Towards a Complete Pipeline for Segmenting Nuclei in Feulgen-Stained Images

• DOI: 10.48550/arxiv.2002.08331
• Publication Date: 2020-01-01
• Authors: Luiz Antonio Busc...; Aldo von Wangenheim; Felipe Perozzo Da...; Alexandre Sherlle...; Fabiana Botelho d...; Marcelo Ricardo S...
• Keywords: Market Segmentation; Feulgen stain; Code (set theory)

Cervical cancer is the second most common type in women around world. In some countries, due to non-existent or inadequate screening, it often detected at late stages, making standard treatment options absent unaffordable. It a deadly disease that could benefit from early detection approaches. usually done by cytological exams which consist of visually inspecting nuclei searching for morphological alteration. Since humans, naturally, subjectivity introduced. Computational methods be used reduce this, where first stage process would segmentation. this context, we present complete pipeline segmentation Feulgen-stained images using Convolutional Neural Networks. Here show entire segmentation, since collection samples, passing through pre-processing, training network, post-processing and results evaluation. We achieved an overall IoU 0.78, showing affordability approach on images. The code available in: https://github.com/luizbuschetto/feulgen_nuclei_segmentation.