Connor Lee, Harvy Chang, Hongyi Niu, Albert Li, Chengbiao Wu, Veronica Gomez-Godinez, Linda Shi
Abstract: Machine learning (ML) has been successfully applied across numerous disciplines to enhance efficiency and accuracy; however, most biological research laboratories continued to rely heavily on manual image processing approaches. While expert analysis was essential during the early experimental stages, the availability of sufficient datasets allowed ML algorithms to significantly accelerate image analysis and reduce human error. This motivation led to the development of a machine learning–based image processing pipeline for biological imaging applications. An ML framework integrating Cellpose-based segmentation with Python automation was implemented and applied to four major areas: protein recruitment following DNA double-strand breaks (DSBs), calcium flux tracking in cortical neurons after laser-induced shockwave (LIS) injury, calcium response characterization in retinal ganglion cells (RGCs), and comparative calcium dynamics in Alzheimer’s disease (AD) cellular models. Controlled DSBs were introduced into U2OS cells by a robotic laser microscope system (Robolase), with proteins of interest labeled using green fluorescent protein (GFP). Sequential fluorescence images were acquired and processed through custom Python code. The algorithm automatically detected protein recruitment by quantifying GFP intensity changes over time and generated kinetic plots within minutes. Minimal code modifications allowed the pipeline to be extended to calcium flux analyses in neurons, RGCs, and AD-affected cells. The automated pipelines considerably shortened the image processing time compared with manual methods, while maintaining high precision and scalability across biological applications. These findings demonstrated the potential of ML-enhanced imaging pipelines to advance studies of DNA damage repair, traumatic brain injury (TBI) modeling, and neurodegenerative disease.
Keywords: DNA Damage Repair, Laser Ablation, Laser-Induced Shockwave, Calcium Imaging, Retinal Ganglion Cells, Alzheimer’s Disease, Cellpose, Automated Analysis.
Date Published: October 30, 2025 DOI: 10.11159/jbeb.2025.003
View Article