1 School of Electrical Science, Department of Electronic and Communication Engineering, Bharath Institute of Higher Education and Research, Chennai, India
2 Department of Forensic Medicine and Toxicology, Saveetha Medical College and Hospital (SMCH), Saveetha Institute of Medical and Technical Sciences (SIMATS), Thandalam, Chennai-602105, Tamil Nadu, India. Email: priyavajjiravelu@gmail.com
3 Department of AI & ML, Faculty of Science & Humanities, SRM Institute of Science and Technology, Ramapuram, Chennai-600089, Tamil Nadu, India
4 Department of Mechatronics Engineering, Hindustan Institute of Technology and Science, Chennai-603103, Tamil Nadu, India. Email: janakir@hindustanuniv.ac.in
5 Department of Computer, Science and Engineering, Dhanalakshmi Srinivasan University, Chennai Campus, Mamandur, Chengalpattu-603111, Tamil Nadu, India. Email: premalathasaro@gmail.com
6 Marine Nanomedicine Laboratory Department of Research, Saveetha College of Nursing (SCON), Saveetha Institute of Medical and Technical Sciences (SIMATS), Thandalam, Chennai-602105, Tamil Nadu, India. Email: mythileeswari@gmail.com
7 Natural Biomedicine Laboratory, Department of Dermatology, Saveetha Medical College and Hospital (SMCH), Saveetha Institute of Medical and Technical Sciences (SIMATS), Thandalam, Chennai-602105, Tamil Nadu, India. Email: saktthivel@gmail.com
8* Department of ENT, Saveetha Medical College and Hospitals (SMCH), Saveetha Institute of Medical and Technical Sciences (SIMATS), Thandalam, Chennai-602105, Tamil Nadu, India. Email: shenbhagaraman@gmail.com (Corresponding Author)
Background:
Cervical cancer (CC) is one of the most common malignancies affecting women worldwide, and early detection plays a crucial role in reducing morbidity and mortality. Traditional screening methods such as Pap smear examination rely heavily on manual interpretation, which can be time-consuming and prone to inter-observer variability. Recent advances in artificial intelligence (AI) and deep learning (DL) have enabled automated analysis of medical images, improving diagnostic accuracy and screening efficiency.
Objective:
This study aims to develop a deep learning–based framework for the early detection and classification of cervical cancer using Pap smear and colposcopy images, and to evaluate the performance of a novel hybrid model for multi-class cervical intraepithelial neoplasia (CIN) prediction.
Methods:
A DL-based pipeline was designed incorporating image pre-processing, lesion segmentation, region of interest (ROI) extraction, and data augmentation. Baseline classification models including VGG16, ResNet50, and DenseNet121 were evaluated. A novel hybrid architecture, TriCervix, was proposed, integrating YOLOv8 for lesion detection, EfficientNet-B3 for feature extraction, and Bi-LSTM for sequential texture learning. The model was trained and tested to classify different stages of cervical intraepithelial neoplasia (CIN1 and CIN2).
Results:
The proposed TriCervix model demonstrated improved classification performance compared to conventional convolutional neural network (CNN) models such as VGG16, ResNet50, and DenseNet121. The hybrid architecture achieved higher accuracy, precision, and recall in detecting early-stage cervical lesions, particularly in distinguishing CIN1 and CIN2 stages.
Conclusion:
The TriCervix hybrid deep learning framework provides an effective and accurate approach for early-stage cervical cancer detection using medical imaging data. Its improved performance highlights its potential as a low-cost automated screening tool, particularly beneficial for large-scale screening programs in resource-limited healthcare settings.
Keywords: Cervical cancer, Deep learning, Pap smear imaging, Cervical intraepithelial neoplasia (CIN), Automated medical image classification
How to cite this article: Bhakthavachalam H, Vajjiravelu P, Kandasamy K, Srinivasan JR, Jayaraman P, Balaguru K, Lakshmikanthan M, Muthu S, Ramalingam S. TriCervix: Advanced Cervical Cancer Screening Using YOLO Detection and Sequential Texture-Based Hybrid Modeling. Int J Drug Deliv Technol. 2026;16(5): 102-109. DOI: 10.25258/ijddt.16.5.10
Source of support: Nil.
Conflict of interest: None