Related Experiment Videos
Automatic Diagnosis of Plus Disease in Retinopathy of Prematurity Using Deep Learning
Eşay Kiran Yenice1, Çağatay Berke Erdaş2
1Department of Ophthalmology, University of Health Sciences, Bilkent City Hospital, Ankara, Türkiye.
Insights
Deep learning algorithms can automatically detect plus disease in infants with retinopathy of prematurity (ROP) from retinal images. This AI-driven approach achieves high accuracy, sensitivity, and specificity for diagnosing this critical condition.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Plus disease is a critical indicator for treatment in retinopathy of prematurity (ROP).
- Accurate diagnosis of plus disease is essential for timely intervention in ROP.
- Current diagnostic methods rely on clinical ophthalmoscopic examinations.
Purpose of the Study:
- To develop and evaluate deep learning (DL) algorithms for the automatic diagnosis of plus disease in ROP.
- To predict plus disease using DL models trained on infant retinal images.
- To assess the performance of DL algorithms in identifying plus disease in ROP.
Main Methods:
- Retinal images from 600 infants screened for ROP were analyzed.
- Images were classified as normal, pre-plus, or plus disease.
- Deep learning models (EfficientNetB7, InceptionResNetV2, VGG16) were trained using 10-fold cross-validation.
Main Results:
- InceptionResNetV2 achieved 0.90 sensitivity, 0.92 specificity, and 0.91 accuracy for plus versus no-plus disease diagnosis.
- The area under the ROC curve was 0.91 for plus disease detection.
- The algorithm demonstrated 0.81 sensitivity, 0.97 specificity, and 0.88 accuracy for detecting pre-plus disease or worse versus normal.
Conclusions:
- Deep learning algorithms can effectively automate the diagnosis of plus disease in ROP.
- The developed DL models exhibit high sensitivity, specificity, and accuracy.
- This AI approach shows promise for improving ROP management and treatment decisions.
Objectives:
Plus disease, which can be diagnosed during clinical ophthalmoscopic examinations, is the most important feature in determining retinopathy of prematurity (ROP) requiring treatment. In the current study, we aimed to automatic diagnose and predict plus disease based on deep learning (DL) from retinal images of infants with ROP.
Methods:
600 retinal images from infants screened for ROP were evaluated. Each image was classified as normal, pre-plus and plus disease. After image pre-processing, the images were distributed into groups equal to the number of normal eye images and were used for training DL algorithms such as EfficientNetB7, InceptionResNetV2 and VGG16. The algorithms were trained with 10-fold cross-validation, and results are reported as sensitivity, specificity, accuracy, receiver operating characteristic (ROC) curve, and area under the curve (AUC).
Results:
Of the 600 retinal images included, 258 images obtained after pre-processing were graded as 86 (33.3%) normal, 86 (33.3%) as pre-plus disease, and 86 (33.3%) as plus disease. For plus versus no-plus disease diagnosis, among the algorithms, InceptionResNetV2 achieved 0.90 sensitivity, 0.92 specificity, and 0.91 accuracy. Area under the ROC curve was 0.91. For detection of pre-plus disease or worse versus normal, the algorithm achieved 0.81 sensitivity, 0.97 specificity, and 0.88 accuracy.
Conclusion:
Our results showed that DL algorithms can automatically diagnose plus disease in ROP with high sensitivity, high specificity, and high accuracy.
