Cell Classification and Counting
Cognex Deep Learning separates cell types by their unique characteristics
Graphical programming environment for deep learning-based industrial image analysis
Cell classification and counting is a key task in clinical diagnosis (for example, in blood smears or mitotic counts). Many processes require an accurate initial cell count to standardize their baseline inputs and measure outcomes. Because the appearance of cells can differ, machine vision can sometimes struggle to accurately locate them against confusing backgrounds or artifacts. Those cells which are close together can also be difficult for machine vision to distinguish as independent.
The Cognex Deep Learning location tool accurately identifies cells by learning from annotated images of microscopic slides. The tool generalizes the normal appearance of cells based on their size, shape, and surface features as well as variable features. Self-learning algorithms learn to differentiate between intact and damaged cells (for example, those which carry a malaria virus), even on noisy backgrounds, and provide a reliable count. The classification tool learns to sort different types of cells, classifying them by unique characteristics such as size and shape.