Explore the vast range of titles available.

With the rapid advancement of retrieval-augmented vision-language models, multimodal medical retrieval-augmented generation (MMed-RAG) systems are increasingly adopted in clinical decision support. These systems enhance medical applications by performing cross-modal retrieval to integrate relevant visual and textual evidence for tasks, e.g., report generation and disease diagnosis. However, their complex architecture also introduces underexplored adversarial vulnerabilities, particularly via visual input perturbations. In this paper, we propose Medusa, a novel framework for crafting cross-modal transferable adversarial attacks on MMed-RAG systems under a black-box setting. Specifically, Medusa formulates the attack as a perturbation optimization problem, leveraging a multi-positive InfoNCE loss (MPIL) to align adversarial visual embeddings with medically plausible but malicious textual targets, thereby hijacking the retrieval process. To enhance transferability, we adopt a surrogate model ensemble and design a dual-loop optimization strategy augmented with invariant risk minimization (IRM). Extensive experiments on two real-world medical tasks, including medical report generation and disease diagnosis, demonstrate that Medusa achieves over 90% average attack success rate across various generation models and retrievers under appropriate parameter configuration, while remaining robust against four mainstream defenses, outperforming state-of-the-art baselines. Our results reveal critical vulnerabilities in the MMed-RAG systems and highlight the necessity of robustness benchmarking in safety-critical medical applications. The code and data are available at https://anonymous.4open.science/r/MMed-RAG-Attack-F05A.

With the widespread adoption of pathology foundation models in both research and clinical decision support systems, exploring their security has become a critical concern. However, despite their growing impact, the vulnerability of these models to adversarial attacks remains largely unexplored. In this work, we present the first systematic investigation into the security of pathology foundation models for whole slide image~(WSI) analysis against adversarial attacks. Specifically, we introduce the principle of local perturbation with global impact and propose a label-free attack framework that operates without requiring access to downstream task labels. Under this attack framework, we revise four classical white-box attack methods and redefine the perturbation budget based on the characteristics of WSI. We conduct comprehensive experiments on three representative pathology foundation models across five datasets and six downstream tasks. Despite modifying only 0.1\\% of patches per slide with imperceptible noise, our attack leads to downstream accuracy degradation that can reach up to 20\\% in the worst cases. Furthermore, we analyze key factors that influence attack success, explore the relationship between patch-level vulnerability and semantic content, and conduct a preliminary investigation into potential defence strategies. These findings lay the groundwork for future research on the adversarial robustness and reliable deployment of pathology foundation models. Our code is publicly available at: https://github.com/Jiashuai-Liu-hmos/Attack-WSI-pathology-foundation-models.

The treatment response to neoadjuvant immunochemotherapy varies among patients with potentially resectable non-small cell lung cancers (NSCLC) and may have severe immune-related adverse effects. We are currently unable to accurately predict therapeutic response. We aimed to develop a radiomics-based nomogram to predict a major pathological response (MPR) of potentially resectable NSCLC to neoadjuvant immunochemotherapy using pretreatment computed tomography (CT) images and clinical characteristics. A total of 89 eligible participants were included and randomly divided into training (N=64) and validation (N=25) sets. Radiomic features were extracted from tumor volumes of interest in pretreatment CT images. Following data dimension reduction, feature selection, and radiomic signature building, a radiomics-clinical combined nomogram was developed using logistic regression analysis. The radiomics-clinical combined model achieved excellent discriminative performance, with AUCs of 0.84 (95% CI, 0.74-0.93) and 0.81(95% CI, 0.63-0.98) and accuracies of 80% and 80% in the training and validation sets, respectively. Decision curves analysis (DCA) indicated that the radiomics-clinical combined nomogram was clinically valuable. The constructed nomogram was able to predict MPR to neoadjuvant immunochemotherapy with a high degree of accuracy and robustness, suggesting that it is a convenient tool for assisting with the individualized management of patients with potentially resectable NSCLC.

Background Angiotensin receptor-neprilysin inhibitors (ARNIs) could improve the outcome of patients with hypertension or heart failure. Yet it remains unclear if such protective effect exists in lung cancer patients with hypertension undergoing immune checkpoint inhibitors (ICIs). Objective This study sought to evaluate the impact of ARNIs on the outcome of lung cancer patients with hypertension treated with ICIs. Methods We performed a retrospective analysis of lung cancer patients with concurrent hypertension undergoing ICIs treatment between September 2019 and May 2024 in Shanghai Chest Hospital. The overall population included patients with or without ARNIs therapy. The primary endpoint was all-cause mortality and the secondary endpoint was a composite of cardiovascular events, including myocarditis, heart failure (HF), acute coronary syndrome (ACS), pericardial effusion (PE), and new-onset arrhythmias. Results The study enrolled 153 patients, with 39 (25.5%) patients in ARNIs group and 114 (74.5%) patients in control group. Over a median follow-up of 12 months, 42 (27.5%) patients died, ARNIs treatment was significantly associated with improved overall survival for lung cancer patients with hypertension and treated with ICIs (89.7% vs. 66.7%, HR = 0.27; 95% CI 0.10–0.75, log-rank P  = 0.007). However, there was no significant difference in cardiovascular outcome between ARNIs and control group (21/114, 18.4% vs. 3/39, 7.7%, P  = 0.112). During follow-up, a total of 24 cardiovascular adverse events occurred, of which the most common cardiovascular event was new-onset arrhythmia in both ARNIs group (2/3, 66.7%) and control group (14/21, 66.7%). Conclusions The use of ARNIs seemed associated with a lower all-cause mortality rate yet similar rate of cardiac events in lung cancer patients with hypertension and treated with ICIs. The findings of this single-center retrospective study warrant validation in larger prospective studies.

Key message One hundred and fifty-five QTL for trace element concentrations in foxtail millet were identified using a genome-wide association study, and a candidate gene associated with Ni–Co–Cr concentrations was detected. Foxtail millet ( Setaria italica ) is an important regional crop known for its rich mineral nutrient content, which has beneficial effects on human health. We assessed the concentrations of ten trace elements (Ba, Co, Cr, Cu, Fe, Mn, Ni, Pb, Sr, and Zn) in the grain of 408 foxtail millet accessions. Significant differences in the concentrations of five elements (Ba, Co, Ni, Sr, and Zn) were observed between two subpopulations of spring- and summer-sown foxtail millet varieties. Moreover, 84.4% of the element pairs exhibited significant correlations. To identify the genetic factors influencing trace element accumulation, a comprehensive genome-wide association study was conducted, identifying 155 quantitative trait locus (QTL) for the ten trace elements across three different environments. Among them, ten QTL were consistently detected in multiple environments, including qZn2.1, qZn4.4, qCr4.1, qFe6.3, qFe6.5, qCo6.1, qPb7.3, qPb7.5, qBa9.1, and qNi9.1 . Thirteen QTL clusters were detected for multiple elements, which partially explained the correlations between elements. Additionally, the different concentrations of five elements between foxtail millet subpopulations were caused by the different frequencies of high-concentration alleles associated with important marker-trait associations. Haplotype analysis identified a candidate gene SETIT_036676mg associated with Ni accumulation, with the GG haplotype significantly increasing Ni–Co–Cr concentrations in foxtail millet. A cleaved amplified polymorphic sequence marker (cNi6676) based on the two haplotypes of SETIT_036676mg was developed and validated. Results of this study provide valuable reference information for the genetic research and improvement of trace element content in foxtail millet.

Cardiac ischemia/reperfusion (I/R) injury is an important therapeutic target for ischemic heart disease. Lipid droplets (LDs) are the key organelles involved in lipid metabolism. This study aimed to identify the LD‐mediated protection against lipotoxicity in cardiac I/R injury. LD accumulation is upregulated in hearts subjected to I/R injury; however, it is insufficient to neutralize lipotoxicity or prevent cardiomyocyte death. Seipin played a central role in LD biogenesis in cardiomyocytes following I/R injury. Seipin deficiency led to reduced LD levels and exacerbated cardiac I/R injury. Whereas increased LD levels, via Seipin overexpression or lipolysis inhibition, ameliorated myocardial I/R injury. I/R‐induced downregulation of Seipin is attributed to the reduced expression of its transcription factor USF1, which is required for metabolic adaptation in acute myocardial ischemia. These findings not only elucidate the pathophysiological roles of LDs and Seipin but also provide a promising therapeutic target for myocardial I/R injury. Lipid droplets (LDs) protect the heart against lipotoxicity in cardiac ischemia/reperfusion (I/R) injury; however, they are insufficient to prevent cardiomyocyte death. Seipin plays a central role in the insufficient formation of LDs, subsequent lipotoxicity, and myocardial injury during cardiac I/R injury. Genetic or pharmacological treatments to improve LDs levels could alleviate I/R‐induced cardiomyocyte death and myocardial injury.

This report summarizes the results of Learning to Understand Aerial Images (LUAI) 2021 challenge held on ICCV 2021, which focuses on object detection and semantic segmentation in aerial images. Using DOTA-v2.0 and GID-15 datasets, this challenge proposes three tasks for oriented object detection, horizontal object detection, and semantic segmentation of common categories in aerial images. This challenge received a total of 146 registrations on the three tasks. Through the challenge, we hope to draw attention from a wide range of communities and call for more efforts on the problems of learning to understand aerial images.