Author(s):
1. Igor Hut, Serbia
2. Branislava Jeftić, Univeristy of Belgrade, Faculty of Mechanical Engineering, 3. Aleksandra Dragičević, Albania
4. Lidija Matija, Mašinski fakultet Univerziteta u Beogradu, Serbia
5. Djuro Koruga, Serbia

Abstract:
of the pathologist. Liquid-based cytology is proven to be more efficient than the conventional Papanicolaou test when it comes to sample preparation and the possibility of conducting several tests on the same sample. However, the specificity and sensitivity of the test are comparable with that of the Papanicolaou test, with false negative results still being the main drawback of these manually performed tests. Advances in technology and the availability of digital data have enabled the successful application of machine learning models in screening. Images of cervical cells are used as input to different deep learning models currently tested in studies concerning computer-aided diagnostic systems. This study explores different architectures of convolutional neural networks for cervical cancer detection based on Optomagnetic imaging spectroscopy and liquid-based cytology samples. The proposed VGG16-based model achieved 93.3% sensitivity and 67.8% specificity in the binary classification problem. Results highlight the need for a more balanced dataset in order for the suggested deep model to achieve better performance.

Key words:
cervical cancer,liquid-based cytology,convolutional neural network,optomagnetic imaging spectroscopy

Thematic field:
SYMPOSIUM B - Biomaterials and nanomedicine

Date of abstract submission:
01.08.2022.

Conference:
Contemporary Materials 2022 - Savremeni materijali

Abstract Final paper