The analysis of the choroid in the eye is crucial for our understanding of a range of ocular diseases and physiological processes. Optical coherence tomography (OCT) imaging provides the ability to capture highly detailed cross-sectional images of the choroid yet only a very limited number of commercial OCT instruments provide methods for automatic segmentation of choroidal tissue. Manual annotation of the choroidal boundaries is often performed but this is impractical due to the lengthy time taken to analyse large volumes of images. Therefore, there is a pressing need for reliable and accurate methods to automatically segment choroidal tissue boundaries in OCT images. In this work, a variety of patch-based and fully-convolutional deep learning methods are proposed to accurately determine the location of the choroidal boundaries of interest. The effect of network architecture, patch-size and contrast enhancement methods was tested to better understand the optimal architecture and approach to maximize performance. The results are compared with manual boundary segmentation used as a ground-truth, as well as with a standard image analysis technique. Results of total retinal layer segmentation are also presented for comparison purposes. The findings presented here demonstrate the benefit of deep learning methods for segmentation of the chorio-retinal boundary analysis in OCT images.

No abstract

Purpose To use a deep learning model to develop a fully automated method (fully semantic network and graph search [FS-GS]) of retinal segmentation for optical coherence tomography (OCT) images from patients with Stargardt disease. Methods Eighty-seven manually segmented (ground truth) OCT volume scan sets (5171 B-scans) from 22 patients with Stargardt disease were used for training, validation and testing of a novel retinal boundary detection approach (FS-GS) that combines a fully semantic deep learning segmentation method, which generates a per-pixel class prediction map with a graph-search method to extract retinal boundary positions. The performance was evaluated using the mean absolute boundary error and the differences in two clinical metrics (retinal thickness and volume) compared with the ground truth. The performance of a separate deep learning method and two publicly available software algorithms were also evaluated against the ground truth. Results FS-GS showed an excellent agreement with the ground truth, with a boundary mean absolute error of 0.23 and 1.12 pixels for the internal limiting membrane and the base of retinal pigment epithelium or Bruch's membrane, respectively. The mean difference in thickness and volume across the central 6 mm zone were 2.10 µm and 0.059 mm 3 . The performance of the proposed method was more accurate and consistent than the publicly available OCTExplorer and AURA tools. Conclusions The FS-GS method delivers good performance in segmentation of OCT images of pathologic retina in Stargardt disease. Translational Relevance Deep learning models can provide a robust method for retinal segmentation and support a high-throughput analysis pipeline for measuring retinal thickness and volume in Stargardt disease.

Deep learning methods have enabled a fast, accurate and automated approach for retinal layer segmentation in posterior segment OCT images. Due to the success of semantic segmentation methods adopting the U-Net, a wide range of variants and improvements have been developed and applied to OCT segmentation. Unfortunately, the relative performance of these methods is difficult to ascertain for OCT retinal layer segmentation due to a lack of comprehensive comparative studies, and a lack of proper matching between networks in previous comparisons, as well as the use of different OCT datasets between studies. In this paper, a detailed and unbiased comparison is performed between eight U-Net architecture variants across four different OCT datasets from a range of different populations, ocular pathologies, acquisition parameters, instruments and segmentation tasks. The U-Net architecture variants evaluated include some which have not been previously explored for OCT segmentation. Using the Dice coefficient to evaluate segmentation performance, minimal differences were noted between most of the tested architectures across the four datasets. Using an extra convolutional layer per pooling block gave a small improvement in segmentation performance for all architectures across all four datasets. This finding highlights the importance of careful architecture comparison (e.g. ensuring networks are matched using an equivalent number of layers) to obtain a true and unbiased performance assessment of fully semantic models. Overall, this study demonstrates that the vanilla U-Net is sufficient for OCT retinal layer segmentation and that state-of-the-art methods and other architectural changes are potentially unnecessary for this particular task, especially given the associated increased complexity and slower speed for the marginal performance gains observed. Given the U-Net model and its variants represent one of the most commonly applied image segmentation methods, the consistent findings across several datasets here are likely to translate to many other OCT datasets and studies. This will provide significant value by saving time and cost in experimentation and model development as well as reduced inference time in practice by selecting simpler models.

No abstract