Multi-streams and multi-features for cell classification

With the development of deep learning technique, cell classification has gained increasing interests from the community. Identifying malignant cells in B-ALL white blood cancer microscopic images is challenging, since the normal and malignant cells have similar appearances. Traditional cell identification approach requires experienced pathologists to carefully read the cell images, which is laborious and suffers from inter-observer variations. Hence, the computer aid diagnosis systems for blood disorders, for example, leukemia, are worthwhile to develop. In this paper, we design a multi-stream model to classify the immature leukemic blasts and normal cells. We evaluated the proposed model on the C-NMC 2019 challenge dataset. The experimental results show that a promising result is achieved by our model.