SciZoom: A Large-scale Benchmark for Hierarchical Scientific Summarization across the LLM Era

The explosive growth of AI research has created unprecedented information overload, increasing the demand for scientific summarization at multiple levels of granularity beyond traditional abstracts. While LLMs are increasingly adopted for summarization, existing benchmarks remain limited in scale, target only a single granularity, and predate the LLM era. Moreover, since the release of ChatGPT in November 2022, researchers have rapidly adopted LLMs for drafting manuscripts themselves, fundamentally transforming scientific writing, yet no resource exists to analyze how this writing has evolved. To bridge these gaps, we introduce SciZoom, a benchmark comprising 44,946 papers from four top-tier ML venues (NeurIPS, ICLR, ICML, EMNLP) spanning 2020 to 2025, explicitly stratified into Pre-LLM and Post-LLM eras. SciZoom provides three hierarchical summarization targets (Abstract, Contributions, and TL;DR) achieving compression ratios up to 600:1, enabling both multi-granularity summarization research and temporal mining of scientific writing patterns. Our linguistic analysis reveals striking shifts in phrase patterns (up to 10x for formulaic expressions) and rhetorical style (23% decline in hedging), suggesting that LLM-assisted writing produces more confident yet homogenized prose. SciZoom serves as both a challenging benchmark and a unique resource for mining the evolution of scientific discourse in the generative AI era. Our code and dataset are publicly available on GitHub (https://github.com/janghana/SciZoom) and Hugging Face (https://huggingface.co/datasets/hanjang/SciZoom), respectively.

Segmentation-before-Staining Improves Structural Fidelity in Virtual IHC-to-Multiplex IF Translation

Multiplex immunofluorescence (mIF) enables simultaneous single-cell quantification of multiple biomarkers within intact tissue architecture, yet its high reagent cost, multi-round staining protocols, and need for specialized imaging platforms limit routine clinical adoption. Virtual staining can synthesize mIF channels from widely available brightfield immunohistochemistry (IHC), but current translators optimize pixel-level fidelity without explicitly constraining nuclear morphology. In pathology, this gap is clinically consequential: subtle distortions in nuclei count, shape, or spatial arrangement propagate directly to quantification endpoints such as the Ki67 proliferation index, where errors of a few percent can shift treatment-relevant risk categories. This work introduces a supervision-free, architecture-agnostic conditioning strategy that injects a continuous cell probability map from a pretrained nuclei segmentation foundation model as an explicit input prior, together with a variance-preserving regularization term that matches local intensity statistics to maintain cell-level heterogeneity in synthesized fluorescence channels. The soft prior retains gradient-level boundary information lost by binary thresholding, providing a richer conditioning signal without task-specific tuning. Controlled experiments across Pix2Pix with U-Net and ResNet generators, deterministic regression U-Net, and conditional diffusion on two independent datasets demonstrate consistent improvements in nuclei count fidelity and perceptual quality, as the sole modifications. Code will be made publicly available upon acceptance.

Evidential Perfusion Physics-Informed Neural Networks with Residual Uncertainty Quantification

Physics-informed neural networks (PINNs) have shown promise in addressing the ill-posed deconvolution problem in computed tomography perfusion (CTP) imaging for acute ischemic stroke assessment. However, existing PINN-based approaches remain deterministic and do not quantify uncertainty associated with violations of physics constraints, limiting reliability assessment. We propose Evidential Perfusion Physics-Informed Neural Networks (EPPINN), a framework that integrates evidential deep learning with physics-informed modeling to enable uncertainty-aware perfusion parameter estimation. EPPINN models arterial input, tissue concentration, and perfusion parameters using coordinate-based networks, and places a Normal--Inverse--Gamma distribution over the physics residual to characterize voxel-wise aleatoric and epistemic uncertainty in physics consistency without requiring Bayesian sampling or ensemble inference. The framework further incorporates physiologically constrained parameterization and stabilization strategies to promote robust per-case optimization. We evaluate EPPINN on digital phantom data, the ISLES 2018 benchmark, and a clinical cohort. On the evaluated datasets, EPPINN achieves lower normalized mean absolute error than classical deconvolution and PINN baselines, particularly under sparse temporal sampling and low signal-to-noise conditions, while providing conservative uncertainty estimates with high empirical coverage. On clinical data, EPPINN attains the highest voxel-level and case-level infarct-core detection sensitivity. These results suggest that evidential physics-informed learning can improve both accuracy and reliability of CTP analysis for time-critical stroke assessment.

MPIB: A Benchmark for Medical Prompt Injection Attacks and Clinical Safety in LLMs

Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) systems are increasingly integrated into clinical workflows; however, prompt injection attacks can steer these systems toward clinically unsafe or misleading outputs. We introduce the Medical Prompt Injection Benchmark (MPIB), a dataset-and-benchmark suite for evaluating clinical safety under both direct prompt injection and indirect, RAG-mediated injection across clinically grounded tasks. MPIB emphasizes outcome-level risk via the Clinical Harm Event Rate (CHER), which measures high-severity clinical harm events under a clinically grounded taxonomy, and reports CHER alongside Attack Success Rate (ASR) to disentangle instruction compliance from downstream patient risk. The benchmark comprises 9,697 curated instances constructed through multi-stage quality gates and clinical safety linting. Evaluating MPIB across a diverse set of baseline LLMs and defense configurations, we find that ASR and CHER can diverge substantially, and that robustness depends critically on whether adversarial instructions appear in the user query or in retrieved context. We release MPIB with evaluation code, adversarial baselines, and comprehensive documentation to support reproducible and systematic research on clinical prompt injection. Code and data are available at GitHub (code) and Hugging Face (data).