Investigation of the Effect of the Number of Clients in Federated Learning on Cervical Cancer Classification Using Pap-Smear Images

The classification of cervical cell images provides important information for diagnosing malignant or precancerous lesions and can assist in making an accurate diagnosis. Current methods generally require collecting all patients' Pap-Smear test images in a central location. However, this method jeopardizes the privacy of patient data. In this study, different classification algorithms and federated learning architectures were used to achieve accurate classification and protect data confidentiality. Firstly the accuracy parameters of the data were examined with different classification algorithms. Then, the success of these results was evaluated, federated learning architectures were added, and the architectures were compared. In this comparison, the relationship between the number of clients and the accuracy value was specifically examined.