Semi-supervised image segmentation using a residual-driven mean teacher and an exponential Dice loss

Semi-supervised segmentation plays an important role in computer vision and medical image analysis and can alleviate the burden of acquiring abundant expert-annotated images. In this paper, we developed a residual-driven semi-supervised segmentation method (termed RDMT) based on the classical mean teacher (MT) framework by introducing a novel model-level residual perturbation and an exponential Dice (eDice) loss. The introduced perturbation was integrated into the exponential moving average (EMA) scheme to enhance the performance of the MT, while the eDice loss was used to improve the detection sensitivity of a given network to object boundaries. We validated the developed method by applying it to segment 3D Left Atrium (LA) and 2D optic cup (OC) from the public LASC and REFUGE datasets based on the V-Net and U-Net, respectively. Extensive experiments demonstrated that the developed method achieved the average Dice score of 0.8776 and 0.7751, when trained on 10% and 20% labeled images, respectively for the LA and OC regions depicted on the LASC and REFUGE datasets. It significantly outperformed the MT and can compete with several existing semi-supervised segmentation methods (i.e., HCMT, UAMT, DTC and SASS).

Accurate C/D ratio estimation with elliptical fitting for OCT image based on joint segmentation and detection network

Proper estimation of the cup-to-disc ratio (C/D ratio) plays a significant role in ophthalmic examinations, and it is urgent to improve the efficiency of C/D ratio automatic measurement. Therefore, we propose a new method for measuring the C/D ratio of OCTs in normal subjects. Firstly, the end-to-end deep convolution network is used to segment and detect the inner limiting membrane (ILM) and the two Bruch's membrane opening (BMO) terminations. Then, we introduce an ellipse fitting technique to post-process the edge of the optic disc. Finally, the proposed method is evaluated on 41 normal subjects using the optic-disc-area scanning mode of three machines: BV1000, Topcon 3D OCT-1, and Nidek ARK-1. In addition, pairwise correlation analyses are carried out to compare the C/D ratio measurement method of BV1000 to existing commercial OCT machines as well as other state-of-the-art methods. The correlation coefficient between the C/D ratio calculated by BV1000 and the C/D ratio calculated by manual annotation is 0.84, which indicates that the proposed method has a strong correlation with the results of manual annotation by ophthalmologists. Moreover, in comparison between BV1000, Topcon and Nidek in practical screening among normal subjects, the proportion of the C/D ratio less than 0.6 calculated by BV1000 accounts for 96.34%, which is the closest to the clinical statistics among the three OCT machines. The above experimental results and analysis show that the proposed method performs well in cup and disc detection and C/D ratio measurement, and compared with the existing commercial OCT equipment, the C/D ratio measurement results are relatively close to reality, which has certain clinical application value.

Cell fate decisions, transcription factors and signaling during early retinal development

The development of the vertebrate eyes is a complex process starting from anterior-posterior and dorso-ventral patterning of the anterior neural tube, resulting in the formation of the eye field. Symmetrical separation of the eye field at the anterior neural plate is followed by two symmetrical evaginations to generate a pair of optic vesicles. Next, reciprocal invagination of the optic vesicles with surface ectoderm-derived lens placodes generates double-layered optic cups. The inner and outer layers of the optic cups develop into the neural retina and retinal pigment epithelium (RPE), respectively. In vitro produced retinal tissues, called retinal organoids, are formed from human pluripotent stem cells, mimicking major steps of retinal differentiation in vivo. This review article summarizes recent progress in our understanding of early eye development, focusing on the formation the eye field, optic vesicles, and early optic cups. Recent single-cell transcriptomic studies are integrated with classical in vivo genetic and functional studies to uncover a range of cellular mechanisms underlying early eye development. The functions of signal transduction pathways and lineage-specific DNA-binding transcription factors are dissected to explain cell-specific regulatory mechanisms underlying cell fate determination during early eye development. The functions of homeodomain (HD) transcription factors Otx2, Pax6, Lhx2, Six3 and Six6, which are required for early eye development, are discussed in detail. Comprehensive understanding of the mechanisms of early eye development provides insight into the molecular and cellular basis of developmental ocular anomalies, such as optic cup coloboma. Lastly, modeling human development and inherited retinal diseases using stem cell-derived retinal organoids generates opportunities to discover novel therapies for retinal diseases.

An Efficient Hierarchical Optic Disc and Cup Segmentation Network Combined with Multi-task Learning and Adversarial Learning

Automatic and accurate segmentation of optic disc (OD) and optic cup (OC) in fundus images is a fundamental task in computer-aided ocular pathologies diagnosis. The complex structures, such as blood vessels and macular region, and the existence of lesions in fundus images bring great challenges to the segmentation task. Recently, the convolutional neural network-based methods have exhibited its potential in fundus image analysis. In this paper, we propose a cascaded two-stage network architecture for robust and accurate OD and OC segmentation in fundus images. In the first stage, the U-Net like framework with an improved attention mechanism and focal loss is proposed to detect accurate and reliable OD location from the full-scale resolution fundus images. Based on the outputs of the first stage, a refined segmentation network in the second stage that integrates multi-task framework and adversarial learning is further designed for OD and OC segmentation separately. The multi-task framework is conducted to predict the OD and OC masks by simultaneously estimating contours and distance maps as auxiliary tasks, which can guarantee the smoothness and shape of object in segmentation predictions. The adversarial learning technique is introduced to encourage the segmentation network to produce an output that is consistent with the true labels in space and shape distribution. We evaluate the performance of our method using two public retinal fundus image datasets (RIM-ONE-r3 and REFUGE). Extensive ablation studies and comparison experiments with existing methods demonstrate that our approach can produce competitive performance compared with state-of-the-art methods.

A multi-scale convolutional neural network with context for joint segmentation of optic disc and cup

Glaucoma is the leading cause of irreversible blindness. For glaucoma screening, the cup to disc ratio (CDR) is a significant indicator, whose calculation relies on the segmentation of optic disc(OD) and optic cup(OC) in color fundus images. This study proposes a residual multi-scale convolutional neural network with a context semantic extraction module to jointly segment the OD and OC. The proposed method uses a W-shaped backbone network, including image pyramid multi-scale input with the side output layer as an early classifier to generate local prediction output. The proposed method includes a context extraction module that extracts contextual semantic information from multiple level receptive field sizes and adaptively recalibrates channel-wise feature responses. It can effectively extract global information and reduce the semantic gaps in the fusion of deep and shallow semantic information. We validated the proposed method on four datasets, including DRISHTI-GS1, REFUGE, RIM-ONE r3, and a private dataset. The overlap errors are 0.0540, 0.0684, 0.0492, 0.0511 in OC segmentation and 0.2332, 0.1777, 0.2372, 0.2547 in OD segmentation, respectively. Experimental results indicate that the proposed method can estimate the CDR for a large-scale glaucoma screening.

Stem cell-based retina models

From the early days of cell biological research, the eye-especially the retina-has evoked broad interest among scientists. The retina has since been thoroughly investigated and numerous models have been exploited to shed light on its development, morphology, and function. Apart from various animal models and human clinical and anatomical research, stem cell-based models of animal and human cells of origin have entered the field, especially during the last decade. Despite the observation that the retina of different species comprises endogenous stem cells, most stem cell-related research in the human retina is now based on pluripotent stem cell models. Herein, systems of two-dimensional (2D) cultures and co-cultures of distinctly differentiated retinal subtypes revealed a variety of cellular aspects but have in many aspects been replaced by three-dimensional (3D) structures-the so-called retinal organoids. These organoids not only contain all major retinal cell subtypes compared to the physiological situation, but also show a distinct layering in close proximity to the in vivo morphology. Nevertheless, all these models have inherent advantages and disadvantages, which are expounded and summarized in this review. Finally, we discuss current application aspects of stem cell-based retina models and the specific promises they hold for the future.

Early Loss of Optic Cup with Increased Intracranial Pressure

A 54-year-old woman developed increased intracranial pressure due to chronic dural venous sinus thrombosis in the absence of significant comorbidity. In the years preceding diagnosis, the patient underwent routine diabetic retinopathy screening; however, gradual loss of optic cup and disc swelling was only retrospectively noted after marked papilloedema was evident at an optometry visit. The patient made a full recovery once the diagnosis was established. This case is novel in the literature as routine photographs, taken for diabetic retinopathy screening, demonstrate the natural history of papilloedema before medical intervention. Interestingly, these images show that a traditionally "late" sign of papilloedema - loss of cup - can be the sentinel abnormality. This finding may have significant clinical implications and reinforces the need for clinicians to compare investigation results with historical patient data.

An efficient optic cup segmentation method decreasing the influences of blood vessels

BACKGROUND

Optic cup is an important structure in ophthalmologic diagnosis such as glaucoma. Automatic optic cup segmentation is also a key issue in computer aided diagnosis based on digital fundus image. However, current methods didn't effectively solve the problem of edge blurring caused by blood vessels around the optic cup.

METHODS

In this study, an improved Bertalmio-Sapiro-Caselles-Ballester (BSCB) model was proposed to eliminate the noising induced by blood vessel. First, morphological operations were performed to get the enhanced green channel image. Then blood vessels were extracted and filled by improved BSCB model. Finally, Local Chart-Vest model was used to segment the optic cup. A total of 94 samples which included 32 glaucoma fundus images and 62 normal fundus images were experimented.

RESULTS

The evaluation parameters of F-score and the boundary distance achieved by the proposed method against the results from experts were 0.7955 ± 0.0724 and 11.42 ± 3.61, respectively. Average vertical optic cup-to-disc ratio values of the normal and glaucoma samples achieved by the proposed method were 0.4369 ± 0.1193 and 0.7156 ± 0.0698, which were also close to those by experts. In addition, 39 glaucoma images from the public dataset RIM-ONE were also used for methodology evaluation.

CONCLUSIONS

The results showed that our proposed method could overcome the influence of blood vessels in some degree and was competitive to other current optic cup segmentation algorithms. This novel methodology will be expected to use in clinic in the field of glaucoma early detection.

How mechanical forces shape the developing eye

In the vertebrate embryo, the eyes develop from optic vesicles that grow laterally outward from the brain tube and contact the overlying surface ectoderm. Within the region of contact, each optic vesicle and the surface ectoderm thicken to form placodes, which then invaginate to create the optic cup and lens pit, respectively. Eventually, the optic cup becomes the retina, while the lens pit closes to form the lens vesicle. Here, we review current hypotheses for the physical mechanisms that create these structures and present novel three-dimensional computer (finite-element) models to illustrate the plausibility and limitations of these hypotheses. Taken together, experimental and numerical results suggest that the driving forces for early eye morphogenesis are generated mainly by differential growth, actomyosin contraction, and regional apoptosis, with morphology mediated by physical constraints provided by adjacent tissues and extracellular matrix. While these studies offer new insight into the mechanics of eye development, future work is needed to better understand how these mechanisms are regulated to precisely control the shape of the eye.

Eye and heart morphogenesis are dependent on melatonin signaling in chick embryos

Calmodulin is vital for chick embryos morphogenesis in the incubation time 48-66 h when the rudimentary C-shaped heart attains an S-shaped pattern and the optic vesicles develop into optic cups. Melatonin is in the extraembryonic yolk sac of the avian egg; melatonin binds calmodulin. The aim of this study was to investigate the function of melatonin in the formation of the chick embryo optic cups and S-shaped heart, by pharmacological methods and immunoassays. Mel1a melatonin receptor immunofluorescence was distributed in the optic cups and rudimentary hearts. We separated embryonated chicken eggs at 48 h of incubation into basal, control and drug-treated groups, with treatment applied in the egg air sac. At 66 h of incubation, embryos were excised from the eggs and analyzed. Embryos from the basal, control (distilled water), melatonin and 6-chloromelatonin (melatonin receptor agonist) groups had regular optic cups and an S-shaped heart, while those from the calmidazolium (calmodulin inhibitor) group did not. Embryos from the luzindole (melatonin receptor antagonist) and prazosin (Mel1c melatonin receptor antagonist) groups did not have regular optic cups. Embryos from the 4-P-PDOT (Mel1b melatonin receptor antagonist) group did not have an S-shaped heart. Previous application of the melatonin, 6-chloromelatonin or forskolin (adenylate cyclase enhancer) prevented the abnormal appearance of chick embryos from the calmidazolium, luzindole, prazosin and 4-P-PDOT groups. However, 6-chloromelatonin and forskolin only partially prevented the development of defective eye cups in embryos from the calmidazolium group. The results suggested that melatonin modulates chick embryo morphogenesis via calmodulin and membrane receptors.

Joint optic disc and cup boundary extraction from monocular fundus images

BACKGROUND AND OBJECTIVE

Accurate segmentation of optic disc and cup from monocular color fundus images plays a significant role in the screening and diagnosis of glaucoma. Though optic cup is characterized by the drop in depth from the disc boundary, most existing methods segment the two structures separately and rely only on color and vessel kink based cues due to the lack of explicit depth information in color fundus images.

METHODS

We propose a novel boundary-based Conditional Random Field formulation that extracts both the optic disc and cup boundaries in a single optimization step. In addition to the color gradients, the proposed method explicitly models the depth which is estimated from the fundus image itself using a coupled, sparse dictionary trained on a set of image-depth map (derived from Optical Coherence Tomography) pairs.

RESULTS

The estimated depth achieved a correlation coefficient of 0.80 with respect to the ground truth. The proposed segmentation method outperformed several state-of-the-art methods on five public datasets. The average dice coefficient was in the range of 0.87-0.97 for disc segmentation across three datasets and 0.83 for cup segmentation on the DRISHTI-GS1 test set. The method achieved a good glaucoma classification performance with an average AUC of 0.85 for five fold cross-validation on RIM-ONE v2.

CONCLUSIONS

We propose a method to jointly segment the optic disc and cup boundaries by modeling the drop in depth between the two structures. Since our method requires a single fundus image per eye during testing it can be employed in the large-scale screening of glaucoma where expensive 3D imaging is unavailable.

Generation of a Three-Dimensional Retinal Tissue from Self-Organizing Human ESC Culture

A three-dimensional (3D) tissue generated in vitro is a promising source to study developmental biology and regenerative medicine. In the last decade, Yoshiki Sasai's group have developed a 3D stem cell culture technique known as SFEBq and demonstrated that embryonic stem cells (ESCs) have an ability to self-organize stratified neural tissue including 3D-retina. Furthermore, we have reported that ESC-derived retinal tissue can form an optic cup and a ciliary margin, which are unique structures in the developing retina. In this review, we focus on self-organizing culture technique to generate 3D-retina from human ESCs.

Agreement among ophthalmologists in marking the optic disc and optic cup in fundus images

The purpose of this paper was to study the agreement between six ophthalmologists who manually marked the optic nerve head using fundus images. Four different parameters were considered from the manual marking images: (1) disc (area and centroid), (2) cup (area and centroid), (3) horizontal and vertical cup-to-disc ratios, and (4) including the previous two parameters for both horizontal and vertical cup-to-disc ratios, and investigated the comprehensive agreement and accuracy among all the ophthalmologists. The best agreement percentage for all the parameters combined was between ophthalmologists number one and three for 44 % of images, and the best accuracy was for ophthalmologist number one with 77.4 % of 315 total tested images. Our analysis shows that more than half of the images in the dataset were not agreed upon when considering all the parameters together.

Proper closure of the optic fissure requires ephrin A5-EphB2-JNK signaling

The development of complex organs such as the eye requires a delicate and coordinated balance of cell division and cell death. Although apoptosis is prevalent in the proximoventral optic cup, the precise role it plays in eye development needs to be investigated further. In this study, we show that reduced apoptosis in the proximoventral optic cup prevents closure of the optic fissure. We also show that expression of ephrin A5 (Efna5) partially overlaps with Eph receptor B2 (Ephb2) expression in the proximoventral optic cup and that binding of EphB2 to ephrin A5 induces a sustained activation of JNK. This prolonged JNK signal promotes apoptosis and prevents cell proliferation. Thus, we propose that the unique cross-subclass interaction of EphB2 with ephrin A5 has evolved to function upstream of JNK signaling for the purpose of maintaining an adequate pool of progenitor cells to ensure proper closure of the optic fissure.

An adaptive threshold based image processing technique for improved glaucoma detection and classification

Glaucoma is an optic neuropathy which is one of the main causes of permanent blindness worldwide. This paper presents an automatic image processing based method for detection of glaucoma from the digital fundus images. In this proposed work, the discriminatory parameters of glaucoma infection, such as cup to disc ratio (CDR), neuro retinal rim (NRR) area and blood vessels in different regions of the optic disc has been used as features and fed as inputs to learning algorithms for glaucoma diagnosis. These features which have discriminatory changes with the occurrence of glaucoma are strategically used for training the classifiers to improve the accuracy of identification. The segmentation of optic disc and cup based on adaptive threshold of the pixel intensities lying in the optic nerve head region. Unlike existing methods the proposed algorithm is based on an adaptive threshold that uses local features from the fundus image for segmentation of optic cup and optic disc making it invariant to the quality of the image and noise content which may find wider acceptability. The experimental results indicate that such features are more significant in comparison to the statistical or textural features as considered in existing works. The proposed work achieves an accuracy of 94.11% with a sensitivity of 100%. A comparison of the proposed work with the existing methods indicates that the proposed approach has improved accuracy of classification glaucoma from a digital fundus which may be considered clinically significant.

Wnt ligands from the embryonic surface ectoderm regulate 'bimetallic strip' optic cup morphogenesis in mouse

The Wnt/β-catenin response pathway is central to many developmental processes. Here, we assessed the role of Wnt signaling in early eye development using the mouse as a model system. We showed that the surface ectoderm region that includes the lens placode expressed 12 out of 19 possible Wnt ligands. When these activities were suppressed by conditional deletion of wntless (Le-cre; Wls^fl/fl) there were dramatic consequences that included a saucer-shaped optic cup, ventral coloboma, and a deficiency of periocular mesenchyme. This phenotype shared features with that produced when the Wnt/β-catenin pathway co-receptor Lrp6 is mutated or when retinoic acid (RA) signaling in the eye is compromised. Consistent with this, microarray and cell fate marker analysis identified a series of expression changes in genes known to be regulated by RA or by the Wnt/β-catenin pathway. Using pathway reporters, we showed that Wnt ligands from the surface ectoderm directly or indirectly elicit a Wnt/β-catenin response in retinal pigment epithelium (RPE) progenitors near the optic cup rim. In Le-cre; Wls^fl/fl mice, the numbers of RPE cells are reduced and this can explain, using the principle of the bimetallic strip, the curvature of the optic cup. These data thus establish a novel hypothesis to explain how differential cell numbers in a bilayered epithelium can lead to shape change.

Summary: During optic cup morphogenesis, Wnt ligands expressed in the surface ectoderm control cell proliferation in the retinal pigmented epithelium, and thus influence bending of the neural retina.

Bmp4 from the optic vesicle specifies murine retina formation

Previous studies of mouse embryos concluded that after the optic vesicle evaginates from the ventral forebrain and contacts the surface ectoderm, signals from the ectoderm specify the distal region of the optic vesicle to become retina and signals from the optic vesicle induce the lens. Germline deletion of Bmp4 resulted in failure of lens formation. We performed conditional deletion of Bmp4 from the optic vesicle to test the function of Bmp4 in murine eye development. The optic vesicle evaginated normally and contacted the surface ectoderm. Lens induction did not occur. The optic cup failed to form and the expression of retina-specific genes decreased markedly in the distal optic vesicle. Instead, cells in the prospective retina expressed genes characteristic of the retinal pigmented epithelium. We conclude that Bmp4 is required for retina specification in mice. In the absence of Bmp4, formation of the retinal pigmented epithelium is the default differentiation pathway of the optic vesicle. Differences in the signaling pathways required for specification of the retina and retinal pigmented epithelium in chicken and mouse embryos suggest major changes in signaling during the evolution of the vertebrate eye.

RPE specification in the chick is mediated by surface ectoderm-derived BMP and Wnt signalling

The retinal pigment epithelium (RPE) is indispensable for vertebrate eye development and vision. In the classical model of optic vesicle patterning, the surface ectoderm produces fibroblast growth factors (FGFs) that specify the neural retina (NR) distally, whereas TGFβ family members released from the proximal mesenchyme are involved in RPE specification. However, we previously proposed that bone morphogenetic proteins (BMPs) released from the surface ectoderm are essential for RPE specification in chick. We now show that the BMP- and Wnt-expressing surface ectoderm is required for RPE specification. We reveal that Wnt signalling from the overlying surface ectoderm is involved in restricting BMP-mediated RPE specification to the dorsal optic vesicle. Wnt2b is expressed in the dorsal surface ectoderm and subsequently in dorsal optic vesicle cells. Activation of Wnt signalling by implanting Wnt3a-soaked beads or inhibiting GSK3β at optic vesicle stages inhibits NR development and converts the entire optic vesicle into RPE. Surface ectoderm removal at early optic vesicle stages or inhibition of Wnt, but not Wnt/β-catenin, signalling prevents pigmentation and downregulates the RPE regulatory gene Mitf. Activation of BMP or Wnt signalling can replace the surface ectoderm to rescue MITF expression and optic cup formation. We provide evidence that BMPs and Wnts cooperate via a GSK3β-dependent but β-catenin-independent pathway at the level of pSmad to ensure RPE specification in dorsal optic vesicle cells. We propose a new dorsoventral model of optic vesicle patterning, whereby initially surface ectoderm-derived Wnt signalling directs dorsal optic vesicle cells to develop into RPE through a stabilising effect of BMP signalling.

Evaluation of the Optic Nerve Head in Glaucoma

Glaucoma is an optic neuropathy leading to changes in the intrapaillary and parapaillary regions of the optic disk. Despite technological advances, clinical identification of optic nerve head characteristics remains the first step in diagnosis. Careful examination of the disk parameters including size, shape, neuroretinal rim shape and pallor; size of the optic cup in relation to the area of the disk; configuration and depth of the optic cup; ratios of cup-to-disk diameter and cup-to-disk area; presence and location of splinter-shaped hemorrhages; occurrence, size, configuration, and location of parapapillary chorioretinal atrophy; and visibility of the retinal nerve fiber layer (RNFL) is important to differentiate between the glaucomatous and nonglaucomatous optic neuropathy. How to cite this article: Gandhi M, Dubey S. Evaluation of the Optic Nerve Head in Glaucoma. J Current Glau Prac 2013;7(3):106-114.

Shaping the eye from embryonic stem cells: Biological and medical implications

Organogenesis is regulated by a complex network of intrinsic cues, diffusible signals and cell/cell or cell/matrix interactions that drive the cells of a prospective organ to differentiate and collectively organize in three dimensions. Generating organs in vitro from embryonic stem (ES) cells may provide a simplified system to decipher how these processes are orchestrated in time and space within particular and between neighboring tissues. Recently, this field of stem cell research has also gained considerable interest for its potential applications in regenerative medicine. Among human pathologies for which stem cell-based therapy is foreseen as a promising therapeutic strategy are many retinal degenerative diseases, like retinitis pigmentosa and age-related macular degeneration. Over the last decade, progress has been made in producing ES-derived retinal cells in vitro, but engineering entire synthetic retinas was considered beyond reach. Recently however, major breakthroughs have been achieved with pioneer works describing the extraordinary self-organization of murine and human ES cells into a three dimensional structure highly resembling a retina. ES-derived retinal cells indeed assemble to form a cohesive neuroepithelial sheet that is endowed with the intrinsic capacity to recapitulate, outside an embryonic environment, the main steps of retinal morphogenesis as observed in vivo. This represents a tremendous advance that should help resolving fundamental questions related to retinogenesis. Here, we will discuss these studies, and the potential applications of such stem cell-based systems for regenerative medicine.