用强度信息指导去噪，提升荧光寿命成像的信噪比和生物信息保真度。
Zero-Shot Denoising for Fluorescence Lifetime Imaging Microscopy with Intensity-Guided Learning
- 通过强度引导的跨通道去噪，保留多模态数据相关性
- 零样本框架在真实生物样本上显著降低噪声并保留寿命信息
- 适合需要高保真生理分子分析的生物成像研究者
荧光寿命成像显微镜（FLIM）等多模态成像技术拓展了传统荧光强度成像的信息维度，但因复杂噪声模式导致图像质量下降。由于强度与寿命信息存在内在关联，各通道噪声呈多变量依赖关系，且未必共享结构特征。为此，本文提出一种基于强度引导学习的零样本去噪框架。该框架采用独立通道处理路径，并利用预训练的强度去噪先验，指导多个通道中寿命成分的优化。通过在真实世界获取的生物样本上的实验验证，该方法在降噪和寿命信息保持方面均优于现有方法，从而实现更可靠的生理与分子信息提取。
原文摘要 · Abstract (English)
Multimodal and multi-information microscopy techniques such as Fluorescence Lifetime Imaging Microscopy (FLIM) extend the informational channels beyond intensity-based fluorescence microscopy but suffer from reduced image quality due to complex noise patterns. For FLIM, the intrinsic relationship between intensity and lifetime information means noise in each channel is a multivariate function across channels without necessarily sharing structural features. Based on this, we present a novel Zero-Shot Denoising Framework with an Intensity-Guided Learning approach. Our correlation-preserving strategy maintains important biological information that might be lost when channels are processed independently. Our framework implements separate processing paths for each channel and utilizes a pre-trained intensity denoising prior to guide the refinement of lifetime components across multiple channels. Through experiments on real-world FLIM-acquired biological samples, we show that our approach outperforms existing methods in both noise reduction and lifetime preservation, thereby enabling more reliable extraction of physiological and molecular information.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。