USE OF ARTIFICIAL INTELLIGENCE IN HEALTH DIAGNOSTICS - A VALIDATION STUDY ON CHORIONIC VILLI

  • Talat zehra Jinnah Sindh Medical University, Karachi Pakistan
  • Asma Shaikh Jinnah Sindh Medical University, Karachi Pakistan
  • Nighat Jamal Bahria University Medical & Dental College, Karachi Pakistan
  • Asma Shabbir Jinnah Sindh Medical University, Karachi Pakistan
  • Binish Arif Jinnah Sindh Medical University, Karachi Pakistan
  • Nausheen Ferozuddin Fazia Ruth Pfau Medical College, Karachi Pakistan

Abstract

Objective: The aim of the study was to validate chorionic villi identification by artificial intelligent software verses manual diagnosis.

Material and Methods: A Cross sectional study carried out at Yasmeen Syed Lab from 1st September 2020 to 30th October 2020 on samples(n=60) previously diagnosed cases of products of conception. Glass slides were scanned into digital slides using Bes Scope camera at 4x objective (0.1 um pixel). Digital slides were imported into Aiforia Oy software for all subsequent steps, including annotation, training, and analysis of digital slides. The minimum confidence level for diagnosing chorionic villi in software was 50%.

Results: Out of 60 cases, the software was able to diagnose 50 cases correctly. The concordance between glass slide and digital slide was 50 (83.33%). Once the software attained maturity, software showed green color chorionic villi present on digital slide after analyzing.

Conclusion: Artificial intelligent software is a novel technique, as it can overcome the scarcity of pathologists all over the world in general and in the developing world particularly. The results of this study were encouraging. We have certain limitations because of the unavailability of digital microscope or a pathology slide scanner.

Key Words: Digital pathology, Artificial intelligence, Chorionic villi.

Published
2021-12-31