NTU Undergraduates' research April 2014 - Biosciences | Seite 8

An assessment of the use of image analysis as a way of helping distinguish
grading of cervical smears especially from the mild, moderate and severe
category using images from traditional conventional PAP smears
H. Plaha.,
Nottingham Trent University
15/04/2014
An assessment of the use of image analysis as a way of helping distinguish grading of cervical smears especially from the
mild, moderate and severe category using images from traditional conventional PAP smears

Abstract
Objective: This study used image analysis to assess the nuclear cytoplasmic (NC) ratio for a range of
samples taken from a pathological database subjectively graded as mild dyskaryosis, moderate
dyskaryosis and severe dyskaryosis.
Methods: Quantitative image analysis using Image J was conducted on 50 samples of each cell from
each grading category of normal superficial, mild, moderate and severe dyskaryotic cells, so in total
200 cell samples were measured. Statistical analysis is conducted on the results from the image
analysis, in an attempt to distinguish the categories of grading between the Bethesda system; a twotier model or the British Society for Clinical Cytology; a three-tier model but newly proposed two-tier
model.
Results: analysis of the area NC ratios as currently stands in the BSCC system indicates that there are
three distinct populations; mild dyskaryosis, moderate dyskaryosis and severe dyskaryosis. The
graphs produced even with error bars seem to indicate there are three different results and there
doesn’t seem to appear to be any overlap. As the BSCC are proposing a new two-tier system of lowgrade and high-grade dyskaryo