Malaria is a highly lethal parasitic disease that affects millions of people every year, particularly in tropical regions. Early and accurate diagnosis is crucial for effective treatment, often relying on microscopic analysis of blood samples. However, obtaining fully focused images is a challenge due to the depth-of-field limitations of microscopes, resulting in blurred areas that may hinder parasite identification. In this context, multifocal image fusion emerges as a solution to obtain fully focused images, enabling automated diagnostic systems. This study aims to evaluate and compare different multifocal image fusion techniques applied to microscopic fields containing malaria parasites. The metrics used to assess the results were the Multicolor Edge Information Preservation and the Brenner Gradient, which measure the reliability of the fusion and the sharpness of the resulting images. Two fusion algorithms were implemented, following the parameters suggested in the literature, and applied to a dataset of thick blood smear images containing malaria parasites. In addition to analyzing the overall performance of the techniques, the study identified the parameters that showed the best results. The technique that demonstrated the best results was the one proposed by Piccinini et al. (2012), using the parameters Mean Filter = 11x11 and Majority Filter = 17x17, where the metric values were MEIP = 0.4631 and Brenner Gradient = 1,388.2. As a result, this research contributes to advancing image processing studies in the medical field through a comparative analysis of fusion techniques, as well as creating an extended-focus image dataset to support further research aimed at automating malaria diagnosis.
|
