Towards Automatic Segmentation: Preprocessing Histological Images with Filters and Contrast Enhancement

Chagas disease and toxoplasmosis are highly relevant to public health concerns. To investigate parasitic load and treatment efficacy, several studies have been conducted, with the quantification of intracellular parasites being a crucial step. Currently, this process is performed manually through microscopic observation, which limits the efficiency and standardization of the results. This work proposes the use of digital image processing techniques by applying filtering, contrast enhancement, and detection algorithms to improve the analysis of these images. In this study, 22 images were evaluated and processed using Python, applying Gaussian, Mean, and Bilateral filters, along with contrast enhancement techniques such as CLAHE and Unsharp Mask. The best combinations of filters and parameters were assessed using the Carneiro Contrast Index (CCI) and Neighborhood Valley Emphasis (NVE) metrics. The results indicated that CLAHE was more effective at enhancing contrast, while Unsharp Mask achieved better segmentability scores. It was concluded that a balance between CCI and NVE provides the most suitable approach to optimize image quality for subsequent segmentation.