Segmentation Techniques Applied to CNNs for Cervical Cancer Classification

Segmentation Techniques Applied to CNNs for Cervical Cancer Classification

Cervical cancer continues to be a significant global health issue, ranking as the fourth most prevalent cancer affecting women. Enhancing population screening programs by refining the examination of cervical samples conducted by skilled pathologists offers a compelling alternative for early detectio...

Full description

Saved in:
Bibliographic Details
Main Authors: Ana Ortiz-Gonzalez, Raquel Martinez-Espana, Juan Morales-Garcia, Baldomero Imbernon, Jose Martinez-Mas, Mauricio A. Alvarez, Oscar David Romero, Juan Pedro Martinez-Cendan, Andres Bueno-Crespo
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Artificial intelligence
deep learning
artificial vision
supervised classification
convolutional neural networks
segmentation
Online Access:https://ieeexplore.ieee.org/document/10971179/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Cervical cancer continues to be a significant global health issue, ranking as the fourth most prevalent cancer affecting women. Enhancing population screening programs by refining the examination of cervical samples conducted by skilled pathologists offers a compelling alternative for early detection of this disease. Deep Learning facilitates the development of automatic classification models to aid experts in this task. However, it is increasingly important to bring explainability to the model both to understand how the network learns to identify pathology and to bring confidence to the diagnosis. In this paper, we design an automatic segmentation masks for the classification of cervicovaginal cell images. This automatic segmentation is combined in a classification model that allows the models to improve their performance thanks to the morphological information provided by the combined segmentation in a Global Average Pooling layer with the convolutional network analysis of the original image. The models will be trained with real data so that learning can recognize the diversity of colors, shapes and sizes of human cell nuclei. The results show a robust and explainable model with satisfactory results, obtaining an F1 Score value of 0.935 in binary classification of revisable and non-revisable cell.
ISSN:2169-3536

Similar Items