Lanfen Lin

Synthesizing multi-phase contrast-enhanced CT images from non-contrast CT (NCCT) may provide complementary phase-specific cues for preliminary assessment. After rigorous clinical validation, such images could provide clinical decision support or aid diagnostic triage by identifying cases that warrant further acquired contrast-enhanced CT (CECT) work-up; they are not intended to replace acquired CECT. Arterial-phase (ART) and portal-venous-phase (PV) images share anatomical structures but exhibit distinct enhancement patterns and intensity distributions, which makes simultaneous multi-phase synthesis challenging for conventional single-decoder models. We propose a task-prompt-guided dual-decoder framework that combines a shared Swin Transformer encoder, two phase-specific decoders, learnable task prompts initialized from Qwen3-8B semantic representations, two phase-specific adversarial discriminators, and an independent ART/PV domain classifier. The shared encoder extracts phase-invariant anatomical features, whereas the two decoders independently model ART- and PV-specific enhancement. The Qwen3-derived prompt vectors provide phase-aware initialization and subsequently adapt through prompt–feature interaction at the bottleneck. Experiments on a single-center dataset of 86 patients show improved whole-image and lesion-focused PSNR, SSIM, MSE, and PCC relative to Pix2pix and MedGAN. Controlled ablation experiments indicate the benefits of Qwen3-based prompt initialization, subsequent prompt adaptation, and correct prompt–decoder correspondence; a reduced baseline additionally assesses the combined removal of LLTP and ART/PV domain classification. In a downstream slice-level four-class focal liver lesion classification experiment, synthetic multiphase input improved accuracy from 71.65% with NCCT alone to 85.04%, compared with 91.34% for acquired multiphase CT. These findings provide a proof of concept for LLM-guided multi-phase CT synthesis, although external validation, clinically oriented safety assessment, and reader studies remain necessary before clinical use for decision support or diagnostic triage.

Understanding human perceptions poses a significant multimodal challenge for computers, involving textual, acoustic, and visual signals. Recently, large language models has garnered great attention, leading to numerous methods aimed at efficiently fine-tuning pretrained models for multimodal downstream tasks. However, there remains a scarcity of techniques that prioritize modality-invariant and -specific information during parameter-efficient tuning, despite evidence from previous studies showcasing the effectiveness of modality disentangling. To address this gap, we propose a novel multimodal tuning approach for large language models, termed as Disentangled Multimodal Tuning and Interaction. Specifically, we evaluate the independence among different modalities and disentangle corresponding modality-invariant and specific components, which are subsequently leveraged for prompt tuning. Following tuning, a newly designed independence-guided cross-attention module is introduced for modality interaction, where the attention mechanism is decoupled and bolstered with independence from the modality-disentangling process. This approach not only enables large language models to efficiently assimilate information from various modalities but also cultivates an awareness of both modality-invariant and specific information. Compared to previous methods, our approach facilitates modality interaction at a more granular level, resulting in enhanced performance. We validate our method through experiments on four public datasets, demonstrating significant performance improvements.

Depression, a pervasive mental health condition, affects millions globally, challenging early and accurate diagnosis due to its subtle and varied manifestations. Recognizing the critical link between emotional dysregulation and depressive symptoms, our research introduces a pioneering training paradigm that integrates sentiment analysis with depression detection. This approach is motivated by the potential of sentiment data to enrich models with a deeper understanding of emotional states, crucial for identifying depressive patterns. To leverage the nuanced sentiment information without compromising the pretrained model’s integrity, we employ deep prefix tuning. This novel technique allows for targeted model refinement, ensuring that the valuable pretrained structures are not overshadowed by the sparse and specific nature of depression-related data. The empirical results demonstrate superior performance across standard benchmarks, setting a new precedent for multimodal depression detection.

Explainable artificial intelligence (XAI) is increasingly important for computational pathology, where reliable and interpretable predictions are required for clinical use. Pre-trained Vision-Language Models (VLMs) offer a natural pathway to connect visual evidence with textual concepts, but adapting them to the domain of cancer pathology remains challenging due to fine-grained and heterogeneous morphological patterns. Visual prompt learning enables efficient task adaptation with minimal fine-tuning; however, existing prompting techniques face critical limitations: soft prompts often lack clinical specificity, while manually designed hard prompts reduce adaptability. To address these issues, we propose RAPT, a retrieval-augmented and text-guided visual prompting framework for explainable pathology classification. Given an input image, RAPT retrieves semantically related exemplars and leverages class-specific textual descriptions to construct disease-aware prompt tokens, injecting diagnostic cues without manual annotation. An adaptive weighting mechanism attenuates unreliable retrieval evidence, and bridge prompt tokens facilitate the integration of retrieved cues with image representations. Extensive experiments on three public cancer pathology datasets (PatchGastric, BACH, and LC25000) demonstrate that RAPT consistently outperforms prompting baselines across diverse backbone settings. Additional analyses and qualitative case studies demonstrate clinically actionable cues, robustness to imperfect retrieval, and clear failure-mode boundaries for trustworthy pathology decision support, highlighting the potential of robust and explainable prompting for computational pathology.

Unpaired H&E-to-IHC Stain Translation aims to generate immunohistochemistry (IHC) staining from Hematoxylin and Eosin (H&E) staining. It offers clearer diagnostic insights and potentially expands access to advanced pathology services in resource-limited areas. This task faces two primary challenges: capturing target domain style characteristics and preserving topological features in histological images. Recently, Schrödinger Bridge (SB)-based methods have offered a solution for unpaired image-to-image translation, addressing the mode collapse and artifact issues in CycleGAN-based approaches, as well as the Gaussian prior assumption limitation in diffusion-based methods. While SB-based methods suffer from the curse of dimensionality with high-resolution images, the Unpaired Neural Schrödinger Bridge (UNSB) overcomes this challenge and achieves state-of-the-art (SOTA) performance on natural images. However, UNSB has two key issues in histological images: (1) loss of topological features and (2) IHC staining representation. UNSB focuses only on the optimal path from source to target domains, ignoring local structure paths. Convolutional neural networks (CNNs) do not perfectly preserve critical anatomical structures due to limitations like receptive field size or model capacity. To address these challenges, we introduce the Topology-aware Diffusion Schrödinger Bridge (TDSB), integrating a Topology Guidance (TG) module and Dual-Domain Adaptive Patch-based noise contrastive estimation (DDAP). Experiments on seven translation tasks across three datasets show that our method achieves SOTA performance in unpaired H&E-to-IHC stain translation. Clinical evaluation through pathologists' assessments further validates the effectiveness of our method.