Jian Jiang

Zhitao Zeng, Mengya Xu, Jian Jiang et al.

Surgical intelligence has the potential to improve the safety and consistency of surgical care, yet most existing surgical AI frameworks remain task-specific and struggle to generalize across procedures and institutions. Although multimodal foundation models, particularly multimodal large language models, have demonstrated strong cross-task capabilities across various medical domains, their advancement in surgery remains constrained by the lack of large-scale, systematically curated multimodal data. To address this challenge, we introduce Surg$Σ$, a spectrum of large-scale multimodal data and foundation models for surgical intelligence. At the core of this framework lies Surg$Σ$-DB, a large-scale multimodal data foundation designed to support diverse surgical tasks. Surg$Σ$-DB consolidates heterogeneous surgical data sources (including open-source datasets, curated in-house clinical collections and web-source data) into a unified schema, aiming to improve label consistency and data standardization across heterogeneous datasets. Surg$Σ$-DB spans 6 clinical specialties and diverse surgical types, providing rich image- and video-level annotations across 18 practical surgical tasks covering understanding, reasoning, planning, and generation, at an unprecedented scale (over 5.98M conversations). Beyond conventional multimodal conversations, Surg$Σ$-DB incorporates hierarchical reasoning annotations, providing richer semantic cues to support deeper contextual understanding in complex surgical scenarios. We further provide empirical evidence through recently developed surgical foundation models built upon Surg$Σ$-DB, illustrating the practical benefits of large-scale multimodal annotations, unified semantic design, and structured reasoning annotations for improving cross-task generalization and interpretability.

Long Chen, Jian Jiang, Bozheng Dou et al.

Pain is a significant global health issue, and the current treatment options for pain management have limitations in terms of effectiveness, side effects, and potential for addiction. There is a pressing need for improved pain treatments and the development of new drugs. Voltage-gated sodium channels, particularly Nav1.3, Nav1.7, Nav1.8, and Nav1.9, play a crucial role in neuronal excitability and are predominantly expressed in the peripheral nervous system. Targeting these channels may provide a means to treat pain while minimizing central and cardiac adverse effects. In this study, we construct protein-protein interaction (PPI) networks based on pain-related sodium channels and develop a corresponding drug-target interaction (DTI) network to identify potential lead compounds for pain management. To ensure reliable machine learning predictions, we carefully select 111 inhibitor datasets from a pool of over 1,000 targets in the PPI network. We employ three distinct machine learning algorithms combined with advanced natural language processing (NLP)-based embeddings, specifically pre-trained transformer and autoencoder representations. Through a systematic screening process, we evaluate the side effects and repurposing potential of over 150,000 drug candidates targeting Nav1.7 and Nav1.8 sodium channels. Additionally, we assess the ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties of these candidates to identify leads with near-optimal characteristics. Our strategy provides an innovative platform for the pharmacological development of pain treatments, offering the potential for improved efficacy and reduced side effects.

Zhiyuan Shi, Qibo Qiu, Feng Xue et al.

The linear memory growth of the KV cache poses a significant bottleneck for LLM inference in long-context tasks. Existing static compression methods often fail to preserve globally important information, principally because they overlook the attention drift phenomenon where token significance evolves dynamically. Although recent dynamic retrieval approaches attempt to address this issue, they typically suffer from coarse-grained caching strategies and incur high I/O overhead due to frequent data transfers. To overcome these limitations, we propose HeteroCache, a training-free dynamic compression framework. Our method is built on two key insights: attention heads exhibit diverse temporal heterogeneity, and there is significant spatial redundancy among heads within the same layer. Guided by these insights, HeteroCache categorizes heads based on stability and redundancy. Consequently, we apply a fine-grained weighting strategy that allocates larger cache budgets to heads with rapidly shifting attention to capture context changes, thereby addressing the inefficiency of coarse-grained strategies. Furthermore, we employ a hierarchical storage mechanism in which a subset of representative heads monitors attention shift, and trigger an asynchronous, on-demand retrieval of contexts from the CPU, effectively hiding I/O latency. Finally, experiments demonstrate that HeteroCache achieves state-of-the-art performance on multiple long-context benchmarks and accelerates decoding by up to $3\times$ compared to the original model in the 224K context. Our code will be open-source.

Xinxin Zhao, Jian Jiang, Yan Tian et al.

Tooth image segmentation is a cornerstone of dental digitization. However, traditional image encoders relying on fixed-resolution feature maps often lead to discontinuous segmentation and poor discrimination between target regions and background, due to insufficient modeling of environmental and global context. Moreover, transformer-based self-attention introduces substantial computational overhead because of its quadratic complexity (O(n^2)), making it inefficient for high-resolution dental images. To address these challenges, we introduce a three-stage encoder with hierarchical feature representation to capture scale-adaptive information in dental images. By jointly leveraging low-level details and high-level semantics through cross-scale feature fusion, the model effectively preserves fine structural information while maintaining strong contextual awareness. Furthermore, a bidirectional sequence modeling strategy is incorporated to enhance global spatial context understanding without incurring high computational cost. We validate our method on two dental datasets, with experimental results demonstrating its superiority over existing approaches. On the OralVision dataset, our model achieves a 1.1% improvement in mean intersection over union (mIoU).

Jian Jiang, Oya Celiktutan

Task incremental learning aims to enable a system to maintain its performance on previously learned tasks while learning new tasks, solving the problem of catastrophic forgetting. One promising approach is to build an individual network or sub-network for future tasks. However, this leads to an ever-growing memory due to saving extra weights for new tasks and how to address this issue has remained an open problem in task incremental learning. In this paper, we introduce a novel Neural Weight Search technique that designs a fixed search space where the optimal combinations of frozen weights can be searched to build new models for novel tasks in an end-to-end manner, resulting in scalable and controllable memory growth. Extensive experiments on two benchmarks, i.e., Split-CIFAR-100 and CUB-to-Sketches, show our method achieves state-of-the-art performance with respect to both average inference accuracy and total memory cost.

Yujia Xu, Hak-Keung Lam, Guangyu Jia et al.

COVID-19 pandemic continues to spread rapidly over the world and causes a tremendous crisis in global human health and the economy. Its early detection and diagnosis are crucial for controlling the further spread. Many deep learning-based methods have been proposed to assist clinicians in automatic COVID-19 diagnosis based on computed tomography imaging. However, challenges still remain, including low data diversity in existing datasets, and unsatisfied detection resulting from insufficient accuracy and sensitivity of deep learning models. To enhance the data diversity, we design augmentation techniques of incremental levels and apply them to the largest open-access benchmark dataset, COVIDx CT-2A. Meanwhile, similarity regularization (SR) derived from contrastive learning is proposed in this study to enable CNNs to learn more parameter-efficient representations, thus improving the accuracy and sensitivity of CNNs. The results on seven commonly used CNNs demonstrate that CNN performance can be improved stably through applying the designed augmentation and SR techniques. In particular, DenseNet121 with SR achieves an average test accuracy of 99.44% in three trials for three-category classification, including normal, non-COVID-19 pneumonia, and COVID-19 pneumonia. And the achieved precision, sensitivity, and specificity for the COVID-19 pneumonia category are 98.40%, 99.59%, and 99.50%, respectively. These statistics suggest that our method has surpassed the existing state-of-the-art methods on the COVIDx CT-2A dataset.

Lu Han, Yu Liu, Qiwen Deng et al.

Time Series Foundation Models (TSFMs) have achieved remarkable success through large-scale pretraining. However, their design primarily targets real-valued series, limiting their ability to handle general forecasting tasks involving diverse and often heterogeneous covariates--such as categorical variables and multimodal data (e.g., images, text)--which are typically task-specific and difficult to leverage during pretraining. To address this gap, we propose Unified Covariate Adaptation (UniCA), a framework to bridge TSFMs with general covariate-aware forecasting. UniCA first performs covariate homogenization to transform heterogeneous covariates into high-level homogeneous series representations and then fuses them via a unified attention-based fusion mechanism. UniCA is compatible and universal for adaptation with both homogeneous and heterogeneous covariates, incorporating extra covariate information while preserving the generalization ability of TSFMs.Extensive experiments on multiple unimodal and multimodal covariate-aware forecasting benchmarks demonstrate the superiority of UniCA, highlighting the promise of covariate-aware TSFM adaptation in real-world forecasting scenarios. Codes are released on https://github.com/hanlu-nju/UniCA.

Matej Vitek, Darian Tomašević, Abhijit Das et al.

This paper presents a summary of the 2025 Sclera Segmentation Benchmarking Competition (SSBC), which focused on the development of privacy-preserving sclera-segmentation models trained using synthetically generated ocular images. The goal of the competition was to evaluate how well models trained on synthetic data perform in comparison to those trained on real-world datasets. The competition featured two tracks: $(i)$ one relying solely on synthetic data for model development, and $(ii)$ one combining/mixing synthetic with (a limited amount of) real-world data. A total of nine research groups submitted diverse segmentation models, employing a variety of architectural designs, including transformer-based solutions, lightweight models, and segmentation networks guided by generative frameworks. Experiments were conducted across three evaluation datasets containing both synthetic and real-world images, collected under diverse conditions. Results show that models trained entirely on synthetic data can achieve competitive performance, particularly when dedicated training strategies are employed, as evidenced by the top performing models that achieved $F_1$ scores of over $0.8$ in the synthetic data track. Moreover, performance gains in the mixed track were often driven more by methodological choices rather than by the inclusion of real data, highlighting the promise of synthetic data for privacy-aware biometric development. The code and data for the competition is available at: https://github.com/dariant/SSBC_2025.

Ivan Perov, Daiheng Gao, Nikolay Chervoniy et al.

Deepfake defense not only requires the research of detection but also requires the efforts of generation methods. However, current deepfake methods suffer the effects of obscure workflow and poor performance. To solve this problem, we present DeepFaceLab, the current dominant deepfake framework for face-swapping. It provides the necessary tools as well as an easy-to-use way to conduct high-quality face-swapping. It also offers a flexible and loose coupling structure for people who need to strengthen their pipeline with other features without writing complicated boilerplate code. We detail the principles that drive the implementation of DeepFaceLab and introduce its pipeline, through which every aspect of the pipeline can be modified painlessly by users to achieve their customization purpose. It is noteworthy that DeepFaceLab could achieve cinema-quality results with high fidelity. We demonstrate the advantage of our system by comparing our approach with other face-swapping methods.For more information, please visit:https://github.com/iperov/DeepFaceLab/.

Haonan Ping, Jian Jiang, Cheng Yuan et al.

Accurate segmentation of tissues and instruments in surgical scenes is annotation-intensive due to irregular shapes, thin structures, specularities, and frequent occlusions. While SAM models support point, box, and mask prompts, points are often too sparse and boxes too coarse to localize such challenging targets. We present SCISSR, a scribble-promptable framework for interactive surgical scene segmentation. It introduces a lightweight Scribble Encoder that converts freehand scribbles into dense prompt embeddings compatible with the mask decoder, enabling iterative refinement for a target object by drawing corrective strokes on error regions. Because all added modules (the Scribble Encoder, Spatial Gated Fusion, and LoRA adapters) interact with the backbone only through its standard embedding interfaces, the framework is not tied to a single model: we build on SAM 2 in this work, yet the same components transfer to other prompt-driven segmentation architectures such as SAM 3 without structural modification. To preserve pre-trained capabilities, we train only these lightweight additions while keeping the remaining backbone frozen. Experiments on EndoVis 2018 demonstrate strong in-domain performance, while evaluation on the out-of-distribution CholecSeg8k further confirms robustness across surgical domains. SCISSR achieves 95.41% Dice on EndoVis 2018 with five interaction rounds and 96.30% Dice on CholecSeg8k with three interaction rounds, outperforming iterative point prompting on both benchmarks.

Zhitao Zeng, Zhu Zhuo, Xiaojun Jia et al. · pku

Foundation models have achieved transformative success across biomedical domains by enabling holistic understanding of multimodal data. However, their application in surgery remains underexplored. Surgical intelligence presents unique challenges - requiring surgical visual perception, temporal analysis, and reasoning. Existing general-purpose vision-language models fail to address these needs due to insufficient domain-specific supervision and the lack of a large-scale high-quality surgical database. To bridge this gap, we propose SurgVLM, one of the first large vision-language foundation models for surgical intelligence, where this single universal model can tackle versatile surgical tasks. To enable this, we construct a large-scale multimodal surgical database, SurgVLM-DB, comprising over 1.81 million frames with 7.79 million conversations, spanning more than 16 surgical types and 18 anatomical structures. We unify and reorganize 23 public datasets across 10 surgical tasks, followed by standardizing labels and doing hierarchical vision-language alignment to facilitate comprehensive coverage of gradually finer-grained surgical tasks, from visual perception, temporal analysis, to high-level reasoning. Building upon this comprehensive dataset, we propose SurgVLM, which is built upon Qwen2.5-VL, and undergoes instruction tuning to 10+ surgical tasks. We further construct a surgical multimodal benchmark, SurgVLM-Bench, for method evaluation. SurgVLM-Bench consists of 6 popular and widely-used datasets in surgical domain, covering several crucial downstream tasks. Based on SurgVLM-Bench, we evaluate the performance of our SurgVLM (3 SurgVLM variants: SurgVLM-7B, SurgVLM-32B, and SurgVLM-72B), and conduct comprehensive comparisons with 14 mainstream commercial VLMs (e.g., GPT-4o, Gemini 2.0 Flash, Qwen2.5-Max).

Cheng Yuan, Jian Jiang, Kunyi Yang et al.

Surgical video segmentation is critical for AI to interpret spatial-temporal dynamics in surgery, yet model performance is constrained by limited annotated data. The SAM2 model, pretrained on natural videos, offers potential for zero-shot surgical segmentation, but its applicability in complex surgical environments, with challenges like tissue deformation and instrument variability, remains unexplored. We present the first comprehensive evaluation of the zero-shot capability of SAM2 in 9 surgical datasets (17 surgery types), covering laparoscopic, endoscopic, and robotic procedures. We analyze various prompting (points, boxes, mask) and {finetuning (dense, sparse) strategies}, robustness to surgical challenges, and generalization across procedures and anatomies. Key findings reveal that while SAM2 demonstrates notable zero-shot adaptability in structured scenarios (e.g., instrument segmentation, {multi-organ segmentation}, and scene segmentation), its performance varies under dynamic surgical conditions, highlighting gaps in handling temporal coherence and domain-specific artifacts. These results highlight future pathways to adaptive data-efficient solutions for the surgical data science field.

Qibo Chen, Weizhong Jin, Jianyue Ge et al.

Recent research on universal object detection aims to introduce language in a SoTA closed-set detector and then generalize the open-set concepts by constructing large-scale (text-region) datasets for training. However, these methods face two main challenges: (i) how to efficiently use the prior information in the prompts to genericise objects and (ii) how to reduce alignment bias in the downstream tasks, both leading to sub-optimal performance in some scenarios beyond pre-training. To address these challenges, we propose a strong universal detection foundation model called CP-DETR, which is competitive in almost all scenarios, with only one pre-training weight. Specifically, we design an efficient prompt visual hybrid encoder that enhances the information interaction between prompt and visual through scale-by-scale and multi-scale fusion modules. Then, the hybrid encoder is facilitated to fully utilize the prompted information by prompt multi-label loss and auxiliary detection head. In addition to text prompts, we have designed two practical concept prompt generation methods, visual prompt and optimized prompt, to extract abstract concepts through concrete visual examples and stably reduce alignment bias in downstream tasks. With these effective designs, CP-DETR demonstrates superior universal detection performance in a broad spectrum of scenarios. For example, our Swin-T backbone model achieves 47.6 zero-shot AP on LVIS, and the Swin-L backbone model achieves 32.2 zero-shot AP on ODinW35. Furthermore, our visual prompt generation method achieves 68.4 AP on COCO val by interactive detection, and the optimized prompt achieves 73.1 fully-shot AP on ODinW13.

Jian Jiang, Chenxi Lin, Yiming Gu et al.

Surgical scene understanding demands not only accurate predictions but also interpretable reasoning that surgeons can verify against clinical expertise. However, existing surgical vision-language models generate predictions without reasoning chains, and general-purpose reasoning models fail on compositional surgical tasks without domain-specific knowledge. We present Surg-R1, a surgical Vision-Language Model that addresses this gap through hierarchical reasoning trained via a four-stage pipeline. Our approach introduces three key contributions: (1) a three-level reasoning hierarchy decomposing surgical interpretation into perceptual grounding, relational understanding, and contextual reasoning; (2) the largest surgical chain-of-thought dataset with 320,000 reasoning pairs; and (3) a four-stage training pipeline progressing from supervised fine-tuning to group relative policy optimization and iterative self-improvement. Evaluation on SurgBench, comprising six public benchmarks and six multi-center external validation datasets from five institutions, demonstrates that Surg-R1 achieves the highest Arena Score (64.9%) on public benchmarks versus Gemini 3.0 Pro (46.1%) and GPT-5.1 (37.9%), outperforming both proprietary reasoning models and specialized surgical VLMs on the majority of tasks spanning instrument localization, triplet recognition, phase recognition, action recognition, and critical view of safety assessment, with a 15.2 percentage point improvement over the strongest surgical baseline on external validation.

Qibo Chen, Weizhong Jin, Shuchang Li et al.

Text prompts are crucial for generalizing pre-trained open-set object detection models to new categories. However, current methods for text prompts are limited as they require manual feedback when generalizing to new categories, which restricts their ability to model complex scenes, often leading to incorrect detection results. To address this limitation, we propose a novel visual prompt method that learns new category knowledge from a few labeled images, which generalizes the pre-trained detection model to the new category. To allow visual prompts to represent new categories adequately, we propose a statistical-based prompt construction module that is not limited by predefined vocabulary lengths, thus allowing more vectors to be used when representing categories. We further utilize the category dictionaries in the pre-training dataset to design task-specific similarity dictionaries, which make visual prompts more discriminative. We evaluate the method on the ODinW dataset and show that it outperforms existing prompt learning methods and performs more consistently in combinatorial inference.

Wu Hao Ran, Xi Xi, Furong Li et al.

The advent of large language models (LLMs) has opened new avenues for analyzing complex, unstructured data, particularly within the medical domain. Electronic Health Records (EHRs) contain a wealth of information in various formats, including free text clinical notes, structured lab results, and diagnostic codes. This paper explores the application of advanced language models to leverage these diverse data sources for improved clinical decision support. We will discuss how text-based features, often overlooked in traditional high dimensional EHR analysis, can provide semantically rich representations and aid in harmonizing data across different institutions. Furthermore, we delve into the challenges and opportunities of incorporating medical codes and ensuring the generalizability and fairness of AI models in healthcare.

Jian Jiang, Long Chen, Yueying Zhu et al.

Anesthetics are crucial in surgical procedures and therapeutic interventions, but they come with side effects and varying levels of effectiveness, calling for novel anesthetic agents that offer more precise and controllable effects. Targeting Gamma-aminobutyric acid (GABA) receptors, the primary inhibitory receptors in the central nervous system, could enhance their inhibitory action, potentially reducing side effects while improving the potency of anesthetics. In this study, we introduce a proteomic learning of GABA receptor-mediated anesthesia based on 24 GABA receptor subtypes by considering over 4000 proteins in protein-protein interaction (PPI) networks and over 1.5 millions known binding compounds. We develop a corresponding drug-target interaction network to identify potential lead compounds for novel anesthetic design. To ensure robust proteomic learning predictions, we curated a dataset comprising 136 targets from a pool of 980 targets within the PPI networks. We employed three machine learning algorithms, integrating advanced natural language processing (NLP) models such as pretrained transformer and autoencoder embeddings. Through a comprehensive screening process, we evaluated the side effects and repurposing potential of over 180,000 drug candidates targeting the GABRA5 receptor. Additionally, we assessed the ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties of these candidates to identify those with near-optimal characteristics. This approach also involved optimizing the structures of existing anesthetics. Our work presents an innovative strategy for the development of new anesthetic drugs, optimization of anesthetic use, and deeper understanding of potential anesthesia-related side effects.

Jian Jiang

Detecting microbial biomarkers used to predict disease phenotypes and clinical outcomes is crucial for disease early-stage screening and diagnosis. Most methods for biomarker identification are linear-based, which is very limited as biological processes are rarely fully linear. The introduction of machine learning to this field tends to bring a promising solution. However, identifying microbial biomarkers in an interpretable, data-driven and robust manner remains challenging. We present MIIDL, a Python package for the identification of microbial biomarkers based on interpretable deep learning. MIIDL innovatively applies convolutional neural networks, a variety of interpretability algorithms and plenty of pre-processing methods to provide a one-stop and robust pipeline for microbial biomarkers identification from high-dimensional and sparse data sets.

Jian Jiang, Edoardo Cetin, Oya Celiktutan

Incremental learning aims to enable machine learning models to continuously acquire new knowledge given new classes, while maintaining the knowledge already learned for old classes. Saving a subset of training samples of previously seen classes in the memory and replaying them during new training phases is proven to be an efficient and effective way to fulfil this aim. It is evident that the larger number of exemplars the model inherits the better performance it can achieve. However, finding a trade-off between the model performance and the number of samples to save for each class is still an open problem for replay-based incremental learning and is increasingly desirable for real-life applications. In this paper, we approach this open problem by tapping into a two-step compression approach. The first step is a lossy compression, we propose to encode input images and save their discrete latent representations in the form of codes that are learned using a hierarchical Vector Quantised Variational Autoencoder (VQ-VAE). In the second step, we further compress codes losslessly by learning a hierarchical latent variable model with bits-back asymmetric numeral systems (BB-ANS). To compensate for the information lost in the first step compression, we introduce an Information Back (IB) mechanism that utilizes real exemplars for a contrastive learning loss to regularize the training of a classifier. By maintaining all seen exemplars' representations in the format of `codes', Discrete Representation Replay (DRR) outperforms the state-of-art method on CIFAR-100 by a margin of 4% accuracy with a much less memory cost required for saving samples. Incorporated with IB and saving a small set of old raw exemplars as well, the accuracy of DRR can be further improved by 2% accuracy.