A systematic review of effective data augmentation in cervical cancer detection

The rapid progress of AI has made computer-assisted systems essential in medical fields like cervical cytology analysis. Deep learning requires large datasets, but data scarcity and privacy concerns pose challenges. Data augmentation addresses this by generating additional images and improving model...

Full description

Saved in:
Bibliographic Details
Main Authors: Betelhem Zewdu Wubineh, Andrzej Rusiecki, Krzysztof Halawa
Format: Article
Language:English
Published: Polish Academy of Sciences 2025-06-01
Series:International Journal of Electronics and Telecommunications
Subjects:
Online Access:https://journals.pan.pl/Content/135235/5-4984-Wubineh-sk.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The rapid progress of AI has made computer-assisted systems essential in medical fields like cervical cytology analysis. Deep learning requires large datasets, but data scarcity and privacy concerns pose challenges. Data augmentation addresses this by generating additional images and improving model accuracy and generalizability. This review examines effective augmentation techniques and top-performing deep-learning models for segmentation and classification in cervical cancer detection. Analyzing 57 articles, we found that hybrid deep feature fusion with augmentation (rotation, flipping, shifting, brightness adjustments) achieved 99.8% accuracy in binary and 99.1% in multiclass classification. Augmentation is vital for enhancing model performance in limited data scenarios.
ISSN:2081-8491
2300-1933