AutoFRET: An Image Processing-Based ROI Automated Selection Method for Quantitative FRET Measurements

The emission-based fluorescence resonance energy transfer (E-FRET), renowned for its rapid detection, noninvasiveness towards fluorophores, and compatibility with both wide-field and confocal microscopy, is extensively employed in dynamically monitoring intermolecular interactions within living cells. However, E-FRET requires manual screening of hundreds to thousands of images for regions meeting specific criteria, a labor-intensive process devoid of mature automation solutions. In this article, we introduce AutoFRET, the automated and efficient solution tailored for E-FRET experimentation. AutoFRET harnesses image processing algorithms to swiftly and precisely identify target regions amidst vast image datasets. Furthermore, to mitigate the impact of dead cells in images on experimental results, we devise a novel cell morphology-based approach for their identification and exclusion. AutoFRET significantly reduces the time commitment for E-FRET experimental data analysis, condensing the entire process to the minute level. Comprehensive experimental evaluations reveal an average accuracy exceeding 95% for AutoFRET. This research presents a highly automated and reliable platform that expeditiously quantifies molecular interactions in living cells leveraging FRET technology, poised to contribute to advancements in quantitative biological research.