Ophthalmologic Imaging Research Articles

Outcomes of Adversarial Attacks on Deep Learning Models for Ophthalmology Imaging Domains

This study investigates whether adversarial attacks can confuse deep learning systems based on imaging domains.

Application of Artificial Intelligence in Targeting Retinal Diseases.

Retinal diseases affect an increasing number of patients worldwide because of the aging population. Request for diagnostic imaging in ophthalmology is ramping up, while the number of specialists keeps shrinking. Cutting-edge technology embedding artificial intelligence (AI) algorithms are thus advocated to help ophthalmologists perform their clinical tasks as well as to provide a source for the advancement of novel biomarkers. In particular, optical coherence tomography (OCT) evaluation of the retina can be augmented by algorithms based on machine learning and deep learning to early detect, qualitatively localize and quantitatively measure epi/intra/subretinal abnormalities or pathological features of macular or neural diseases. In this paper, we discuss the use of AI to facilitate efficacy and accuracy of retinal imaging in those diseases increasingly treated by intravitreal vascular endothelial growth factor (VEGF) inhibitors (i.e. anti-VEGF drugs), also including integration and interpretation features in the process. We review recent advances by AI in diabetic retinopathy, age-related macular degeneration, and retinopathy of prematurity that envision a potentially key role of highly automated systems in screening, early diagnosis, grading and individualized therapy. We discuss benefits and critical aspects of automating the evaluation of disease activity, recurrences, the timing of retreatment and therapeutically potential novel targets in ophthalmology. The impact of massive employment of AI to optimize clinical assistance and encourage tailored therapies for distinct patterns of retinal diseases is also discussed.

Peripapillary microperimetry for the diagnosis and follow-up of papilledema in cases treated for idiopathic intracranial hypertension

ABSTRACT Objectives To evaluate the use of peripapillary microperimetry in the diagnosis and follow-up of medically and surgically treated cases of papilledema due to idiopathic intracranial hypertension (IIH). Methods This study was a prospective non-comparative observational case series of patients with IIH. All patients underwent full ophthalmological examination, radiological imaging, and lumbar puncture, as well as microperimetric measurements around the optic nerve head. Patients were classified into a medical group, treated by weight reduction and acetazolamide, and a surgical group, treated by optic nerve sheath fenestration. Peripapillary microperimetric examinations were done at presentation and after one-month of treatment. Findings were compared to values obtained from an age-matched healthy control group. Results The study included 78 eyes of 39 patients diagnosed with IIH and 32 eyes of 16 healthy control subjects. The average of the sum of the 12 peripapillary microperimetric thresholds for both eyes of IIH cases before treatment was 53.4±59.9 dB, while for controls it was 171±14.0 dB (p<0.001). After 1 month of treatment, it improved to 86.7±56.1 dB (p<0.001) in IIH cases. Analyzed separately, the medical group improved from 68.9±43.4 dB to 105.2±76.5 dB (p<0.001), while the surgical group improved from 13.5±21.9 dB to 54.9±35.5 dB (p<0.001). Several potential roles for peripapillary microperimetry use in IIH management were identified including confirming papilledema diagnosis, follow up of treatment, and selection of cases for surgery. Conclusion Microperimetry can accurately and quantitatively monitor changes in peripapillary retinal sensitivity in patients undergoing treatment for papilledema due to IIH.

An update on the diagnosis and management of ocular sarcoidosis.

Ocular sarcoidosis is one of the most common causes of uveitis worldwide. The diagnosis and treatment of patients with ocular sarcoidosis remains challenging in some cases. It is important for clinicians to keep up to date with new diagnostic and treatment tools for this disease. The International Workshop on Ocular Sarcoidosis diagnostic criteria were first proposed in 2009 and revised in 2017. The new criteria contained two parts: ocular presentation and systemic investigation. The diagnostic value of liver enzymes was reduced in the new criteria, whereas the value placed of lymphopenia and the CD4/CD8 ratio in bronchoalveolar lavage fluid were increased. Despite not being included in the criteria, recent studies have also highlighted the diagnostic value of serum soluble interleukin-2 receptors. Recent ophthalmologic imaging also provides useful insights for the differential diagnosis.Many new treatments for ocular sarcoidosis have been developed in recent years. The introduction of biological immunomodulatory agents for uveitis treatment represents a big improvement. Antitumor necrosis factor-alpha antibodies, including adalimumab, have been proven to be effective for treating ocular sarcoidosis. Many studies have also suggested that other biological agents could be effective and well tolerated. Newer intravitreal dexamethasone and fluocinolone implants have been developed. Patients treated with these implants have experienced good and sustained control of their intraocular inflammation. Diagnosis and treatment options for ocular sarcoidosis have changed over time. However, challenges still exist in some difficult patients. Future studies should focus on finding more sensitive biomarkers and developing more effective immunomodulatory treatments with longer efficacy and less side effects.

3-Dimensional magnetic resonance imaging of the freely moving human eye.

Eye motion is a major confound for magnetic resonance imaging (MRI) in neuroscience or ophthalmology. Currently, solutions toward eye stabilisation include participants fixating or administration of paralytics/anaesthetics. We developed a novel MRI protocol for acquiring 3-dimensional images while the eye freely moves. Eye motion serves as the basis for image reconstruction, rather than an impediment. We fully reconstruct videos of the moving eye and head. We quantitatively validate data quality with millimetre resolution in two ways for individual participants. First, eye position based on reconstructed images correlated with simultaneous eye-tracking. Second, the reconstructed images preserve anatomical properties; the eye's axial length measured from MRI images matched that obtained with ocular biometry. The technique operates on a standard clinical setup, without necessitating specialized hardware, facilitating wide deployment. In clinical practice, we anticipate that this may help reduce burdens on both patients and infrastructure, by integrating multiple varieties of assessments into a single comprehensive session. More generally, our protocol is a harbinger for removing the necessity of fixation, thereby opening new opportunities for ethologically-valid, naturalistic paradigms, the inclusion of populations typically unable to stably fixate, and increased translational research such as in awake animals whose eye movements constitute an accessible behavioural readout.

Concurrent tuberculous chorioretinitis with choroidal neovascularization and tuberculous meningitis: a case report

BackgroundTuberculosis (TB) remains a severe health burden worldwide. The manifestation of concurrent tuberculous cerebral and ocular involvements associated with TB is uncommon.Case presentationWe report a 17-year-old girl with concurrent tuberculous cerebral and ocular involvements and visual impairment due to choroidal neovascularization. This study emphasizes the definite diagnosis with the combination of ophthalmological examination, multimodal imaging and routine tuberculosis testing, and the proper management with intravitreal anti-VEGF injection accompanied by systemic anti-tuberculosis therapy.ConclusionCombined applications of routine TB tests, fundus multimodal imaging and diagnostic therapy greatly help the clinician to establish a precise diagnosis and in monitoring the therapeutic response.

Teleophthalmology Using Whole Slide Imaging (WSI): Clinical-Pathological Correlation Even Without Own Histological Laboratory

Telepathology has experienced enormous growth over the last two decades due to significant advances in IT and has led to the introduction of innovative solutions for digital imaging in ophthalmology. Whole Slide Imaging (WSI), where conventional glass slides are fully scanned, offers a wide range of applications, including remote (conciliary) diagnostics, second opinions, (remote) teaching and research and presentation. In addition to qualified and experienced personnel, a comparatively expensive basic equipment has so far also been an essential requirement. The focus ophthalmology at the Eye Centre in Freiburg is able to involve a de facto unlimited number of users and consulting partners (also overseas) in the digital microscopy of identical sections independent of location, device and time. This is done simply by using the internet. This procedure has proven to be very successful and opens up completely new avenues, not only for correct diagnosis (clinical-pathological correlation), but also for education and training. Preparation and distribution of further histological slides in addition to the scanned slide is not required for small specimens with scant material, for example, nor is it necessary to purchase microscopes for the participants if funds are lacking or the infrastructure is unsuitable. When dealing with transferring data, it is essential to comply with data protection regulations.

Functional Imaging of the Ocular Fundus Using an 8-Band Retinal Multispectral Imaging System

Application of functional imaging in ophthalmology requires efficient imaging techniques that can detect and quantify chromophores to visualise processes in vivo. The aim of the present study was to develop and evaluate a fast and affordable imaging system. We describe an eight-band retinal multispectral imaging (MSI) system and compare it with a hyperspectral imaging (HSI) device. Determination of blood oxygen saturation was studied as proof of principle. Reflectance of incident light is measured as 1/absorbance at different wavelengths between 440 nm and 580 nm. Both devices have incorporated optical bandpass filters in a mydriatic fundus camera. The MSI system scans the retina at eight pre-defined wavelengths specific for the spectrum of haemoglobin. The HSI system acquires a full scan from 480 to 720 nm in 5 nm steps. A simple assessment of the ratio between the absorbance peaks of oxygenated haemoglobin (HbO2) and reduced haemoglobin (HbR) was not suitable for generating validated oxygenation maps of the retina. However, a correction algorithm that compares the measured reflectance with reflectance spectra of fully oxygenated and fully deoxygenated blood allowed our MSI setup to estimate relative oxygen saturation at higher levels, but underestimated relative oxygen saturation at lower levels. The MSI device generated better quality images than the HSI device. It allows customisation with filter sets optimised for other chromophores of interest, and augmented with extrinsic contrast imaging agents, it has the potential for a wider range of ophthalmic molecular imaging applications.

Diagnostic Value of Systematic Imaging Examination in Embedded Optic Disc Drusen in Adolescents with Mild Visual Impairment.

Aim To evaluate the diagnostic value of systematic ophthalmologic imaging examination in the diagnosis of embedded optic disc drusen (ODD) in adolescents with mild visual impairment. Methods Eleven patients were evaluated through optometric examination, fundus photography, visual field inspection, optical coherence tomography (OCT), ultrasonography (US), and fundus fluorescein angiography (FFA). Of the 11 patients, three also underwent cranial and orbital magnetic resonance imaging (MRI). Results All 11 patients had either no apparent abnormality or only mild refractive abnormalities. In all patients, fundus inspection revealed flushing the optic disc with varying degrees of limited boundary ambiguity and optic disc congestion with disappearance of the fovea. One patient had a visual field defect during the period of edema of ODD, but the visual field returned to normal after the optic disc edema subsided. US revealed discoid acousto-optic masses in front of the optic disc in six patients. OCT showed a slight elevation and thinning of the retinal nerve fiber layer (RNFL) of the optic disc in all patients. Quasicircular, hyperreflex signals of different sizes could be observed below the RNFL. Late-stage FFA revealed focal staining at the edge of the optic disc without fluorescence leakage in all patients. Orbital and cranial MRI findings were normal in the three patients. Conclusion A systematic ophthalmologic imaging examination can not only improve the detection rate of embedded ODD but also avoid excessive examinations and treatments.

Radiochemistry: A Useful Tool in the Ophthalmic Drug Discovery.

Positron Emission Tomography (PET) and Single Photon Emission Computerized Tomography (SPECT) are ultra-sensitive, fully translational and minimally invasive nuclear imaging techniques capable of tracing the spatiotemporal distribution of positron (PET) or gamma (SPECT) emitter-labeled molecules after administration into a living organism. Besides their impact in the clinical diagnostic, PET and SPECT are playing an increasing role in the process of drug development, both during the evaluation of the pharmacokinetic properties of new chemical entities as well as in the proof of concept, proof of mechanism and proof of efficacy studies. However, they have been scarcely applied in the context of ophthalmic drugs. In this paper, the basics of nuclear imaging and radiochemistry are briefly discussed, and the few examples of the use of these imaging modalities in ophthalmic drug development reported in the literature are presented and discussed. Finally, in a purely theoretical exercise, some labeling strategies that could be applied to the preparation of selected ophthalmic drugs are proposed and potential applications of nuclear imaging in ophthalmology are projected.

Potential of methods of artificial intelligence for quality assurance

Procedures with artificial intelligence (AI), such as deep neural networks, show promising results in automatic analysis of ophthalmological imaging data. This article discusses to what extent the application of AI algorithms can contribute to quality assurance in the field of ophthalmology. Relevant aspects from the literature are discussed. Systems based on artificial deep neural networks achieve remarkable results in the diagnostics of eye diseases, such as diabetic retinopathy and are very helpful, for example by segmenting optical coherence tomographic (OCT) images and detecting lesion components with high fidelity. To train these algorithms large data sets are required. The quality and availability of such data sets determine the continuous improvement of the algorithms. The comparison between the AI algorithms and physicians for image interpretation has also enabled insights into the diagnostic concordance between physicians. Current challenges include the development of methods for modelling decision uncertainty and improved interpretability of automated diagnostic decisions. Systems based on AI can support decision making for physicians and thereby contribute to a more efficient quality assurance.

A novel artificial neural network-based algorithm for detecting vascular occlusive diseases in retinal images

Introduction: Central retinal vascular occlusions cause sudden painless complete blindness of one eye or partial visual loss, depending on the retinal vessel involved. These acute conditions may occur due to thrombosis, diabetic retinopathy, hypertensive retinopathy, glaucoma or due to other systemic conditions like connective tissue disorders. Central retinal artery occlusion (CRAO) is considered analogous to a cerebral stroke and it is an ophthalmological emergency. Central retinal vein occlusion (CRVO) is one of the most common causes of visual loss worldwide and the second most common cause of blindness due to retinal vascular disorders after diabetic retinopathy. Artificial intelligence (AI) based methodologies have been used in automated diagnosis support systems in detecting many ophthalmological conditions like diabetic retinopathy, retinal micro aneurisms, and retinal haemorrhages. Methods: The developed algorithm consists of modules for, image pre-processing, extraction of the areas of interest, feature extraction, feature presentation and image classification. A feed-forward back-propagation multilayer artificial neural network (ANN), with 27 input layer neurones, three output layer neurones and 81 first hidden layer and nine second hidden layer neurones, was used as the image classifier. This methodology was trained with 90 retinal images. Results: The percentage accuracy of the developed ANN-based ophthalmological image classification algorithm is 97.8% when it is trained to diagnose CRAO and CRVO. The sensitivity is 98.3%, and the specificity is 100%. Conclusions: This study presents the first AI-based approach in automatic detection of both CRAO and CRVO. Once this algorithm is trained to minimise false negatives, it can be used as an automated diagnosis support system in ophthalmological screening programs.

Smartphone-based Gonio-Imaging: A Novel Addition to Glaucoma Screening Tools.

Gonioscopy is one of the most difficult skill acquired by ophthalmologists. It is one of the key components of glaucoma diagnosis and management. Smartphone based imaging solutions have revolutionised the imaging in ophthalmology by virtue of their ease of use and portability. Authors hereby describe a smartphone based gonioimaging technique with a 4-mirror goniolens, which can detect the gross findings in anterior chamber angle, and thus can be utilised as screening tool for glaucoma.

Using Anisotropic Diffusion in the Multiscale Ridge Detection Method

A ridge detection algorithm is proposed for tracing blood vessels on images of the ocular fundus. Multiscale non-maximum suppression is applied to the image Laplacian. The multiscale algorithm exploits the pyramidal fine structure similarly to the SIFT method. Anisotropic diffusion is used in preprocessing, which makes it possible to boost the value of the convolution of the Laplacian with the Gaussian on ridge structures. The proposed algorithm has been tested on the ophthalmological image database DRIVE. The proposed preprocessing has substantially improved the ridge detection quality.

Guidelines for ultrasound examination in ophthalmology. Part III: Color Doppler ultrasonography.

The aim of this article was to present the possibilities of use and application of color-coded Doppler ultrasonography in the diagnosis of various diseases of the eyeball and orbit which result from vascular disorders. Color-coded Doppler ultrasonography is recommended for the assessment of blood flow velocity in the retrobulbar arteries. That is why the article contains current recommendations for Doppler imaging in ophthalmology. The paper provides detailed recommendations for patient’s preparation for the examination, presents the scanning technique and safety of the examination, and lists ophthalmological diseases of vascular origin for which color-coded Doppler ultrasonography can be applied. Furthermore, the article also presents other techniques applied in clinical practice for the assessment of blood flow or imaging of vasculature of a given eyeball structure, inter alia: power Doppler ultrasonography, 3D and 4D ultrasonography, magnetic resonance angiography, spiral computer tomography, transcranial ultrasonography and modern microvascular imaging. The authors emphasize the usefulness of color-coded Doppler ultrasonography in the diagnosis of diseases which result from blood flow disorders within the eyeball, such as amaurosis fugax, ocular ischemic syndrome, insufficiency in vessels supplying the carotid and vertebral arteries, posterior ischemic optic neuropathy, glaucoma, age-related macular degeneration, vascular vision disorders, vascular malformations, such as arteriovenous fistula, orbital varices, systemic connective tissue diseases in retinopathy of prematurity, diabetes, thyroid disorders or strabismus. The application of color-coded Doppler ultrasonography is especially important in the assessment of the vasculature of intrabulbar tumorous lesions and in the differential diagnosis of intrabulbar tumors.

Applications of Manganese-Enhanced Magnetic Resonance Imaging in Ophthalmology and Visual Neuroscience.

Understanding the mechanisms of vision in health and disease requires knowledge of the anatomy and physiology of the eye and the neural pathways relevant to visual perception. As such, development of imaging techniques for the visual system is crucial for unveiling the neural basis of visual function or impairment. Magnetic resonance imaging (MRI) offers non-invasive probing of the structure and function of the neural circuits without depth limitation, and can help identify abnormalities in brain tissues in vivo. Among the advanced MRI techniques, manganese-enhanced MRI (MEMRI) involves the use of active manganese contrast agents that positively enhance brain tissue signals in T1-weighted imaging with respect to the levels of connectivity and activity. Depending on the routes of administration, accumulation of manganese ions in the eye and the visual pathways can be attributed to systemic distribution or their local transport across axons in an anterograde fashion, entering the neurons through voltage-gated calcium channels. The use of the paramagnetic manganese contrast in MRI has a wide range of applications in the visual system from imaging neurodevelopment to assessing and monitoring neurodegeneration, neuroplasticity, neuroprotection, and neuroregeneration. In this review, we present four major domains of scientific inquiry where MEMRI can be put to imperative use — deciphering neuroarchitecture, tracing neuronal tracts, detecting neuronal activity, and identifying or differentiating glial activity. We deliberate upon each category studies that have successfully employed MEMRI to examine the visual system, including the delivery protocols, spatiotemporal characteristics, and biophysical interpretation. Based on this literature, we have identified some critical challenges in the field in terms of toxicity, and sensitivity and specificity of manganese enhancement. We also discuss the pitfalls and alternatives of MEMRI which will provide new avenues to explore in the future.

High-Resolution SD-OCT and EDI-OCT in the Evaluation and Management of Multifocal Serpiginoed Choroditis

Purpose To describe spectral domain optical coherence tomography (SD-OCT) and enhanced depth image OCT (EDI-OCT) findings of multifocal serpiginoid choroditis (MSC) , including affected layer of retinal involvement, changes at the vitreoretinal interface, and response to therapy. Methods A retrospective review of 20 eyes (14 patients) with MSC. Each patient underwent a complete ophthalmologic examination, fundus photography, fundus autoflorecence (FAF) and OCT imaging of the affected retina at the initial visit and on each follow-up. Results In acute stage, SD-OCT showed hyperreflective areas involving the outer retinal layers which include retinal pigment epithelium (RPE), photoreceptor outer segment tips (POST), inner segment–outer segment (IS/OS) junction, external limiting membrane (ELM), and outer nuclear layer (ONL) with choroidal and intraretinal layer cells infiltrate. EDI-OCT showed increase choroidal thickness. As the lesions began to heal, irregular, knobby elevations of outer retinal layers appeared (RPE, POST, IS/OS junction, and ELM could not be distinguished) with significant decrease in choroidal and intraretinal cells. On complete healing, loss of RPE, POST, IS/OS junction, and ELM in SD-OCT scan and absent of the choroidal and intraretinal cells and continous hyperreflactivity of the choroid (increased penetrance). Conclusion SD-OCT and EDI-OCT provides high-resolution detail regarding ultrastructural changes in vitreoretinal interface, outer retina and choroid during the course of the lesion. Serial SD-OCT and EDI-OCT also provides further insight into response to therapy by observing choroidal and intraretinal cells.

Automatic segmentation of the foveal avascular zone in ophthalmological OCT-A images.

Angiography by Optical Coherence Tomography (OCT-A) is a non-invasive retinal imaging modality of recent appearance that allows the visualization of the vascular structure at predefined depths based on the detection of the blood movement through the retinal vasculature. In this way, OCT-A images constitute a suitable scenario to analyze the retinal vascular properties of regions of interest as is the case of the macular area, measuring the characteristics of the foveal vascular and avascular zones. Extracted parameters of this region can be used as prognostic factors that determine if the patient suffers from certain pathologies (such as diabetic retinopathy or retinal vein occlusion, among others), indicating the associated pathological degree. The manual extraction of these biomedical parameters is a long, tedious and subjective process, introducing a significant intra and inter-expert variability, which penalizes the utility of the measurements. In addition, the absence of tools that automatically facilitate these calculations encourages the creation of computer-aided diagnosis frameworks that ease the doctor’s work, increasing their productivity and making viable the use of this type of vascular biomarkers. In this work we propose a fully automatic system that identifies and precisely segments the region of the foveal avascular zone (FAZ) using a novel ophthalmological image modality as is OCT-A. The system combines different image processing techniques to firstly identify the region where the FAZ is contained and, secondly, proceed with the extraction of its precise contour. The system was validated using a representative set of 213 healthy and diabetic OCT-A images, providing accurate results with the best correlation with the manual measurements of two experts clinician of 0.93 as well as a Jaccard’s index of 0.82 of the best experimental case in the experiments with healthy OCT-A images. The method also provided satisfactory results in diabetic OCT-A images, with a best correlation coefficient with the manual labeling of an expert clinician of 0.93 and a Jaccard’s index of 0.83. This tool provides an accurate FAZ measurement with the desired objectivity and reproducibility, being very useful for the analysis of relevant vascular diseases through the study of the retinal micro-circulation.

Artificial Intelligence and Optical Coherence Tomography Imaging.

This review article aimed to highlight the application and use of artificial intelligence (AI) in optical coherence tomography (OCT) imaging in ophthalmology. Artificial intelligence programs seek to simulate intelligent human behavior in computers. With an abundance of patient data, especially with the advent and growing use of imaging modalities such as OCT, AI programs provide us with the unique opportunity to analyze this plethora of information and assist in making clinical decisions in the field of ophthalmology. Groups around the world have developed and evaluated AI programs that gather data from diagnostic modalities, such as OCT, that assist in the diagnosis and management of ophthalmological diseases with a high accuracy. Artificial intelligence programs using OCT have the potential to play a significant role in the diagnosis and management of ophthalmological disease in the near future. Incorporation of AI in medicine, however, is not without its pitfalls. Some limitations of AI in ophthalmology are also discussed in this review. These include the deskilling of physicians due to increase in reliance on automation, inability of AI programs to take a holistic approach to clinical encounters with patients, requirement of pre-existing strong datasets to train AI programs, and the inability of AI programs to incorporate the ambiguity and variability that is intrinsic to the nature of clinical medicine.

Automatic wide field registration and mosaicking of OCTA images using vascularity information

Optical Coherence Tomography Angiography (OCTA) constitutes a novel ophthalmological image modality that is characterized for being a non-invasive capture technique that allows a profound analysis of the vascular characteristics of the eye fundus. Given the restricted field of view of the eye fundus that offers each scan, the specialists frequently capture several complementary images that may be simultaneously analyzed to offer a complete and accurate diagnosis of the patient.In this work, we propose a fully automatic method to register complementary OCTA images and provide compositions for the same patient, generating a wide field of representation that allows a simpler and more direct analysis than the traditional tedious manual procedures. To achieve this, we based our proposal in a robust combination of representative features that are filtered by an accurate identification of the main retinal vasculature. This way, given the characteristic high irregularity in the fundus of the OCTA images, we avoid many variable areas that may interfere in the registration process, restricting the analysis to the most representative and stable structure of this image modality, the main retinal vasculature. In particular, we use Speeded-Up Robust Features (SURF) algorithm to extract representative features in the main vascular region that is extracted using a method that combines the analysis of the Hessian matrix followed by an hysteresis threshold process. Then, using a K-NN model, we perform the registration of the resulting features from the different OCTA images to be analyzed. Finally, the Random sample consensus (RANSAC) method is exploited to produce the final target mosaic. The proposed method presented satisfactory results in the validation experiments, with accurate values for the MSE index of 1.2566 and 1.6725 pixels for the registration of paired images an mosaics, respectively.