Asymmetry between right and left fundus images identified using convolutional neural networks

Abstract

We analyzed fundus images to identify whether convolutional neural networks (CNNs) can discriminate between right and left fundus images. We gathered 98,038 fundus photographs from the Gyeongsang National University Changwon Hospital, South Korea, and augmented these with the Ocular Disease Intelligent Recognition dataset. We created eight combinations of image sets to train CNNs. Class activation mapping was used to identify the discriminative image regions used by the CNNs. CNNs identified right and left fundus images with high accuracy (more than 99.3% in the Gyeongsang National University Changwon Hospital dataset and 91.1% in the Ocular Disease Intelligent Recognition dataset) regardless of whether the images were flipped horizontally. The depth and complexity of the CNN affected the accuracy (DenseNet121: 99.91%, ResNet50: 99.86%, and VGG19: 99.37%). DenseNet121 did not discriminate images composed of only left eyes (55.1%, p = 0.548). Class activation mapping identified the macula as the discriminative region used by the CNNs. Several previous studies used the flipping method to augment data in fundus photographs. However, such photographs are distinct from non-flipped images. This asymmetry could result in undesired bias in machine learning. Therefore, when developing a CNN with fundus photographs, care should be taken when applying data augmentation with flipping.

Asymmetry between right and left optical coherence tomography images identified using convolutional neural networks

“…For Set 6(R v L v D 121 ) images, CAM highlighted not only the parafovea, but also the fovea. In a previous study using fundus photography 13 , CAM brightly highlighted the temporal parafovea and moderately highlighted the fovea. It is possible that the asymmetric differ on the location of the retina, and the temporal parafovea may have a larger asymmetric than the superior and inferior parafovea.…”

“…Our previous study 13 showed that left fundus images are not mirror-symmetric with respect to right fundus images. CNNs are capable of distinguishing the left from right fundus with an accuracy greater than 99.9%.…”

“…We used a pre-trained model to improve accuracy 31 – 34 , as previous studies have demonstrated that the features of pre-trained models are transferable to ophthalmology data 33 . Densenet121 35 , ResNet50 36 , and VGG19 37 , embedded in the Keras package, showed excellent ability to distinguish left and right fundus photographs 13 . The outputs of DenseNet121, ResNet50, and VGG19 were connected to the fully connected layer and Softmax layer.…”

“…A fundus image of the left eye appears as a mirror image of the fundus image of the right eye. In a previous study of CNNs, we identified asymmetries in fundus photographs 13 . However, fundus images can also be affected by several factors such as the camera lens, flash, and room lighting conditions.…”

Estimation of best corrected visual acuity based on deep neural network

“…Specifically, CNN-based detection and classification of DR, which is a diabetes complication causing blindness, has been investigated 42 – 45 . Moreover, a study confirmed that the CNN-based approach could be used for the estimation of gender from retinal fundus image 46 and the identification of left and right fundus image 47 . Since the spatial feature of fundus images can be extracted by CNNs, we examined the feasibility of CNN-based BCVA estimation in this study.…”