Classification of Cervical Precancerous Cell of ThinPrep Images Based on Deep Learning Model AlexNet and InceptionV3

• Published in: Proceedings - 2022 IEEE 11th International Conference on Communication Systems and Network Technologies, CSNT 2022
• Main Author: 2-s2.0-85133141911
• Format: Conference paper
• Language: English
• Published: Institute of Electrical and Electronics Engineers Inc. 2022
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85133141911&doi=10.1109%2fCSNT54456.2022.9787658&partnerID=40&md5=dc90da12f1b6f53889929a7b0ccedfed

Cervical cancer is one of the deadliest diseases in the world, responsible for the greatest number of fatalities. Around 569.847 new cervical cancer cases are recorded every year. Efforts to prevent this condition can be conducted by early identification. There are several methods to detect cervical cancer, one of which is ThinPrep. In identifying cervical cancer, a neural network can be utilized as an alternative. AlexNet and InceptionV3 are neural network frequently applied to detect various diseases. In this study cervical cell images were classified based on cell severity, using deep learning models AlexNet and InceptionV3. The results it can be known that Inception V3 has a better performance based on the performance matrix analysis of the both models. The best performance matrix results for InceptionV3 are 89.80% for accuracy, 89.81% for precision, 91.17% for sensitivity, 94.49% for specificity, and 89.26% for F-score. However, AlexNet's training time have much faster than InceptionV3, with an average training time 57 seconds and fastest training time 55 seconds. © 2022 IEEE.