Deep Learning for Diabetic Retinopathy Prediction

Diabetic retinopathy is a complication of diabetes mellitus. Its early diagnosis can prevent its progression and avoid the development of other major complications such as blindness. Deep learning and transfer learning appear in this context as powerful tools to aid in diagnosing this condition. The present work proposes to experiment with different models of pre-trained convolutional neural networks to determine which one fits best the problem of predicting diabetic retinopathy. The Diabetic Retinopathy Detection dataset supported by the EyePACS competition is used for evaluation. Seven pre-trained CNN models implemented in the Keras library developed in Python and, in this case, executed in the Kaggle platform, are used. Results show that no architecture performs better in all evaluation metrics. From a balanced behaviour perspective, the MobileNetV2 model stands out, with execution times almost half that of the slowest CNNs and without falling into overfitting with 20 learning epochs. InceptionResNetV2 stands out from the perspective of best performance, with a Kappa coefficient of 0.7588.

Automated summary

This study experiment with different pre-trained convolutional neural network models to predict diabetic retinopathy, showing that no architecture outperforms in all evaluation metrics. MobileNetV2 model stands out from a balanced behavior perspective with almost half the execution time of the slowest CNNs and no overfitting in 20 learning epochs. InceptionResNetV2 excels in terms of best performance, with a Kappa coefficient of 0.7588.