Definiens Developer

Definiens is a commercial image segmentation and classification tool. The user designs a signal processing workflow by combining built-in filtering, thresholding and object classification modules. Object detection is typically done on hierarchical object levels, e.g cell level for cell objects and organelle level Nucleus and ER obejcts inside a cell object. For each object, a huge set of features (shape-based, intensity based, relations to neighbor objects...) is available and can be used for object classification or merging with neighboring objects. The classical definiens workflow is the so called bottom-up approach: In a first step, the image is segmented in numerous small objects, resulting in a heavy oversegmentation of of the target objects. Objects are then fused step by step on basis of features like “relative border to neighbor object” or “elliptic fit of resulting (fused) object”. Objects can assigned to different classes (like “nucleus” or “cancer cell”), based on their features. Weaknesses: -complex to use -closed (no API) -very expensive -relatively slow (you have to buy one license for each core) -bad 3D-visualization -time lapse analysis is possible but complicated Strengths: -powerful method to classify objects based on multiple features -2D data, especially histological data -good training material to learn software usage -detailed documentation