Retinal automatic segmentation method based on prior information and optimized boundary tracking algorithm

At the present time optical coherence tomography (OCT) is among the most commonly used non-invasive imaging methods for the acquisition of large volumetric scans of human retinal tissues and vasculature. The substantial increase of accessible highly resolved 3D samples at the optic nerve head and the macula is directly linked to medical advancements in early detection of eye diseases. To resolve decisive information from extracted OCT volumes and to make it applicable for further diagnostic analysis, the exact measurement of retinal layer thicknesses serves as an essential task be done for each patient separately. However, manual examination of OCT scans is a demanding and time consuming task, which is typically made difficult by the presence of tissue-dependent speckle noise. Therefore, the elaboration of automated segmentation models has become an important task in the field of medical image processing. We propose a novel, purely data driven geometric approach to order-constrained 3D OCT retinal cell layer segmentation which takes as input data in any metric space and can be implemented using only simple, highly parallelizable operations. As opposed to many established retinal layer segmentation methods, we use only locally extracted features as input and do not employ any global shape prior. The physiological order of retinal cell layers and membranes is achieved through the introduction of a smoothed energy term. This is combined with additional regularization of local smoothness to yield highly accurate 3D segmentations. The approach thereby systematically avoid bias pertaining to global shape and is hence suited for the detection of anatomical changes of retinal tissue structure. To demonstrate its robustness, we compare two different choices of features on a data set of manually annotated 3D OCT volumes of healthy human retina. The quality of computed segmentations is compared to the state of the art in automatic retinal layer segmention as well as to manually annotated ground truth data in terms of mean absolute error and Dice similarity coefficient. Visualizations of segmented volumes are also provided.

Excitotoxicity from the impairment of glutamate uptake constitutes an important mechanism in neurodegenerative diseases such as Alzheimer’s, multiple sclerosis, and Parkinson's disease. Within the eye, excitotoxicity is thought to play a critical role in retinal ganglion cell death in glaucoma, diabetic retinopathy, retinal ischemia, and optic nerve injury, yet how excitotoxic injury impacts different retinal layers is not well understood. Here, we investigated the longitudinal effects of N-methyl-D-aspartate (NMDA)-induced excitotoxic retinal injury in a rat model using deep learning-assisted retinal layer thickness estimation. Before and after unilateral intravitreal NMDA injection in nine adult Long Evans rats, spectral-domain optical coherence tomography (OCT) was used to acquire volumetric retinal images in both eyes over 4 weeks. Ten retinal layers were automatically segmented from the OCT data using our deep learning-based algorithm. Retinal degeneration was evaluated using layer-specific retinal thickness changes at each time point (before, and at 3, 7, and 28 days after NMDA injection). Within the inner retina, our OCT results showed that retinal thinning occurred first in the inner plexiform layer at 3 days after NMDA injection, followed by the inner nuclear layer at 7 days post-injury. In contrast, the retinal nerve fiber layer exhibited an initial thickening 3 days after NMDA injection, followed by normalization and thinning up to 4 weeks post-injury. Our results demonstrated the pathological cascades of NMDA-induced neurotoxicity across different layers of the retina. The early inner plexiform layer thinning suggests early dendritic shrinkage, whereas the initial retinal nerve fiber layer thickening before subsequent normalization and thinning indicates early inflammation before axonal loss and cell death. These findings implicate the inner plexiform layer as an early imaging biomarker of excitotoxic retinal degeneration, whereas caution is warranted when interpreting the ganglion cell complex combining retinal nerve fiber layer, ganglion cell layer, and inner plexiform layer thicknesses in conventional OCT measures. Deep learning-assisted retinal layer segmentation and longitudinal OCT monitoring can help evaluate the different phases of retinal layer damage upon excitotoxicity.

As a function of the spatial position of the optical coherence tomography (OCT) image, retinal layer thickness is an important diagnostic indicator for many retinal diseases. Reliable segmentation of the retinal layer is necessary for extracting useful clinical information. However, manual segmentation of these layers is time-consuming and prone to bias. Furthermore, due to speckle noise, low image contrast, retinal detachment, and also irregular morphological features make the automatic segmentation task challenging. To alleviate these challenges, in this paper, we propose a new coarse-fine framework combining the full convolutional network (FCN) with a multiphase level set (named FCN-MLS) for automatic segmentation of nine boundaries in retinal spectral OCT images. In the coarse stage, FCN is used to learn the characteristics of specific retinal layer boundaries and achieve classification of four retinal layers. The boundaries are then extracted and the remaining boundaries are initialized based on a priori information about the thickness of the retinal layer. In order to prevent the overlapping of the segmentation interfaces, a regional restriction technique is used in the multi-phase level to evolve the boundaries to achieve fine nine retinal layers segmentation. Experimental results on 1280 B-scans show that the proposed method can segment nine retinal boundaries accurately. Compared with the manual delineation, the overall mean absolute boundary location difference and the overall mean absolute thickness difference were 5.88 ± 2.38μm and 5.81 ± 2.19μm, which showed a good consistency with manual segmentation by the physicians. Our experimental results also outperform state-of-the-art methods.

Quantification of retinal layers using automated segmentation of optical coherence tomography (OCT) images allows for longitudinal studies of retinal and neurological disorders in mice. The purpose of this study was to compare the performance of automated retinal layer segmentation algorithms with data from manual segmentation in mice using the Spectralis OCT. Spectral domain OCT images from 55 mice from three different mouse strains were analyzed in total. The OCT scans from 22 C57Bl/6, 22 BALBc, and 11 C3A.Cg-Pde6b+Prph2Rd2 /J mice were automatically segmented using three commercially available automated retinal segmentation algorithms and compared to manual segmentation. Fully automated segmentation performed well in mice and showed coefficients of variation (CV) of below 5% for the total retinal volume. However, all three automated segmentation algorithms yielded much thicker total retinal thickness values compared to manual segmentation data (P < 0.0001) due to segmentation errors in the basement membrane. Whereas the automated retinal segmentation algorithms performed well for the inner layers, the retinal pigmentation epithelium (RPE) was delineated within the sclera, leading to consistently thicker measurements of the photoreceptor layer and the total retina. The introduction of spectral domain OCT allows for accurate imaging of the mouse retina. Exact quantification of retinal layer thicknesses in mice is important to study layers of interest under various pathological conditions.

This work investigates the ability of texture analysis to yield discrimination of retinal tissue layers in the images provided by Optical Coherence Tomography (OCT). In fact, this relatively new imaging technology allows noninvasive visualization of retinal layers. Their segmentation is a prerequisite for any computer method that aims to objectively extract valuable information, regarding the condition and the progression of disease and therapy. Since the regularities of biological tissue can be captured by texture analysis in a straightforward way, a computer approach is proposed based on co-occurence matrices and artificial neural networks (ANN) for the classification and analysis of single retinal layers. A subset of ten normal eyes has been used for the training phase, and another subset of ten normal eyes has been used for testing the system performance. For inner retinal layers, accuracy was 79%, specificity about 71% and sensibility was 87%. Slightly lower values were obtained for outer retinal layers. These preliminary results suggest that this approach may be useful as a prototype system for a quantitative characterization of retinal tissue.

This study investigated the retinal segmental thicknesses in individuals with pseudoexfoliation syndrome and ocular hypertension. Maximum thinning was found at 6 mm inferior to the inner plexiform layer. This layer is very important for the early diagnosis of glaucoma. To analyze the thickness of the peripapillary retinal nerve fiber layer and 8 macular layers using optical coherence tomography in eyes with ocular hypertension (OHT) and pseudoexfoliation syndrome (PXS) and healthy eyes and to evaluate between-group differences. In a prospective study, the macular segmentation of retinal layers in 120 eyes of 120 participants was performed automatically using current Heidelberg Spectralis optical coherence tomography software, which provides measurements for 8 retinal layers. Thickness maps divided into nine subfields (ie, 1, 3, and 6mm) were extracted from the software for each retinal layer and compared between groups. The thinnest macular layers appeared in the ocular hypertensive PXS, normotensive PXS, and OHT groups in that order. In the inner retinal layers (macular retinal nerve fiber layer, ganglion cell layer, and inner plexiform layer), statistically significant differences emerged between the PXS and control groups in all quadrants of the 3 and 6mm areas. No significant difference between the OHT group and control group appeared except in the 6mm temporal quadrant of the inner plexiform layer (IPL). Receiving operating characteristic analysis revealed quadrants with high area-under-the-curve values at 3 and 6mm in macular segments in all 3 groups compared with the control group. In macular segment analysis, the inner retinal layers showed the most thinning in patients with ocular hypertensive PXS. According to receiving operating characteristic curve analysis, examinations performed 6mm inferior to the IPL, as the quadrant with the highest area under the curve in all 3 groups, are critical for the early diagnosis of glaucoma.

Segmentation of retinal layers in optical coherence tomography (OCT) is an essential step in OCT image analysis for screening, diagnosis, and assessment of retinal disease progression. Real-time segmentation together with high-speed OCT volume acquisition allows rendering of en face OCT of arbitrary retinal layers, which can be used to increase the yield rate of high-quality scans, provide real-time feedback during image-guided surgeries, and compensate aberrations in adaptive optics (AO) OCT without using wavefront sensors. We demonstrate here unprecedented real-time OCT segmentation of eight retinal layer boundaries achieved by 3 levels of optimization: 1) a modified, low complexity, neural network structure, 2) an innovative scheme of neural network compression with TensorRT, and 3) specialized GPU hardware to accelerate computation. Inferencing with the compressed network U-NetRT took 3.5 ms, improving by 21 times the speed of conventional U-Net inference without reducing the accuracy. The latency of the entire pipeline from data acquisition to inferencing was only 41 ms, enabled by parallelized batch processing. The system and method allow real-time updating of en face OCT and OCTA visualizations of arbitrary retinal layers and plexuses in continuous mode scanning. To the best our knowledge, our work is the first demonstration of an ophthalmic imager with embedded artificial intelligence (AI) providing real-time feedback.

Diabetic mellitus causes microvasculature changes in the retina which leads to diabetic retinopathy and may cause blindness if left unchecked. Spectral-Domain Optical Coherence Tomography (SDOCT) is a noninvasive imaging modality which could give precise information about the retinal layers. SDOCT retinal images of 75 subjects with uncontrolled diabetic mellitus for more than 2 years duration and images of 30 subjects with controlled diabetes or in normal condition are considered. The speckle noise in the images is smoothened using anisotropic diffusion filtering technique, and segmentation of Retinal Nerve Fiber layer (RNFL) along with Ganglion Cell Layer (GCL) and Inner Plexiform Layer (IPL) complex is performed using the axial gradient canny edge detection combined with a level set method. Textural features are obtained from the segmented layers, and classification of abnormality is done using SVM. The results showed that the retinal nerve fiber layer along with GCL+IPL complex thickness was reduced in subjects with even minimal diabetic retinopathy.

The correct segmentation of blood vessels in optical coherence tomography (OCT) images is an important requirement for better diagnosis of many retinal diseases. Although OCT blood vessel segmentation is often performed by applying vessel detection methods on 2D projection of OCT datasets, some papers investigate the vessel segmentation on OCT slices. The presence of shadows in outer retinal layers is established as the main factor for vessel localization; however, the shadow information fails to localize many important blood vessels. The proposed method is based on anatomical changes of Retinal Nerve Fiber Layer (RNFL) in presence of vessels. In this paper we find the thickening of RNFL by applying a layer segmentation algorithm on OCT slices and combine this information with shadow localization. Furthermore, a vessel detection method based on curvelet transform is also applied on 2D projection of OCTs to be added to localized vessels from OCTs. The results show that combination of vessel detection on 2D projection with vessel localization on OCTs can improve the accuracy up to 0.96 which is promisingly higher than older methods.

Background: Optical coherence tomography can be used to assess retinal degeneration in multiple sclerosis (MS). Thinning of the retinal nerve fibre layer and macular thickness have been well characterized, but newer devices allow quantification of all retinal layers. Objectives: The objective of this study was to evaluate the thickness of the paramacular retina, peripapillary retinal nerve fibre layer, and deeper paramacular layers in MS patient subgroups, using state-of-the-art optical coherence tomography. Methods: Using a Heidelberg Engineering Spectralis device, we performed paramacular volumetric retinal scans and circular peripapillary fibre-layer scans, manually segmenting different retinal layers into single horizontal foveal scans in 95 patients with definite MS (42 relapsing–remitting, 41 secondary progressive, 12 primary progressive), plus 91 age- and sex-matched controls. Results: Even without a history of optic neuritis, all MS subgroups had significant thinning of the peripapillary retinal nerve fibre layer, the paramacular retinal thickness and the retinal ganglion cell- and inner plexiform layer. Only in primary progressive MS was the inner nuclear layer significantly reduced. Conclusions: Our findings indicate a primary retinal pathology involving the inner nuclear layer in primary progressive MS. Results in eyes without history of optic neuritis suggest possible subclinical episodes of optic neuritis or retrograde trans-synaptic degeneration of retinal ganglion cells and their axons.

To evaluate changes in the segmented retinal layers of pituitary adenoma (PA) patients and to identify the relationship between these changes and visual function. A total of 47 (PA patients) and 22 (healthy subjects) eyes were reviewed from the medical records. The PA patients performed a visual field (VF) test before surgery and 1month after surgery. By optical coherence tomography scanning, eight retinal layers were measured: retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer, outer nuclear layer, retinal pigment epithelium, and photoreceptor layer. The PA group showed reduced RNFL, GCL, and IPL thicknesses (p = 0.004,< 0.001,< 0.001) and thicker INL thickness (p = 0.012) than did the controls. The mean deviation of preoperative VF in the PA group was positively correlated with RNFL, GCL, and IPL thicknesses (R = 0.664, 0.720, 0.664; p < 0.001,< 0.001,< 0.001) and negatively correlated with the INL thickness (R = -0.400; p = 0.010). Among the 47 eyes, 32 eyes (68%) were included for subgroup analysis. Preoperative RNFL, GCL, and IPL thicknesses were thicker in the postoperatively improved VF group (p = 0.019, 0.009, 0.005). The preoperative cutoff values for visual recovery were 23.6μm for RNFL thickness, 30.6μm for GCL thickness, and 28.9μm for IPL thickness. During chiasmal compression, the thickening of the INL has presented in addition to thinning of the inner retinal layers. Also, changes in retinal anatomical structures are related to the extent of VF defect and can be used as a predictor of postoperative visual recovery.

Optical coherence tomography (OCT) is one of imaging technologies to diagnose ophthalmic diseases and to monitor progresses of treatments. Therefore, OCT and OCT-angiography (OCTA) have actively developed in the field of ophthalmology, and retinal phantoms for evaluation of performance of OCT and OCTA also have been studied. In this presentation, we will show methods and results of a retinal phantom that can evaluate the optical performance of OCT and OCTA. The retinal phantom not only mimicked superficial vascular networks more realistically, but also implemented retinal layers with curvature. It consists of two microfluidic channels corresponding to superficial and deep retinal vessels, two multi-layered thin films, and base plate with retinal curvature. Each was attached with oxygen plasma for 1-min and fluorinated ethylene propylene (FEP) tubes was connected to the inlet and outlet holes. We could obtain cross-sectional OCT images and en-face OCTA images using lab-made OCT system.

7 - Optical coherence tomography: A review

Medical image segmentation is a critical field in the domain of computer vision and with the growing acclaim of deep learning based models, research in this field is constantly expanding. Optical coherence tomography (OCT) is a non-invasive method that scans the human's retina with depth. It has been hypothesized that the thickness of the retinal layers extracted from OCTs could be an efficient and effective biomarker for early diagnosis of AD. In this work, we aim to design a self-training model architecture for the task of segmenting the retinal layers in OCT scans. Neural architecture search (NAS) is a subfield of AutoML domain, which has a significant impact on improving the accuracy of machine vision tasks. We integrate the NAS algorithm with a Unet auto-encoder architecture as its backbone. Then, we employ our proposed model to segment the retinal nerve fiber layer in our preprocessed OCT images with the aim of AD diagnosis. In this work, we trained a super-resolution generative adversarial network on the raw OCT scans to improve the quality of the images before the modeling stage. In our architecture search strategy, different primitive operations suggested to find down- & up-sampling Unet cell blocks and the binary gate method has been applied to make the search strategy more practical. Our architecture search method is empirically evaluated by training on the Unet and NAS-Unet from scratch. Specifically, the proposed NAS-Unet training significantly outperforms the baseline human-designed architecture by achieving 95.1% in the mean Intersection over Union metric and 79.1% in the Dice similarity coefficient.

PurposeWe correlate optical coherence tomography (OCT) retinal layer thickness measurements with histology in wild-type and retinal degenerative pigs.MethodsOCT scans were obtained using the Bioptigen Envisu R2200. In normal pigs, three eyes were imaged in vivo, and three eyes were imaged after enucleation. In the Pro23His retinal degeneration pigs (P23H), one eye was imaged in vivo and four eyes were imaged after enucleation. All eyes were fixed in 4% paraformaldehyde and processed for histology. Corresponding retinal locations on OCT and histology were identified using anatomic landmarks (optic nerve, retinal vessels, visual streak). Individual retinal layer thicknesses were measured by two independent, masked graders, and intraclass correlation coefficients were used to determine agreement. OCT and histologic retinal thickness measurements were averaged and compared.ResultsOCT and histologic measurements correlated highly in normal and diseased eyes (R2 = 0.91 and 0.92, respectively), and scans performed in vivo and ex vivo did not differ significantly. Despite good overall correlation, certain individual retinal layers (e.g., retinal nerve fiber layer [NFL], inner [INL] and outer [ONL] nuclear layers) appeared thicker on OCT compared to histology, while other layers (e.g., retinal pigment epithelium) appeared thinner. No statistically significant difference was found between OCT and histology for any retinal layer thickness measurement.ConclusionsRetinal layer thickness measurements correlate well with histology in pig eyes, but differences in individual retinal layers may be seen.Translational RelevanceOCT may be used in pigs to measure retinal thicknesses with good overall correlation to histologic measurements.