Presentation Information
[MS11-02]In situ stemness assessment of human pluripotent stem cells through deep-learning-assisted holotomography
*Ki-Jun Yoon1, Hoewon Park1, Geon Kim1, Yongkeun Park1 (1. KAIST (Korea))
Keywords:
Pluripotent stem cell,Holotomography,Artificial Intelligence,quality control,non-labeled imaging
Human pluripotent stem cells (hPSCs) play a pivotal role in regenerative medicine, offering unparalleled opportunities for studying human biology and developing therapies for various diseases. However, hPSCs are highly sensitive to culture conditions, resulting in variability that poses challenges for their effective use in clinical applications. Therefore, quality control of hPSCs is essential to ensure the consistency and safety in downstream applications like cell transplantation therapies. Traditional methods, such as immunostaining and transcriptomic profiling, are invasive, population-level seneitive, and labor-intensive. Conversely, current non-invasive evaluation methods that depend on qualitative morphological criteria are susceptible to subjective bias and inconsistency, particularly when selecting stable human induced pluripotent stem cells (hiPSCs) during reprogramming. Here, we introduce PLUTO (PLUripotency and stemness evaluation model by holoTOmography), a novel method for assessing hPSC pluripotency using holotomography (HT), a label-free imaging system, integrated with artificial intelligence (AI). By refleting the 3D colony structure during early differentiation, PLUTO was constructed to facilitate real-time, non-invasive evaluation of pluripotency. This system reveals previously undetectable cellular and colonial information, providing a robust, unbiased approach for stem cell quality control. PLUTO accurately classified early retinoic acid-induced differentiation and predicted pluripotency levels in newly reprogrammed hiPSCs. This innovative method offers significant potential for improving stem cell culture consistency and clinical applications, such as personalized cell therapy.