Analysis of thin blood images for automated malaria diagnosis

The paper describes novel pre-processing techniques for the analysis of thin-film microscopic images of blood infected with the malaria parasite. An investigation to determine the optimum gray-scale conversion process was conducted and a tool to identify the optimum weights for a contrast enhanced gray-scale conversion was developed. The paper also introduces a new algorithm for morphological filtering of the blood images as a pre-processing tool for segmentation. Conventional morphological closing on blood images removes the unwanted components but also leads to loss of valuable information. The proposed morphological filtering preserves the necessary information of foreground components while removing noise and artefacts. This method could be modified for use in the pre-processing of other pathological images as well such as tissue analysis and cell differential analysis.