Classification of single-cell routine Pap smear images based on deep learning algorithms
Guanhao Zhang · Theoretical and Natural Science · 2024
In this study, we classified single-cell routine Pap smear images by applying deep learning algorithms such as AlexNet, VggNet, GoogleNet and MobileNet and compared their classification effects. The results show that the loss of all four models on both the training and test sets shows a trend of gradually decreasing and stabilising. Specifically, the loss of AlexNet gradually decreases from 0.637 to 0.212, VggNet from 0.777 to 0.278, GoogleNet from 1.77 to 0.31, and MobileNet from 0.809 to 0.267. At the same time, MobileNet exhibits the highest maximum and average accuracies which reached 93.9% and 88.3%, respectively, followed by GoogleNet model with 92.9% and 88.0%, AlexNet with 92% and 88.0%, and VggNet with 90.1% and 86.7%. The results show that MobileNet exhibits superior classification results in this task, which provides strong support for its potential application in the classification of single-cell routine Pap smear images. These findings are of great significance for further exploring the application of deep learning in the field of medical imaging and provide a useful reference for future related research.