SpheroidJ: An Open-Source Set of Tools for Spheroid Segmentation

Spheroids are the most widely used 3D models for studying effects of different micro-environmental characteristics on tumour behaviour, and testing preclinical clinical treatments. In order to speed up study spheroids, imaging methods that automatically segment measure spheroids instrumental; and, several approaches automatic segmentation spheroid images exist in literature. However, those fail generalise a diversity experimental conditions. The aim this work is development set tools works settings. work, we have tackled task by first developing generic algorithm can be easily adapted scenarios. This has been employed reduce burden annotating dataset that, turn, train deep learning architectures semantic segmentation. Both our constructed tested with datasets where were grown under conditions, acquired using equipment. developed particularised scenarios; however, particular algorithms By contrast, best model, HRNet-Seg architecture, generalises properly facilitate dissemination use models, present SpheroidJ, open-source an trained Thanks analysis conditions will more reliable comparable; help advance understanding behaviour.