Label-Free Visualization and Segmentation of Endothelial Cell Mitochondria Using Holotomographic Microscopy and U-Net

Chemical & Biomedical Imaging, Journal Year: 2025, Volume and Issue: 3(4), P. 225 - 231

Published: Feb. 18, 2025

Understanding the physiological processes underlying cardiovascular disease (CVD) requires examination of endothelial cell (EC) mitochondrial networks, because function and adenosine triphosphate production are crucial in EC metabolism, consequently influence CVD progression. Although current biochemical assays immunofluorescence microscopy can reveal how influences cellular they cannot achieve live observation tracking changes networks through fusion fission events. Holotomographic (HTM) has emerged as a promising technique for real-time, label-free visualization ECs their organelles, such mitochondria. This nondestructive, noninterfering imaging method offers unprecedented opportunities to observe network dynamics. However, existing image processing tools based on techniques incompatible with HTM images, machine-learning model is required. Here, we developed using U-net learner Resnet18 encoder identify four classes within images: borders, ECs, background. accurately identifies structures positions. With high accuracy similarity metrics, output successfully provides images ECs. approach enables study effects, holds promise advancing understanding mechanisms.

Language: Английский

Label-Free Visualization and Segmentation of Endothelial Cell Mitochondria Using Holotomographic Microscopy and U-Net

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 2, 2024

Understanding the physiological processes underlying age-related cardiovascular disease (CVD) requires examination of endothelial cell (EC) mitochondrial networks, because function and adenosine triphosphate production are crucial in EC metabolism, consequently influence CVD progression. Although current biochemical assays immunofluorescence microscopy can reveal how influences cellular they cannot achieve live observation tracking changes networks through fusion fission events. Holotomographic (HTM) has emerged as a promising technique for real-time, label-free visualization ECs their organelles, such mitochondria. This non-destructive, non-interfering imaging method offers unprecedented opportunities to observe network dynamics. However, existing image processing tools based on techniques incompatible with HTM images, machine-learning model is required. Here, we developed using U-net learner Resnet18 encoder identify four classes within images: borders, ECs, background. accurately identifies structures positions. With high accuracy similarity metrics, output successfully provides images ECs. approach enables study effects, holds promise advancing understanding mechanisms.

Language: Английский