Scalable and lightweight deep learning for efficient high accuracy single-molecule localization microscopy.

Journal: Nature communications
Published Date: Aug 5, 2025

Abstract

Deep learning has significantly improved the performance of single-molecule localization microscopy (SMLM), but many existing methods remain computationally intensive, limiting their applicability in high-throughput settings. To address these challenges, we present LiteLoc, a scalable analysis framework for high-throughput SMLM data analysis. LiteLoc employs a lightweight neural network architecture and integrates parallel processing across central processing unit (CPU) and graphics processing unit (GPU) resources to reduce latency and energy consumption without sacrificing localization accuracy. LiteLoc demonstrates substantial gains in processing speed and resource efficiency, making it an effective and scalable tool for routine SMLM workflows in biological research.

Authors

  • Yue Fei
    Department of Biomedical Engineering, Southern University of Science and Technology, Shenzhen, China.
  • Shuang Fu
    Guangdong Provincial Key Laboratory of Advanced Biomaterials, Department of Biomedical Engineering, Southern University of Science and Technology, Shenzhen, China.
  • Wei Shi
    Department of Orthopedics, Shenzhen Pediatrics Institute of Shantou University Medical College, Shenzhen, China.
  • Ke Fang
    College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou 310020, China.
  • Ruixiong Wang
    Department of Biomedical Engineering, Southern University of Science and Technology, Shenzhen, China.
  • Tianlun Zhang
    School of Electrical and Electronic Engineering, Hubei University of Technology, Wuhan, China.
  • Yiming Li
    Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China.

Keywords

No keywords available for this article.

External Resources

Popular Topics
Recent Journals