Privacy Preserved Cervical Cancer Detection Using Convolutional Neural Networks Applied to Pap Smear Images

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Pathology

Background:

• Manual examination of Pap smear slides is time-consuming, error-prone, and costly.
• Accurate classification of cervical cells is crucial for early detection of cervical cancer.
• Image processing and machine learning offer potential for automated analysis.

Purpose of the Study:

• To develop and evaluate a convolutional neural network (CNN) for automated cervical cell classification.
• To distinguish between healthy, precancerous, and benign cervical cells using Pap smear images.
• To improve the efficiency and accuracy of cervical cancer screening.

Main Methods:

• Utilized the publicly available SIPaKMeD dataset containing five cervical cell categories.
• Applied image segmentation to Pap smear images.
• Developed a deep CNN with four convolutional layers for augmented cell image classification.

Main Results:

• The proposed CNN achieved an accuracy of 91.13% in classifying cervical cells.
• The model successfully distinguished between healthy, precancerous, and benign cervical cells.
• The CNN architecture is simple yet efficient for cervical cell classification.

Conclusions:

• The developed CNN model offers a faster and more accurate method for cervical cell classification.
• This approach can significantly reduce diagnostic time and computational costs.
• The model provides a foundation for future research to enhance cervical cancer prediction accuracy.