Automatic detection of keratoconus on Pentacam images using feature selection based on deep learning

Fırat, Murat; Çankaya, Cem; Çınar, Ahmet; Tuncer, Taner

doi:10.1002/ima.22717

Cited by 10 publications

(7 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A comparison of our model with other models previously published is shown in Table 5. It should be noted that suspect KCN was not included in many previous studies [2,13,15,19,22]. To tackle that, we developed and validated our proposed method on the three-class problem, including suspect KCN.…”

Section: Discussionmentioning

confidence: 99%

“…In this study, our aim was to improve both the accuracy and robustness of detecting subclinical forms of KCN, based on an ensemble of DL features, which is an emerging approach in which the features are extracted from a pretrained DL network and fed into a machine learning classifier, such as SVM, to make the final classification. This approach has previously been applied to detect two corneal classes (normal and KCN) with four corneal topographic maps and Alexnet with SVM [22], and has also been examined in three-class detection (normal, KCN, and suspect KCN) with seven corneal topographic maps and EfficientNet-b0 architecture and SVM [23].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Deep Feature Fusion of Improved Suspected Keratoconus Detection with Deep Learning

et al. 2023

View full text Add to dashboard Cite

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.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Deep Feature Fusion of Improved Suspected Keratoconus Detection with Deep Learning

et al. 2023

View full text Add to dashboard Cite

show abstract

“…1) Images: In most of contributions, using deep learning architectures, authors have used image type data for the classification of keratoconus. Processed images are in different types, such as corneal topography [12], tomography [25], and Placido disc [37].…”

Section: Rq3 What Are the Data Types Used By The Different Classifier...mentioning

confidence: 99%

“…Table IV indicates the distribution of studies by the number of corneal classes considered in keratoconus classification. 2 classes [24], [37], [28], [35], [10], [12], [14], [15], [18], [20], [22], [23], [34], [39], [46] 15…”

Section: F Rq6 What Is the Number Of Keratoconus Classesmentioning

confidence: 99%

Machine Learning Techniques in Keratoconus Classification: A Systematic Review

Aatila¹,

Lachgar²,

Hrimech³

et al. 2023

IJACSA

View full text Add to dashboard Cite

Machine learning (ML) algorithms are being integrated into several disciplines. Ophthalmology is one field of health sector that has benefited from the advantages and capacities of ML in processing of different types of data. In a large number of studies, the detection and classification of various diseases, such as keratoconus, was carried out by analyzing corneal characteristics, in different data types (images, measurements, etc.), using ML tools. The main objective of this study was to conduct a rigorous systematic review of the use of ML techniques in the detection and classification of keratoconus. Papers considered in this study were selected carefully from Scopus and Web of Science digital databases, according to their content and to the adoption of ML methods in the classification of keratoconus. The selected studies were reviewed to identify different ML techniques implemented and the data types handled in the diagnosis of keratoconus. A total of 38 articles, published between 2005 and 2022, were retained for review and discussion of their content.

show abstract

“…ResNet18, ResNet50, and ResNet101 were used as feature extractors in this paper. The most different one among the optimizations and innovations made in Deep Networks is the ResNet architecture, where 'residual' connections are made [26]. Table 2 shows the architecture of the ResNet models.…”

Section: Positionmentioning

confidence: 99%

Determining the Type of Document Read Using Eye Movement Properties by Hybrid CNN Method

Çetintaş¹,

Tuncer²

2022

Self Cite

View full text Add to dashboard Cite

By examining the change of eye movements during reading, it is possible to determine the type of document read. Returning to the previous position (negative saccade), blink, fixation, and position are important indicators in determining the type of document being read. In this paper, a hybrid deep learning model is proposed to determine the type of document read. The MPIIDPEye dataset, which includes eye movement data of 10-minute comic, newspaper and text document readings from 20 participants, was used. First, the eye movements obtained over time were augmented by the non-linear interpolation technique. In order to process the data of each class with convolutional neural network, spectrogram images of the signals were created. Spectrogram images were given as input to Resnet architectures and the features in the Fc1000 layer were combined. Concatenated feature vectors were given as an input to feature selection algorithms. The most effective features in classification accuracy were determined and classified using the SVM algorithm. The classification was carried out for 3 different cases, and the highest accuracy of 98.41% was obtained for case-2, where Fixation, Position, and Blink properties were used.

show abstract

Automatic detection of keratoconus on Pentacam images using feature selection based on deep learning

Cited by 10 publications

References 29 publications

A Deep Feature Fusion of Improved Suspected Keratoconus Detection with Deep Learning

A Deep Feature Fusion of Improved Suspected Keratoconus Detection with Deep Learning

Machine Learning Techniques in Keratoconus Classification: A Systematic Review

Determining the Type of Document Read Using Eye Movement Properties by Hybrid CNN Method

Contact Info

Product

Resources

About