Confocal Laser Scanning Microscopy Image Analysis (MOONSHOT Goal 3)

AI-based analysis, segmentation, reconstruction, and quantitative assessment of vascular structures and thrombi. A topic of MOONSHOT Goal 3

Overview

This is a topic in MOONSHOT Goal 3 (Pioneering Science through the Creative Co-evolution of Humans and AI Robots) under the cooperation of Nagoya University and Tokyo University of Science.

This topic focuses on the analysis of confocal laser scanning microscopy images for quantitative investigation of vascular structures, thrombus formation, and related biological phenomena.

Computer vision and deep learning techniques are used to segment biological structures, reconstruct three-dimensional vascular networks, visualize thrombi, and extract quantitative measurements from microscopy images.

Research Topics

  • Confocal laser scanning microscopy image analysis
  • Blood vessel and thrombus segmentation
  • Neutrophil localization and measurement
  • Three-dimensional blood vessel reconstruction
  • Vessel tracking
  • Quantitative biological image analysis
  • Deep learning-based microscopic image segmentation
  • Transformer-based segmentation models

Research Period

2021–2025

Collaborations

  • Nagoya University
  • Tokyo University of Science

Representative Publications

  • Blood Vessel Segmentation from Low-Contrast and Wide-Field Optical Microscopic Images of Cranial Window by Attention-Gate-Based Network, CVPR Workshops, 2022.
  • Anatomical Structure-constrained Thrombus Region Segmentation and Measurement Using Confocal Laser Scanning Microscopic Images, SPIE Medical Imaging, 2024.
  • Preliminary Study on Neutrophil Localization and Measurement in Confocal Laser Scanning Microscope Images, CARS, 2024.
  • Prior Knowledge-based Blood Vessel Reconstruction and Thrombus Visualization Using Confocal Laser Scanning Microscopic Images, SPIE Medical Imaging, 2025.
  • Three-Dimensional Blood Vessel Structure Reconstruction from Confocal Laser Scanning Microscopy Images Based on Vessel Tracking, JAMIT, 2025.
  • Blood Vessel and Thrombus Segmentation from Images Using an Encoder-Decoder Model with Transformer, JSCAS, 2025.
  • Generation of Comparative Thrombosis Reports from Confocal Laser-Scanning Microscope Images Using Large Language Models, SPIE Medical Imaging, 2026.

Keywords

Confocal Microscopy, Bioimaging, Biological Image Analysis, Blood Vessel Segmentation, Thrombus Analysis, 3D Reconstruction, Deep Learning, Transformer