Detecting glaucoma based on spectral domain optical coherence tomography imaging of peripapillary retinal nerve fiber layer: a comparison study between hand-crafted features and deep learning model.
Zheng Ce, Xie Xiaolin, Huang Longtao, Chen Binyao, Yang Jianling, Lu Jiewei, Qiao Tong, Fan Zhun, Zhang Mingzhi
AI Summary
Deep learning models using SD-OCT pRNFL images significantly outperform traditional measurements for glaucoma detection, offering a highly accurate diagnostic tool for clinical use.
Abstract
Purpose
To develop a deep learning (DL) model for automated detection of glaucoma and to compare diagnostic capability against hand-craft features (HCFs) based on spectral domain optical coherence tomography (SD-OCT) peripapillary retinal nerve fiber layer (pRNFL) images.
Methods
A DL model with pre-trained convolutional neural network (CNN) based was trained using a retrospective training set of 1501 pRNFL OCT images, which included 690 images from 153 glaucoma patients and 811 images from 394 normal subjects. The DL model was further tested in an independent test set of 50 images from 50 glaucoma patients and 52 images from 52 normal subjects. A customized software was used to extract and measure HCFs including pRNFL thickness in average and four different sectors. Area under the receiver operator characteristics (AROC) curves was calculated to compare the diagnostic capability between DL model and hand-crafted pRNFL parameters.
Results
In this study, the DL model achieved an AROC of 0.99 [CI: 0.97 to 1.00] which was significantly larger than the AROC values of all other HCFs (AROCs 0.661 with 95% CI 0.549 to 0.772 for temporal sector, AROCs 0.696 with 95% CI 0.549 to 0.799 for nasal sector, AROCs 0.913 with 95% CI 0.855 to 0.970 for superior sector, AROCs 0.938 with 95% CI 0.894 to 0.982 for inferior sector, and AROCs 0.895 with 95% CI 0.832 to 0.957 for average).
Conclusion
Our study demonstrated that DL models based on pre-trained CNN are capable of identifying glaucoma with high sensitivity and specificity based on SD-OCT pRNFL images.
MeSH Terms
Shields Classification
Key Concepts5
The deep learning (DL) model achieved an Area Under the Receiver Operating Characteristics (AROC) curve of 0.99 [CI: 0.97 to 1.00] for detecting glaucoma based on spectral domain optical coherence tomography (SD-OCT) peripapillary retinal nerve fiber layer (pRNFL) images.
The deep learning (DL) model's AROC of 0.99 [CI: 0.97 to 1.00] was significantly larger than the AROC values of all other hand-crafted features (HCFs) for detecting glaucoma, which included temporal sector (AROC 0.661, 95% CI 0.549 to 0.772), nasal sector (AROC 0.696, 95% CI 0.549 to 0.799), superior sector (AROC 0.913, 95% CI 0.855 to 0.970), inferior sector (AROC 0.938, 95% CI 0.894 to 0.982), and average (AROC 0.895, 95% CI 0.832 to 0.957) peripapillary retinal nerve fiber layer (pRNFL) thickness.
Deep learning (DL) models based on pre-trained convolutional neural networks (CNN) are capable of identifying glaucoma with high sensitivity and specificity based on spectral domain optical coherence tomography (SD-OCT) peripapillary retinal nerve fiber layer (pRNFL) images.
A deep learning (DL) model based on a pre-trained convolutional neural network (CNN) was trained using a retrospective training set of 1501 peripapillary retinal nerve fiber layer (pRNFL) OCT images, which included 690 images from 153 glaucoma patients and 811 images from 394 normal subjects.
The deep learning (DL) model was tested in an independent test set of 50 images from 50 glaucoma patients and 52 images from 52 normal subjects.
Related Articles5
Age-related changes in optical coherence tomography glaucoma-related parameters: A systematic review.
Systematic ReviewArtificial Intelligence for Optical Coherence Tomography in Glaucoma.
ReviewArtificial Intelligence Deep Learning Models to Predict Spaceflight Associated Neuro-Ocular Syndrome.
Observational StudyAxial elongation in nonpathologic high myopia: Ocular structural changes and glaucoma diagnostic challenges.
ReviewPhysiological change in ganglion cell inner plexiform layer and nerve fibre layer thickness over six years.
Cohort StudyIs this article assigned to the wrong chapter(s)? Let us know.