Two-stage adversarial learning based unsupervised domain adaptation for retinal OCT segmentation.

BACKGROUND: Deep learning based optical coherence tomography (OCT) segmentation methods have achieved excellent results, allowing quantitative analysis of large-scale data. However, OCT images are often acquired by different devices or under different imaging protocols, which leads to serious domain shift problem. This in turn results in performance degradation of segmentation models. PURPOSE: Aiming at the domain shift problem, we propose a two-stage adversarial learning based network (TSANet) that accomplishes unsupervised cross-domain OCT segmentation. METHODS: In the first stage, a Fourier transform based approach is adopted to reduce image style differences from the image level. Then, adversarial learning networks, including a segmenter and a discriminator, are designed to achieve inter-domain consistency in the segmentation output. In the second stage, pseudo labels of selected unlabeled target domain training data are used to fine-tune the segmenter, which further improves its generalization capability. The proposed method was tested on cross-domain datasets for choroid or retinoschisis segmentation tasks. For choroid segmentation, the model was trained on 400 images and validated on 100 images from the source domain, and then trained on 1320 unlabeled images and tested on 330 images from target domain I, and also trained on 400 unlabeled images and tested on 200 images from target domain II. For retinoschisis segmentation, the model was trained on 1284 images and validated on 312 images from the source domain, and then trained on 1024 unlabeled images and tested on 200 images from the target domain. RESULTS: The proposed method achieved significantly improved results over that without domain adaptation, with improvement of 8.34%, 55.82% and 3.53% in intersection over union (IoU) respectively for the three test sets. The performance is better than some state-of-the-art domain adaptation methods. CONCLUSIONS: The proposed TSANet, with image level adaptation, feature level adaptation and pseudo-label based fine-tuning, achieved excellent cross-domain generalization. This alleviates the burden of obtaining additional manual labels when adapting the deep learning model to new OCT data.

Group-wise context selection network for choroid segmentation in optical coherence tomography.

Choroid thickness measured from optical coherence tomography (OCT) images has emerged as a vital metric in the management of retinal diseases such as high myopia. In this paper, we propose a novel group-wise context selection network (referred to as GCS-Net) to segment the choroid of either normal or high myopia eyes. To deal with the diverse choroid thickness and the variable shape of the pathological retina, GCS-Net adopts the group-wise channel dilation (GCD) module and the group-wise spatial dilation module, which can automatically select group-wise multi-scale information under the guidance of channel attention or spatial attention, and enhance the consistency between the receptive field and the target area. Furthermore, a boundary optimization network with a new edge loss is incorporated to improve the resulting choroid boundary by deep supervision. Experimental results evaluated on a dataset composed of 1650 clinically obtained OCT B-scans show that the proposed GCS-Net can achieve a Dice similarity coefficient of 95.97 ± 0.54%, which outperforms some state-of-the-art segmentation networks.