Automated Evaluation of Cell Viability in Microfluidic Spheroid Arrays

Three-dimensional (3D) spheroid arrays promise improved predictability due to their higher physiological relevance. They have the potential to improve drug screening outcomes in preclinical studies. Despite the advances, they can often lead to non-reproducible and unpredictable results. To support the development and subsequent analyses of spheroid arrays, we present a method for analyzing and evaluating cell viability in these. We provide a fast and easy-to-use fully automated workflow for the viability analysis in fluorescence images of cell aggregates within these arrays. The algorithm consists of multiple image processing algorithms for the segmentation and mapping of a priori knowledge about the array layout. The segmentation step is based on Otsu's thresholding followed by morphological filtering to obliterate the necessity of input parameters. No preprocessing is required. Besides, the algorithm offers the possibility of applying an additional flood fill algorithm. Subsequently, a k-means algorithm allows a fast image independent mapping of the grid to identify the cell aggregates. The complete workflow allows the extraction of essential metrics describing the viability of each cell aggregate. With our automated approach, we can show an increase in accuracy compared to simple manual segmentation. Additionally, the objectivity is increased by reducing human intervention. Furthermore, the needed analysis time is shortened and the information extraction and evaluation process is simplified.