Cervical Cancer Classification From Pap Smear Images Using Deep Convolutional Neural Network Models.

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Cervical cancer is a common female malignancy with a prolonged pre-cancerous stage.
• Traditional detection methods require complex cell segmentation and feature extraction.
• Deep learning models (CNNs) often need large datasets, posing challenges for rare disease detection.

Purpose of the Study:

• To develop deep learning models for automated cervical cancer detection without segmentation or custom features.
• To leverage transfer learning with pre-trained CNNs for Pap smear image analysis.
• To evaluate and compare the performance of 13 pre-trained deep CNN models.

Main Methods:

• Utilized transfer learning with 13 pre-trained deep convolutional neural network (CNN) models.
• Applied models directly to Pap smear images for a seven-class classification task.
• Employed the publicly available Herlev dataset and the Keras package in Google Collaboratory.

Main Results:

• DenseNet-201 demonstrated superior accuracy and performance among the evaluated models.
• Pre-trained CNN models yielded promising experimental results.
• The models required minimal computational time for analysis.

Conclusions:

• Deep learning, particularly DenseNet-201, offers an effective approach for automated cervical cancer detection from Pap smear images.
• Transfer learning mitigates data limitations, enabling robust classification.
• This method provides an efficient and accurate tool for early cervical cancer screening.