Pixel Super-Resolved Fluorescence Lifetime Imaging Using Deep Learning

2025-12-18

AI Summary

FLIM clinical adoption is hindered by long pixel dwell times and low signal-to-noise ratio, preventing simultaneous high resolution and imaging speed. To address this, we propose FLIM_PSR_k—a novel multi-channel conditional generative adversarial network (cGAN) architecture for pixel-level fluorescence lifetime image super-resolution reconstruction. Our method jointly integrates multi-channel pixel super-resolution (PSR) with physics-informed fluorescence lifetime modeling, ensuring robustness, real-time inference, and hardware compatibility. In blind testing on patient-derived tumor tissues, FLIM_PSR_k consistently achieves 5× spatial super-resolution (25× increase in space-bandwidth product) and faithfully recovers subcellular structural details. All quantitative image quality metrics show statistically significant improvements (p < 0.01). This work represents the first systematic integration of deep learning–driven PSR into FLIM, overcoming conventional optical diffraction and detection limits.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Fluorescence lifetime imaging microscopy (FLIM) is a powerful quantitative technique that provides metabolic and molecular contrast, offering strong translational potential for label-free, real-time diagnostics. However, its clinical adoption remains limited by long pixel dwell times and low signal-to-noise ratio (SNR), which impose a stricter resolution-speed trade-off than conventional optical imaging approaches. Here, we introduce FLIM_PSR_k, a deep learning-based multi-channel pixel super-resolution (PSR) framework that reconstructs high-resolution FLIM images from data acquired with up to a 5-fold increased pixel size. The model is trained using the conditional generative adversarial network (cGAN) framework, which, compared to diffusion model-based alternatives, delivers a more robust PSR reconstruction with substantially shorter inference times, a crucial advantage for practical deployment. FLIM_PSR_k not only enables faster image acquisition but can also alleviate SNR limitations in autofluorescence-based FLIM. Blind testing on held-out patient-derived tumor tissue samples demonstrates that FLIM_PSR_k reliably achieves a super-resolution factor of k = 5, resulting in a 25-fold increase in the space-bandwidth product of the output images and revealing fine architectural features lost in lower-resolution inputs, with statistically significant improvements across various image quality metrics. By increasing FLIM's effective spatial resolution, FLIM_PSR_k advances lifetime imaging toward faster, higher-resolution, and hardware-flexible implementations compatible with low-numerical-aperture and miniaturized platforms, better positioning FLIM for translational applications.

Problem

• Overcoming low resolution and slow speed in fluorescence lifetime imaging
• Addressing signal-to-noise limitations in autofluorescence-based FLIM applications
• Enabling high-resolution imaging with low-numerical-aperture and miniaturized platforms

Innovation

• Deep learning-based pixel super-resolution for FLIM
• cGAN framework enables robust reconstruction with fast inference
• Achieves 5x super-resolution factor enhancing image quality