Towards Digital Diagnosis of Oral Cancer: A Study on Optimum Preferences of Histopathological Techniques and Features

Abstract

Accurate diagnosis is dependent on various factors in the pathological domain, like types of slides used and

features scrutinized. Sometimes a diagnosis is evident due to clear symptoms. But under adverse constraints,

like improper acquisition etc., it is very difficult to give a quick and clear diagnosis. The study aims to

conduct a survey from the well-known histopathologists of the country to gather an understanding of the

techniques preferably used by them for diagnosis of the disease and summarize it, for arriving at optimum

options which may be adopted for automation. This was a cross-sectional study conducted from March 2018

to May 2018 using a pre-tested structured questionnaire of multiple answer choices. The study subjects

comprised of resident histopathologists of the hospitals covering states/locations all over the country, by

selective sampling. The hospitals were selected based on the availability of on-site pathology centres and

wide coverage. The analysis of the data was done using Ms-Excel and SPSS. The Non-Parametric Friedman

Test was conducted to test for significance of the responses. Oral Squamous Cell Carcinoma (OSCC) of

buccal mucosa with both moderately and well-differentiated grades were reported. For diagnosis, H&E

stain for the slides of 4? thickness is mostly used. Further, invasion of basement membrane was the most

important architectural feature and increased nucleo-cytoplasmic ratio the most important cytological

feature. This type of survey will help in carrying out a directed diagnosis or further research for automated

diagnosis using the results.