A Deep Learning Model for Cervical Cancer Screening on Liquid-Based Cytology Specimens in Whole Slide Images

Abstract

Liquid-based cytology (LBC) for cervical cancer screening is now more common than the conventional smears, which when digitised from glass slides into whole-slide images (WSIs), opens up the possibility of artificial intelligence (AI)-based automated image analysis. Since conventional screening processes by cytoscreeners and cytopathologists using microscopes is limited in terms of human resources, it is important to develop new computational techniques that can automatically and rapidly diagnose a large amount of specimens without delay, which would be of great benefit for clinical laboratories and hospitals. The goal of this study was to investigate the use of a deep learning model for the classification of WSIs of LBC specimens into neoplastic and non-neoplastic. To do so, we used a dataset of 1605 cervical WSIs. We evaluated the model on three test sets with a combined total of 1468 WSIs, achieving ROC AUCs for WSI diagnosis in the range of 0.89–0.96, demonstrating the promising potential use of such models for aiding screening processes.

Sensor detection in gynaecological medicine

“…Liquid-based cytology (LBC) is an alternative technique and requires sample processing before microscopic analysis. LBC is more common than conventional Pap tests due to improved reliability; sample processing in LBC isolates cells of interest from interfering materials (e.g., mucous and blood), improving sample visualization and analysis [31,34]. As residual materials are available, LBC can also incorporate molecular tests such as the detection of the human papillomavirus (HPV) [34,35].…”

“…Pap tests have poor sensitivity and reproducibility as they are qualitative and rely on a cytologist's expertise [29,31,39]. To reduce bias, Kanavati et al developed a deep learning model for aiding cytologists in assessing sample images [34]. However, standalone HPV testing is recommended as the primary screening method for cervical cancer since HPV testing is objective, has greater sensitivity, with slightly lower specificity [31,39,43].…”

“…LBC is more common than conventional Pap tests due to improved reliability; sample processing in LBC isolates cells of interest from interfering materials (e.g., mucous and blood), improving sample visualization and analysis. 31,34 As residual materials are available, LBC can also incorporate molecular tests such as the detection of the human papillomavirus (HPV). 34,35 HPV testing is significant since infection with certain HPV strains are a cause 29,36,37 of about 90% of cervical cancers.…”

“…31,34 As residual materials are available, LBC can also incorporate molecular tests such as the detection of the human papillomavirus (HPV). 34,35 HPV testing is significant since infection with certain HPV strains are a cause 29,36,37 of about 90% of cervical cancers. 38 While a positive HPV result with a high-risk subtype is not a definitive diagnosis of cervical cancer, HPV-positive women can be further examined to determine HPV strain type, infection stage, and potential precancer/cancer presence.…”

“…29,31,39 To reduce bias, Kanavati et al developed a deep learning model for aiding cytologists in assessing sample images. 34 However, standalone HPV testing is recommended as the primary screening method for cervical cancer since HPV testing is objective, has greater sensitivity, with slightly lower specificity. 31,39,43 The specificity of HPV testing increases with age as younger women are more likely to have transient HPV infections, causing guidelines to suggest starting HPV testing for women above 30 or 35 years old.…”