Supervised Segmentation Model for Improved Detection of OSSN using Slit Lamp Images


Kajal Singh (Indian Institute of Technology Mandi), Shagun Bhatt (Indian Institute of Technology Mandi), Ramkailash Gujar (Dr Shroff’s Charity Eye Hospital ), Arnav Bhavsar (Indian Institute of Technology Mandi), Dinesh Singh (Indian Institute of Technology Mandi)
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)

Abstract

Ocular surface squamous neoplasia (OSSN) is a rare eye disease that affects the conjunctiva and the cornea and can cause blindness if not diagnosed early. At an early stage, the diagnosis of OSSN is often confused with other ocular surface diseases (OOSD) such as pterygium, pinguecula, episcleritis, papilloma, etc. Recently, AI algorithms have been explored for OSSN classification using OOSD and healthy patients as a control population. However, these algorithms either use traditional or deep models pre-trained on natural images or mild fine-tuning from some limited examples with image-level annotations, and finally perform poorly, and the results are unreliable. In this paper, we present a self-explainable deep neural network-based segmentation architecture to predict the pixel-level probability of OSSN in slit lamp images. The proposed method is more accurate and reliable than the existing methods as it eliminates the noisy variations in the images and focuses on the causal parts. The experiments in the collected data set for 163 OSSN, 200 OOSD, and 269 healthy people show the efficacy of the proposed methods in screening OSSN.

Citation

@inproceedings{Singh_2025_BMVC,
author    = {Kajal Singh and Shagun Bhatt and Ramkailash Gujar and Arnav Bhavsar and Dinesh Singh},
title     = {Supervised Segmentation Model for Improved Detection of OSSN using Slit Lamp Images},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_1095/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection