Clinical keratoconus (KCN) detection is a challenging and time-consuming task. In the diagnosis process, ophthalmologists must revise demographic and clinical ophthalmic examinations. The latter include slit-lamb, corneal topographic maps, and Pentacam indices (PI). We propose an Ensemble of Deep Transfer Learning (EDTL) based on corneal topographic maps. We consider four pretrained networks, SqueezeNet (SqN), AlexNet (AN), ShuffleNet (SfN), and MobileNet-v2 (MN), and fine-tune them on a dataset of KCN and normal cases, each including four topographic maps. We also consider a PI classifier. Then, our EDTL method combines the output probabilities of each of the five classifiers to obtain a decision based on the fusion of probabilities. Individually, the classifier based on PI achieved 93.1% accuracy, whereas the deep classifiers reached classification accuracies over 90% only in isolated cases. Overall, the average accuracy of the deep networks over the four corneal maps ranged from 86% (SfN) to 89.9% (AN). The classifier ensemble increased the accuracy of the deep classifiers based on corneal maps to values ranging (92.2% to 93.1%) for SqN and (93.1% to 94.8%) for AN. Including in the ensemble-specific combinations of corneal maps’ classifiers and PI increased the accuracy to 98.3%. Moreover, visualization of first learner filters in the networks and Grad-CAMs confirmed that the networks had learned relevant clinical features. This study shows the potential of creating ensembles of deep classifiers fine-tuned with a transfer learning strategy as it resulted in an improved accuracy while showing learnable filters and Grad-CAMs that agree with clinical knowledge. This is a step further towards the potential clinical deployment of an improved computer-assisted diagnosis system for KCN detection to help ophthalmologists to confirm the clinical decision and to perform fast and accurate KCN treatment.

<span lang="EN-US">Light penetrates the human eye through the cornea, which is the outer part of the eye, and then the cornea directs it to the pupil to determine the amount of light that reaches the lens of the eye. Accordingly, the human cornea must not be exposed to any damage or disease that may lead to human vision disturbances. Such damages can be revealed by topographic images used by ophthalmologists. Consequently, an important priority is the early and accurate diagnosis of diseases that may affect corneal integrity through the use of machine learning algorithms, particularly, use of local feature extractions for the image. Accordingly, we suggest a new algorithm called local information pattern (LIP) descriptor to overcome the lack of local binary patterns that loss of information from the image and solve the problem of image rotation. The LIP based on utilizing the sub-image center intensity for estimating neighbors' weights that can use to calculate what so-called contrast based centre (CBC). On the other hand, calculating local pattern (LP) for each block image, to distinguish between two sub-images having the same CBC. LP is the sum of transitions of neighbors' weights, from sub-image center value to one and vice versa. Finally, creating histograms for both CBC and LP, then blending them to represent a robust local feature vector. Which can use for diagnosing, detecting.</span>

Detection of early clinical keratoconus (KCN) is a challenging task, even for expert clinicians. In this study, we propose a deep learning (DL) model to address this challenge. We first used Xception and InceptionResNetV2 DL architectures to extract features from three different corneal maps collected from 1371 eyes examined in an eye clinic in Egypt. We then fused features using Xception and InceptionResNetV2 to detect subclinical forms of KCN more accurately and robustly. We obtained an area under the receiver operating characteristic curves (AUC) of 0.99 and an accuracy range of 97–100% to distinguish normal eyes from eyes with subclinical and established KCN. We further validated the model based on an independent dataset with 213 eyes examined in Iraq and obtained AUCs of 0.91–0.92 and an accuracy range of 88–92%. The proposed model is a step toward improving the detection of clinical and subclinical forms of KCN.

No abstract

Machine learning techniques become more related to medical researches by using medical images as a dataset. It is categorized and analyzed for ultimate effectiveness in diagnosis or decision-making for diseases. Machine learning techniques have been exploited in numerous researches related to corneal diseases, contribution to ophthalmologists for diagnosing the diseases and comprehending the way automated learning techniques act. Nevertheless, confusion still exists in the type of data used, whether it is images, data extracted from images or clinical data, the course reliant on the type of device for obtaining them. In this study, the researches that used machine learning were reviewed and classified in terms of the kind of utilized machine for capturing data, along with the latest updates in sophisticated approaches for corneal disease diagnostic techniques.