Classification and Determination of Severity of Corneal Ulcer with Vision Transformer Based on the Analysis of Public Image Dataset of Fluorescein-Stained Corneas

Talha Burak Alakuş,Muhammet Baykara

DOI: https://doi.org/10.3390/diagnostics14080786

IF: 3.6

2024-04-10

Diagnostics

Abstract:A corneal ulcer is a condition in which an injury to the corneal surface occurs as a result of infection. This can lead to severe vision loss and even blindness. For this reason, early diagnosis of this disease is of great importance. Deep learning algorithms are used in many critical health applications and are used effectively in the early diagnosis stages of diseases. Thus, a deep learning algorithm was applied in this study and corneal ulcer and severity were predicted. The study consisted of four stages over three different scenarios. In the first scenario, the types of corneal ulcers were predicted. In the second scenario, the grades of corneal ulcer types were classified. In the last scenario, the severity of corneal ulcers was classified. For each scenario, data were obtained in the first stage and separated according to the relevant labels. In the second stage, various image processing algorithms were employed, and images were analyzed. At this stage, the images were also augmented by various processes. In the third stage, ViT architecture, a new deep learning model, was used, and the images were classified. In the last stage, the performance of the classifier was determined by accuracy, precision, recall, F1-score, and AUC score. At the end of the study, the ViT deep learning model performed an effective classification, and accuracy scores of 95.77% for the first scenario, 96.43% for the second scenario, and 97.27% for the third scenario were calculated.

medicine, general & internal

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to analyze the public data set of fluorescein - stained corneal images by using the deep - learning algorithm based on Vision Transformer (ViT) to achieve the classification of corneal ulcers and the determination of their severity. Specifically, the research is divided into three different scenarios:

1. **Scenario 1**: Predict the type of corneal ulcer. The goal of this part is to distinguish different types of corneal ulcers, such as punctate corneal ulcer, mixed punctate - patchy corneal ulcer and patchy corneal ulcer.
2. **Scenario 2**: Classify the grades of corneal ulcer types. The goal here is to classify different grades of corneal ulcers according to the TG (Type Grade) rating method, which includes no ulcer, tiny punctate ulcer, large punctate ulcer, confluent large punctate ulcer and ulcer plaque exceeding 1 millimeter.
3. **Scenario 3**: Determine the severity of corneal ulcer. The goal of this part is to evaluate the proportion of the cornea affected by the ulcer and divide it into five grades, from no ulcer to ulcer with the central area affected.

Through these classifications, the research aims to improve the accuracy of early diagnosis of corneal ulcers, so as to take timely treatment measures and prevent the condition from deteriorating and leading to vision loss or blindness. In addition, the research also hopes to show the application potential of deep - learning technology in medical image analysis by introducing the ViT model, especially when dealing with complex and diverse corneal ulcer images.