Novel generalized workflow for cell counting

Counting of cells present in a microscopy image is essential in many biological and pathological studies. The concentration of various cells present in a patient´s blood can provide critical information about the health status of the patient. In addition, it can also be used for the timely detection of parasitic diseases like malaria which is one of the major life threatening blood disease. In this paper we propose a new supervised learning workflow in Ilastik [9] for estimating the number of cells in a microscopic image. Ilastik [9] has an inbuilt workflow for counting objects based on density counting framework [1]. This workflow estimates directly the density of objects in an image and infers the number of objects by integrating over an object density map that is predicted from an input image and thus not requiring segmentation. However, the existing framework in Ilastik does not work well for large sized objects (for example malaria cells). The image has to be resized appropriately for density counting to work accurately. We provide a workflow which dynamically resizes the image by an appropriate resize factor and estimate cell count using Ilastik. Finally, we validate the workflow by considering numerous blood smear images and infer that cell count is estimated with high percentage accuracy (94% Accuracy).