Skip to main content
Logo Institute of Oncology LjubljanaLogo Institute of Oncology Ljubljana
Donate
[email protected]
Donate
Donate
Back

Reliability of Digital Cervical Cytology for the Detection of High-Grade Squamous Intraepithelial Lesions and Cervical Cancer

 

The development of equipment for digitising cytology slides and software for image analysis has also enabled significant progress in the field of cervical liquid based cytology (LBC). The first automated digital cervical cytology (DCC) systems have been developed, allowing distinction between normal and pathologically altered cells in cervical smears. DCC involves capturing and digitising images and using artificial intelligence for their analysis. Despite this progress, the quality of digital images is not yet optimal, and an additional challenge is their size (1-10 GB or more), which results in high storage costs. There are also no clear recommendations regarding optimal scanning parameters, such as the number of Z-planes, the spacing between them, and the number of focal points, although image quality is crucial for reliable assessment. 

The principle of DCC is based on automatic slide analysis, where the software identifies pathological cells and creates a gallery of representative images, which are then evaluated by experts on high-resolution screens. Morphological criteria from the Bethesda classification, developed for microscopic cervical cytology (MCC), are currently used for interpretation. However, it is not known whether all of these criteria are also suitable for digitised slides, as image quality may affect their performance. 

The literature on DCC is limited; initial data indicate comparable performance to MCC, but studies on clinical sensitivity, specificity, longitudinal outcomes, and cumulative incidence for the detection of high-grade intraepithelial lesions (HSIL) and cervical cancer are lacking. 

The aim of the doctoral thesis is to determine the optimal conditions for LBC scanning using the NanoZoomer S360 scanner and AI software. The research will evaluate the relevance of morphological criteria, compare the concordance of MCC and DCC, assess the reliability of the methods in detecting HSIL and cervical cancer, and determine the cumulative inc'idence for APC-N+.