A novel cervical image recognition framework based on feature cognitive enhancement for improved performance

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Cervical cancer is a leading global cancer in women.
• Early diagnosis is crucial for effective treatment.
• Current diagnostic methods face challenges due to increasing patient numbers and doctor workload, potentially leading to misdiagnosis.

Purpose of the Study:

• To develop a novel deep learning model for accurate classification of cervical images.
• To address limitations in feature extraction of existing neural network models for cervical image recognition.

Main Methods:

• Proposed a novel model named FSMO (Feature Cognitive Screening Module, Multi-scale Feature Classification Module, and Overlapping Sampling Module).
• FSMO integrates global and local feature extraction, multi-scale feature fusion, and enhanced edge capturing capabilities.
• The model was evaluated on a self-constructed dataset and a Kaggle dataset.

Main Results:

• FSMO achieved 91.88% accuracy, 92.91% precision, 91.92% recall, and 91.99% F1-Score on the self-constructed dataset.
• Achieved 97.5% accuracy on the Kaggle dataset.
• Outperformed other advanced models in cervical image recognition.

Conclusions:

• The FSMO model demonstrates significant potential for rapid auxiliary diagnosis in cervical imaging.
• This AI-driven approach can contribute to the early detection and treatment of cervical cancer.
• The model's enhanced feature extraction capabilities improve prediction accuracy for cervical disease classification.