Computer vision for human stem cell derived cardiomyocyte classification: The induced pluripotent vs embryonic stem cell case study

Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) represent a potential valuable alternative to human embryonic stem cell-derived cardiomyocytes (hESC-CMs) for the lack of ethical issues and their patient-specificity. Recent studies showed hiPSC-CMs behave like hESC-CMs in terms of membrane currents, contractility and frequency of spontaneous pacing but their similarities need to be studied more. In this work we applied a well tested image processing approach based on texture feature extraction (local binary patterns, Haralick features and threshold adjacency statistics) and classification by support vector machines to hiPSC-CMs and hESC-CMs images from contrast phase microscopy. Our results suggested a good discrimination power of the chosen feature set meaning that some structural differences may still remain between the two cell classes.