Abstract:

Biomedical imaging is crucial for elucidating the mechanisms underlying physiological and pathological processes; however, manual analysis of these images is often time-consuming, labor-intensive, and prone to human error and subjectivity. Although advanced AI-based imaging models, such as You-Only-Look-Once (YOLO), have demonstrated exceptional performance in object detection, segmentation, and classification for natural-scene images, their off-the-shelf performance deteriorates markedly when applied to in vitro and ex vivo biomedical images and videos, which often feature lower resolution, high-level noise, and contrast variations. To overcome these challenges, we develop an integrated pipeline that combines YOLO with advanced imaging-classification models, such as Vision Transformers (ViT), to substantially improve the accuracy and robustness of cell and protein detection, tracking, and classification in both in vitro and ex vivo imaging datasets. Our results show that this framework reliably tracks red blood cell (RBC) deformation in sickle cell disease during dynamic oxygenation–deoxygenation cycles and quantifies the motion of proteins and cells across sequential image frames. The resulting real-time velocity estimates enable detailed assessments of cell deformability and motility under physiologically relevant conditions. Altogether, this YOLO-based framework efficiently analyzes sequential cellular and protein from multi-modal images. While initially validated on hematologic images, the approach is readily extensible to diverse molecular and cellular imaging analysis, offering a versatile tool for translational research across a range of diseases.

Bio:

Dr. He Li developed multiscale computational models based on physics laws using various numerical methods, such as molecular dynamics, dissipative particle dynamics and spectral element method, to simulate biological systems that span multiple spatial

scales, including molecular level, protein level, sub-cellular level, cellular level, multi-cell systems, vasculature and organ systems. His work has demonstrated that computational modeling can bridge the gap between the microscopic and macroscopic physiological processes, and provide innovative approaches to study key problems in biology, medicine and biomedical engineering, such as building mechanistic models to investigate the pathogenesis of human diseases and developing predictive models to examine the existing hypotheses and derive new hypotheses to steer experimental and computational studies. Dr. He Li’s current research interest is to employ AI techniques to develop advanced multiscale models and build predictive AI models that can assimilate data from different sources (e.g., biophysical, biochemical, genomics, proteomics data), to improve digital health technologies.