Currently, commercial OCT systems are equipped to segment only two or three layers. The key challenges to segmentation are a) low signal-to-noise ratio due to speckle noise b) vessel shadows leading to intra-layer inhomogeneity c) lack of distinct boundaries and d) morphological changes in boundary smoothness.
A deep learning framework for segmentation of retinal layers from OCT images
We explored deep learning techniques to automate all these steps and handle the presence/absence of pathologies. A model is proposed consisting of a combination of Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM). The CNN is used to extract layers of interest image and extract the edges, while the LSTM is used to trace the layer boundary. This model is trained on a mixture of normal and AMD cases using minimal data. Validation results on three public datasets show that the pixel-wise mean absolute error obtained with our system is 1.30 ± 0.48 which is lower than the inter-marker error of 1.79 ± 0.76. Our model's performance is also on par with the existing methods.
End-to-end learning of a conditional random field for layer segmentation
Existing Energy minimization based methods employ handcrafted cost terms to define their energy and are not robust to the presence of abnormalities. We propose a novel, Linearly Parameterized, Conditional Random Field (LP-CRF) model whose energy is learnt from a set of training images in an end-to-end manner. The proposed LP-CRF comprises two convolution filter banks to capture the appearance of each tissue region and boundary, the relative weights of the shape priors and an additional term based on the appearance similarity of the adjacent boundary points. All the energy terms are jointly learnt using the Structured Support Vector Machine. The proposed method segments all retinal boundaries in a single step. Our method was evaluated on 107 Normal and 220 AMD B-scan images and found to outperform three publicly available OCT segmentation software. The average unsigned boundary localization error is pixels for segmentation of 8 boundaries on the Normal dataset and pixels for 3 boundaries on the combined AMD and Normal dataset establishing the robustness of the proposed method.
A supervised joint multi-layer segmentation framework using CRFs
The existing energy minimization based methods employ multiple, manually handcrafted cost terms and often fail in the presence of pathologies. In this work, we eliminate the need to handcraft the energy by learning it from training images in an end-to-end manner. Our method can be easily adapted to pathologies by re-training it on an appropriate dataset. We propose a Conditional Random Field (CRF) framework for the joint multi-layer segmentation of OCT B-scans. The appearance of each retinal layer and boundary is modeled by two convolutional filter banks and the shape priors are modeled using Gaussian distributions. The total CRF energy is linearly parameterized to allow a joint, end-to-end training by employing the Structured Support Vector Machine formulation.
The proposed method outperformed three benchmark algorithms on four public datasets. The NORMAL-1 and NORMAL-2 datasets contain healthy OCT B-scans while the AMD-1 and DME-1 dataset contain B-scans of AMD and DME cases respectively. The proposed method achieved an average unsigned boundary localization error (U-BLE) of 1.52 pixels on NORMAL-1, 1.11 pixels on NORMAL-2 and 2.04 pixels on the combined NORMAL-1 and DME-1 dataset across the eight layer boundaries, outperforming the three benchmark methods in each case. The Dice coefficient was 0.87 on NORMAL-1, 0.89 on NORMAL-2 and 0.84 on the combined NORMAL-1 and DME-1 dataset across the seven retinal layers. On the combined NORMAL-1 and AMD-1 dataset, we achieved an average U-BLE of 1.86 pixels on the ILM, inner and outer RPE boundaries and a Dice of 0.98 for the ILM-RPEin region and 0.81 for the RPE layer.