A deep ensemble learning approach for squamous cell classification in cervical cancer.

Area of Science:

• Oncology
• Medical Imaging
• Computer Science

Background:

• Cervical cancer is a significant global health issue, with Pap smear analysis being a key diagnostic tool.
• Current Pap smear analysis is labor-intensive and time-consuming, particularly in resource-limited settings.
• There is a need for efficient, computer-aided methods for cervical cancer pre-analysis.

Purpose of the Study:

• To develop and evaluate an ensemble machine learning technique for classifying cervical squamous cells.
• To assess the accuracy of individual models (CNN, AlexNet, SqueezeNet) and the proposed ensemble method.
• To provide a more efficient diagnostic tool for cervical cancer, especially in resource-constrained environments.

Main Methods:

• Utilized a dataset of over 4096 cervical cell images from SimpakMed (Kaggle).
• Employed an ensemble technique integrating Convolutional Neural Network (CNN), AlexNet, and SqueezeNet for image classification.
• Classified squamous cells into five distinct categories to assess cervical cancer severity.

Main Results:

• Individual models achieved accuracies of 90.8% (CNN), 92% (AlexNet), and 91% (SqueezeNet).
• The proposed ensemble technique significantly outperformed individual models, reaching an accuracy of 94%.
• The ensemble approach demonstrated high efficacy in precise squamous cell classification.

Conclusions:

• The developed ensemble technique offers a highly accurate and efficient method for squamous cell classification.
• This approach shows significant promise for improving cervical cancer diagnosis in resource-limited settings.
• The study highlights the potential of advanced computational methods in enhancing cervical cancer screening programs.