorts and reliability criteria, with prompt changes and control group composition significantly affecting AUROC, and that high training‑set performance does not reliably transfer to external cohorts.
By Mushir Akhtar, M. Tanveer, Mohd. Arshad
The study evaluated ten chest X‑ray AI classification models on 12 datasets from six countries, focusing on image‑level classification and measuring specificity at 95% sensitivity. MAVL and MedKLIP achieved the highest specificity (≈0.37) but still produced many false positives, with performance dropping on institutional cohorts compared to public benchmarks. No single model excelled across all findings, countries, or lesion types, highlighting limited generalisability.
By Qinmei Xu, Yiheng Li, Wenya Zhao, Zihao Zhao, Chuanjun Xu, Ahmet Gorkem Er, Xianghao Zhan, Brittany Dashevsky, Mohammed Alawad, Marc von der St\"uck, Marvin Gazibari\'c, Daniel Truhn, Shunsuke Shibata, Ryosuke Horino, Mio Ito, Akio Hiwatashi, Suyu Li, Mengya Yang, Ya Liu, Zhiyuan Sun, Changsheng Zhou, Haruka Itakura, Xiao Li, Olivier Gevaert
arXiv:2606. 17710v1 Announce Type: cross Abstract: Medical vision-language models report strong chest radiograph accuracy, and this is increasingly read as evidence that they use the image.
By Mahshad Lotfinia, Sebastian Ziegelmayer, Lisa Adams, Daniel Truhn, Andreas Maier, Soroosh Tayebi Arasteh
Med-AR introduces two autoregressive vision‑language models, Med‑AR‑8B and Med‑AR‑2B, pretrained on structured radiology reports, abnormality‑focused text, and region annotations to address long‑tailed chest X‑ray classification. The models outperform existing contrastive, self‑supervised, and supervised encoders—including Med‑CLIP, CheXFound, EVA‑Base, ARK, and BioViL‑T—across PadChest, MIMIC‑CXR, and CheXpert, achieving higher mean AUROC and AUPRC for head, medium, and tail findings and lower excess area under the risk‑coverage curve. Med‑AR also demonstrates improved selective‑prediction performance, with Med‑AR‑8B raising tail‑label mean AUPRC on MIMIC‑CXR from 0.1033 to 0.1441 and Med‑AR‑2B delivering the strongest discrimination on PadChest.
By Janhavi Prabhu, Sahil, Akshay V, Shivam Shukla, Manoj Tadepalli, Preetham Putha
The study introduces a gaze-informed transformer framework that classifies radiologist expertise during thoracic CT interpretation by integrating eye‑tracking data into volumetric feature learning. Using a DINOv2 backbone, the model incorporates a learnable log‑space bias in self‑attention and gaze‑weighted pooling of patch embeddings. Trained on 182 CT reading sessions from five radiologists, it achieved an ROC‑AUC of 0.91 and an F1 score of 0.86, outperforming adapted baseline methods.
By Leila Khaertdinova, Anna Anikina, Claudia Mello-Thoms, Bulat Ibragimov
The study evaluates a deep‑learning model for pediatric pneumonia detection across three countries, testing not only discrimination but also probability calibration, fixed operating‑point transport, shortcut signals, and limited‑label recoverability. Using a DenseNet121 ensemble trained on Guangzhou data, the model achieved high internal AUROC (0.976) but performance dropped to 0.798 and 0.742 on Bangladeshi and Vietnamese datasets, respectively. Limited‑label adaptation with Platt recalibration restored sensitivity but introduced significant specificity variability, highlighting the need for comprehensive cross‑dataset evaluation.
The study evaluates a deep‑learning model for pediatric pneumonia detection across chest X‑ray datasets from three countries, assessing discrimination, calibration, operating‑point transport, shortcut signals, and limited‑label recovery. Using a frozen DenseNet121 ensemble trained on Guangzhou data, the model achieved high internal AUROC (0.976) but performance dropped when applied zero‑shot to Bangladesh (AUROC 0.798) and Vietnam (AUROC 0.742). Limited‑label adaptation with Platt recalibration restored sensitivity but introduced significant specificity variability, highlighting the need to evaluate multiple performance dimensions in cross‑dataset transport studies.
By Nazim-E-Alam
The study identifies a specific attention head (L8H10) in Vision Transformers that reads annotation text on chest X‑ray images instead of lung pathology, affecting 62% of COVID‑19 images. Causal inpainting experiments confirm that the head’s behavior is driven by annotation content, and removing the head increases lung attention by 1.52×. This spurious annotation‑attending behavior is consistent across multiple ViT architectures and datasets, persists after full fine‑tuning, and is invisible to standard output‑level evaluation, highlighting the need for mechanistic auditing.
By Joshua Muhumuza, Cornelius Kalenzi
Co-Annotator is a clinical AI system that distills expert gaze and dictation into two guidance components: a gaze‑aligned Vision Transformer that highlights fixation‑aligned areas of interest (AOIs) and an ontology‑bounded vision‑language model that pre‑fills editable biomarker summaries for retinal OCT. In controlled studies, each modality independently improved diagnostic accuracy and biomarker generation, and when combined across two academic institutions, the system increased correct diagnoses per minute by 40% and reduced comment editing time by 67% without compromising accuracy.
By Ziheng "Leo" Li, Benjamin Freeman, Akshay Raman, Kavin Aravindhan Rajkumar, Xinxin Fang, Rishabh Srivastava, Steven Feiner, Kaveri A. Thakoor
Current evaluation protocols for Vision-Language Models (VLMs) in Radiology Report Generation (RRG) rely on report-level metrics that measure lexical overlap or aggregate clinical correctness. However, such metrics do not test whether individual diagnostic statements stem from the actual pathological evidence visible in the image.
arXiv:2606. 15129v1 Announce Type: cross Abstract: Color fundus photography (CFP) is the mainstay for large-scale retinal screening, yet its diagnostic capacity is constrained by the lack of depth-resolved structural information.
By Zhuo Deng, Ruiheng Zhang, Ziheng Zhang, Weihao Gao, Yitong Li, Qian Wang, Lei Shao, Jiaoyue Dong, Zhixi Zeng, Lijian Fang, Haibo Wang, Xiaobin Lin, Tao Liu, Zhicheng Du, Zhengwei Zhang, Lin Yang, Zheng Gong, Xinyu Zhao, Zhenquan Wu, Fang Li, Zhiguang Zhou, Guoming Zhang, Sun Jing, Han Lv, Wenbin We, Lan Ma