Segmentation of Hepatocytes Nuclei Using YOLO and Mathematical Morphology

In this work, we describe two segmenters that enable morphometry (extraction of geometric measures such as area and perimeter of cell nuclei, quantity of nuclei, among others) in preclinical microphotographs of rat liver tissue. The first segmenter was built using Mathematical Morphology and Digital...

Full description

Saved in:
Bibliographic Details
Published inInternational Conference on Systems, Signals, and Image Processing (Online) pp. 1 - 7
Main Authors Flores, Franklin C., Carvalho, Mateus F. T., Da Silva, Sergio A., Berton, Leandro E. F., Bernardo, Carla C. O., Sevilha, Andre L. R. G., Perles, Juliana V. C. M., Zanoni, Jacqueline N., Felipe, Gustavo Z., Costa, Yandre M. G.
Format Conference Proceeding
LanguageEnglish
Published IEEE 09.07.2024
Subjects
Alternating sequential filter
Area measurement
Digital images
Image segmentation
Liver
liver tissue
Mathematical morphology
Morphology
Rats
segmentation
YOLO
Online AccessGet full text

Cover

More Information
Summary:In this work, we describe two segmenters that enable morphometry (extraction of geometric measures such as area and perimeter of cell nuclei, quantity of nuclei, among others) in preclinical microphotographs of rat liver tissue. The first segmenter was built using Mathematical Morphology and Digital Image Processing to segment the nuclei, while YOLO was used to detect the hepatocytes in the image. The second segmenter was constructed using the Alternating Sequential Filter (ASF), which is based on the interleaving of morphological operators (i.e., opening and closing) by reconstruction to segment and detect the hepatocytes. After applying ASF, we use a minimum local function to binarize the image, highlighting the hepatocytes in white, while the background remains black. Experiments conducted on datasets developed by the Enteric Neural Plasticity Laboratory (ENPL) at the State University of Maringá (UEM) showed that both segmenters provide good results. The best result was achieved with the second segmenter, reaching an IoU of 88.23%. The first segmenter obtained an IoU of 86.36%. The results obtained demonstrate that it is possible to automatically segment cell hepatocytes in rat liver tissue, proving useful in supporting researchers in laboratories dealing with this type of image to perform analyses more quickly and accurately.
ISSN:2157-8702
DOI:10.1109/IWSSIP62407.2024.10634016