Detection and diagnosis of cervical cancer in Pap smear cell images using hybrid CNN.

Saved in:

Bibliographic Details
Title:	Detection and diagnosis of cervical cancer in Pap smear cell images using hybrid CNN.
Authors:	Arulkarthick, E. K.¹ ekarulkarthick@gmail.com, Sukumar, P.²
Source:	Current Science (00113891). 8/10/2025, Vol. 129 Issue 3, p251-259. 9p.
Subjects:	Cervical cancer, Pap test, Detection algorithms, Data augmentation, Image recognition (Computer vision), Convolutional neural networks, Medical screening, Cell segmentation
Abstract:	Cervical cancer is screened in women patients using either Pap smear cell testing or the Cervigram analysis method. The most dominant accuracy has been obtained for the cervical cancer earlier detection system through the analysis of Pap smear cell images. In this article, they are automatically classified using the proposed hybrid convolutional neural networks (HCNN) structure. This classification system consists of enhancement, along with data augmentation and classification with the nucleus segmentation. The adaptive histogram equalisation enhancement algorithm enhances the image as a preprocessing method, and the imaging count is increased using the data augmentation method for obtaining a higher classification rate. The data-augmented images are further classified into four cases (normal, dysplasia, carcinoma in situ (CiS) and superficial) using the proposed hybrid CNN structure. Then, the dilation-erosion method was used to obtain the abnormal pixels in classified Pap smear cell images. Further, the morphological features are computed from the segmented nucleus region and are classified into either 'moderate' or 'severe' based on the computed features. The average diagnosis rate for dysplasia cell images is 90.4%. The average diagnosis rate for dysplasia cell images is 94.2%, and the average diagnosis rate for dysplasia cell images is 89.6%. From these extensive experimental results, the proposed methods are more suitable for a fully automated cervical cancer detection system. [ABSTRACT FROM AUTHOR]
	Copyright of Current Science (00113891) is the property of Indian Academy of Sciences and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	Cervical cancer is screened in women patients using either Pap smear cell testing or the Cervigram analysis method. The most dominant accuracy has been obtained for the cervical cancer earlier detection system through the analysis of Pap smear cell images. In this article, they are automatically classified using the proposed hybrid convolutional neural networks (HCNN) structure. This classification system consists of enhancement, along with data augmentation and classification with the nucleus segmentation. The adaptive histogram equalisation enhancement algorithm enhances the image as a preprocessing method, and the imaging count is increased using the data augmentation method for obtaining a higher classification rate. The data-augmented images are further classified into four cases (normal, dysplasia, carcinoma in situ (CiS) and superficial) using the proposed hybrid CNN structure. Then, the dilation-erosion method was used to obtain the abnormal pixels in classified Pap smear cell images. Further, the morphological features are computed from the segmented nucleus region and are classified into either 'moderate' or 'severe' based on the computed features. The average diagnosis rate for dysplasia cell images is 90.4%. The average diagnosis rate for dysplasia cell images is 94.2%, and the average diagnosis rate for dysplasia cell images is 89.6%. From these extensive experimental results, the proposed methods are more suitable for a fully automated cervical cancer detection system. [ABSTRACT FROM AUTHOR]
ISSN:	00113891
DOI:	10.18520/cs/v129/i3/251-259