Cell annotation method and annotation system using adaptive additional learning

摘要

A learning process is continuously performed when a new class of cell images arrives and / or when a new class of cell images is received, and the prediction capability for the new class of cell images is gradually expanded. Disclosed are methods, computer-readable media and systems for cell annotation. The method receives at least one new cell image, extracts cell features, predicts the closest class based on the extracted cell features, and detects cell pixels from the extracted cell features Generate likelihood maps, extract individual cells based on likelihood maps, and machine-annotate the extracted individual cells to identify cells, non-cell pixels and / or cell boundaries And calculating a reliability level of the machine annotation, and modifying the machine annotation if the reliability level is lower than a predetermined threshold. [Selection] Figure 1