High-Performance Detection of Corneal Ulceration Using Image Classification with Convolutional Neural Networks

Abstract

Corneal Ulcer, also known as keratitis, represents the most frequently appearing symptom among corneal diseases, the second leading cause of ocular morbidity worldwide. Consequences such as irreversible eyesight damage or blindness require an innovative approach that enables a distinction to be made between patterns of different ulcer stages to lower the global burden of visual disability. This paper describes a Convolutional Neural Network-based image classification approach that allows the identification of different types of Corneal Ulcers based on fluorescein staining images. With a balanced accuracy of 92.73 percent, our results set a benchmark in distinguishing between general ulcer patterns. Our proposed method is robust against light reflections and allows automated extraction of meaningful features, manifesting a strong practical and theoretical relevance. By identifying Corneal Ulcers at an early stage, we aid reduction of aggravation by preventively applying and consequently tracking the efficacy of adapted medical treatment, which contributes to IT-based healthcare.

Automated Defect Detection of Screws in the Manufacturing Industry Using Convolutional Neural Networks

“…The current studies achieve an accuracy from the range of 79.40% to 99.00% for defect detection. However, those approaches rely on additional imaging techniques or simply focus on the detection of only single defect types [19]. If different defects are considered, several models were needed to detect those with a high level of accuracy [23].…”

“…Sensors and the Internet of Things cause an increase in data recorded in the manufacturing process. Machine learning approaches have proven to be an effective tool for the evaluation of different sensor data [18,19]. Especially Deep Learning shows superior performance for the analysis of image data [20].…”

Using Slit-Lamp Images for Deep Learning-Based Identification of Bacterial and Fungal Keratitis: Model Development and Validation with Different Convolutional Neural Networks

“…According to their findings, the wrong prediction of the model was due to incorrect focusing on the eyelid, eyelash, and sclera [ 10 ]. The use of fluorescein-staining images was reported in only one study; however, the researchers aimed to identify early corneal ulcers by recognizing point-like patterns, which could not differentiate fungal ulcers that were difficult to manage from other corneal ulcers [ 23 ]. Xu et al reported diagnostic rates of 80% for MK, 53.3% for BK, and 83.3% for FK by using a deep sequential-level learning model with slit-beam slit-lamp images.…”

Benchmarking Transfer Learning Strategies in Time-Series Imaging: Recommendations for Analyzing Raw Sensor Data

“…DIM3 The last dimension of our benchmarking study includes the evaluation of different TL strategies. TL has already established itself as successful in various image analysis-related applications [33] and is already being used in several time-series imagingrelated tasks [38]. Thus, we are implementing and investigating a variety of TL pipelines.…”

“…As the sample size strongly influences the performance of CNNs, Transfer Learning (TL) can be used complementary to overcome the sampling challenge. TL greatly reduces the amount of required training data by utilizing knowledge acquired for a related domain to solve a task for another, improving the model's performance and robustness [33], [34]. Although the current time-series imaging literature lacks TL-based approaches for optimized model performance, frameworks from different domains (e.g., healthcare [35], engineering [31]) allow subclassifying the CNN-based classification of visualized time-series data into three main steps across domains [30], [31], [35], [36]: 1) Data foundation (i.e., signal acquisition and data preprocessing) 2) Imaging (i.e., encoding time-series as images) 3) Evaluation (i.e., DL model for label classification) Companies can use these three influential dimensions (see: figure 1) to gain competitive advantages over their competitors by optimizing their approaches for data analysis [29].…”