National Institute of Biomedical Imaging and Bioengineering 5P41EB035084-02

TR&D3

This NIH-funded project led by University of Michigan focuses on advancing live fluorescence imaging through machine learning to improve image quality and speed, enabling better in vivo cellular analysis and digital pathology, ultimately enhancing biomedical research and reducing animal use.

Description

The project aims to develop machine learning models to denoise live fluorescence images captured at higher frame rates to mitigate motion artifacts critical for in vivo analysis. It addresses challenges in microendoscopy motion sensitivity and biomedical image segmentation at the cellular level, leveraging deep learning for feature extraction and real-time digital readouts correlating in vivo cellular activity with molecular properties via postmortem tissue imaging. The project integrates in vivo optical sections with machine learning for digital pathology and tumor assessment using photoacoustic imaging, enhancing biomedical imaging techniques to better understand cellular functions and drug effects, reducing animal use and increasing research confidence.

View original record