Ocular Fundus Images Research Articles

Characterization of the retinal changes of the 3\xd7Tg-AD mouse model of Alzheimer\u2019s disease

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder whose diagnosis remains a notable challenge. The literature suggests that cerebral changes precede AD symptoms by over two decades, implying a significantly advanced stage of AD by the time it is usually diagnosed. In the study herein, texture analysis was applied to computed optical coherence tomography ocular fundus images to identify differences between a group of the transgenic mouse model of the Alzheimer’s disease (3×Tg-AD) and a group of wild-type mice, at the ages of one and two-months-old. A substantial difference between groups was found at both time-points across all neuroretina’s layers. Here, the inner nuclear layer stands out both in the level of statistically significant differences and on the extension of these differences which span through the imaged area. Also, the progression of AD is suggested to be spotted by texture analysis as demonstrated by the significant difference found in the inner plexiform and the outer nuclear layers from the age of one to the age of two-months-old. These findings demonstrate the potential of the use of the retina and texture analysis to the diagnosis of AD and monitor AD progression. Besides, the differences between groups found in this study suggest that the 3×Tg-AD model may be inappropriate to study early changes associated with the AD and other animal models should be tested following the same path and rationale. Moreover, these results also suggest that the human genes present in these transgenic mice may have an impact on the neurodevelopment of offspring which would justify the significant changes found at the age of one-month-old.

Retina–electrode interface properties and vision restoration by two generations of retinal prostheses in one patient—one in each eye

Objective. The Argus I implant is a first-generation epiretinal prosthesis approved for an investigational clinical trial in the U.S. Its successor, the Argus II implant, has a higher electrode density for increased spatial resolution and covers a larger retinal area to accommodate a wider visual angle. Both generations of Argus restored some vision to end-stage RP patients, but it remains unclear how the increased electrode count affected the visual percepts. Here we report a study of the first person on earth with two ‘bionic eyes’, with an Argus I implanted in one eye and Argus II in the other, to compare the retina–electrode interface and the visual outcome of the two devices. Approach. The retina–electrode interface was examined by electrode impedance, perceptual threshold, and ophthalmic images such as ocular coherence tomography data and fundus imaging. The subject’s visual outcomes were evaluated by computer-based visual function tests and subjective feedback. Main results. The electrode impedance of both Argus I and II slowly decreased overtime after implantation, accompanied by a gradual increase in the perceptual threshold. A quantitative analysis of the impedance and retina–electrode distance revealed somewhat different causes of impedance change in Argus I vs. II. Evaluation of the visual functions restored and feedback from the subject suggest that the Argus II device enables improved spatial visual ability over Argus I, but adaptation to prosthetic vision did not lead to a measurable performance improvement in the standard visual function tests. Significance. This study of Argus I and II in the same subject directly compares for the first time the interface properties and prosthetic vision in two eyes that share the same disease mechanism and converged visual pathway in higher visual centers, offering exciting new insights into the influence of the electrode parameters and layout to prosthetic vision.Trial registration: Data collected within clinical trials registered at NIH (NCT00279500 & NCT01860092).

Artificial intelligence for detection of optic disc abnormalities.

The aim of this review is to highlight novel artificial intelligence-based methods for the detection of optic disc abnormalities, with particular focus on neurology and neuro-ophthalmology. Methods for detection of optic disc abnormalities on retinal fundus images have evolved considerably over the last few years, from classical ophthalmoscopy to artificial intelligence-based identification methods being applied to retinal imaging with the aim of predicting sight and life-threatening complications of underlying brain or optic nerve conditions. Artificial intelligence and in particular newly developed deep-learning systems are playing an increasingly important role for the detection and classification of acquired neuro-ophthalmic optic disc abnormalities on ocular fundus images. The implementation of automatic deep-learning methods for detection of abnormal optic discs, coupled with innovative hardware solutions for fundus imaging, could revolutionize the practice of neurologists and other non-ophthalmic healthcare providers.

Multivariate Regression-Based Convolutional Neural Network Model for Fundus Image Quality Assessment

Objectively assessing the perceptual quality of an ocular fundus image is essential for the reliable diagnosis of various ocular diseases. A fair amount of work has been done in this field to date. However, the generalizability of the current work is limited, as the existing quality models were developed and evaluated with data-sets built with limited subjective inputs. This paper aims at addressing this limitation with the following two contributions. First, a new fundus image quality assessment (FIQuA) data-set is presented, containing 1500 fundus images with three classes of quality: Good, Fair, and Poor. Also, for each image, subjective scores (in the range [0-10]) were collected for six quality parameters, including structural and generic properties of the fundus images. Second, a new multivariate regression based convolutional neural network (CNN) model is proposed to predict the fundus image quality. The proposed model consists of two individually trained blocks. The first block consists of four pre-trained models, trained against the subjective scores for the six quality parameters, and aims at deriving the optimized features for classification. Next, the optimized features from each of the four models are ensembled together and transferred to the second block for final classification. The proposed model achieves a strong correlation with the subjective scores, with the values 0.941 , 0.954 , 0.853 , and 0.401 obtained for SROCC, LCC, KCC, and RMSE respectively. Its classification accuracy is 95.66% over the FIQuA data-set, and 98.96% and 88.43% respectively over the two publicly available data-sets DRIMDB and EyeQ.

Scanning laser ophthalmoscopy – basic optical principles

AbstractScanning Laser Ophthalmoscopy (SLO) produces an ocular fundus image by moving a focused laser beam via scanning mirrors over the retina in a grid pattern and registering the reflected light from each scanned point. In confocal SLO, a pinhole is placed in front of the detector to cut off scattered or defocused light coming from outside the point of interest, which otherwise can blur the image. This results in a focused, high‐contrast image of a single tissue layer located at the focal plane. Tomographic information can be extracted by moving the plane of the pinhole. The use of various wavelengths allows for different applications, such as fluorescein angiography, indocyanine green angiography, and autofluorescence imaging. This lecture will present the basic principles of SLO, as well as discuss notable applications and variants of the technology.

A multi-context CNN ensemble for small lesion detection.

In this paper, we propose a novel method for the detection of small lesions in digital medical images. Our approach is based on a multi-context ensemble of convolutional neural networks (CNNs), aiming at learning different levels of image spatial context and improving detection performance. The main innovation behind the proposed method is the use of multiple-depth CNNs, individually trained on image patches of different dimensions and then combined together. In this way, the final ensemble is able to find and locate abnormalities on the images by exploiting both the local features and the surrounding context of a lesion. Experiments were focused on two well-known medical detection problems that have been recently faced with CNNs: microcalcification detection on full-field digital mammograms and microaneurysm detection on ocular fundus images. To this end, we used two publicly available datasets, INbreast and E-ophtha. Statistically significantly better detection performance were obtained by the proposed ensemble with respect to other approaches in the literature, demonstrating its effectiveness in the detection of small abnormalities.

Activation of transient receptor potential vanilloid 4 channels dilates rat retinal arterioles through nitric oxide- and BKCa channel-dependent mechanisms in vivo.

Transient receptor potential vanilloid 4 (TRPV4) channel, a cation channel expressed in nearly all cell types, plays an important role in the regulation of vascular tone. In the present study, we examined the effect of GSK1016790A, an activator of TRPV4 channels, on the diameter of retinal blood vessels in rats and the underlying mechanisms. Ocular fundus images were captured with an original high-resolution digital fundus camera in vivo and diameters of retinal blood vessels were measured. Intravenous infusion of GSK1016790A (0.2-2μgkg-1min-1) increased retinal arteriolar diameter in a dose-dependent manner. The higher dose of GSK1016790A (2μgkg-1min-1) slightly decreased blood pressure. These responses to GSK1016790A were significantly attenuated by intravenous injection of GSK2193874 (0.3mg/kg), an antagonist of TRPV4 channels. Intravitreal injection of Nω-nitro-L-arginine methyl ester, an inhibitor of nitric oxide (NO) synthase or iberiotoxin, an inhibitor of large-conductance Ca2+-activated K+ (BKCa) channel, significantly attenuated the GSK1016790A-induced increases in retinal arteriolar diameter. These results suggest that activation of TRPV4 channels dilates rat retinal arterioles through NO- and BKCa channel-dependent mechanisms in vivo.

Morphological changes of focal choroidal excavation.

To investigate morphological changes of focal choroidal excavation (FCE). The medical records of 18 FCE patients (25 eyes) were retrospectively reviewed. Clinical features as well as images of ocular fundus, fundus fluorescein angiography (FFA), indocyanine green angiography (ICGA), and spectral domain optical coherence tomography (OCT) were analyzed. OCT images were employed as a major criterion to evaluate the changes which occurred during the follow-up. Of the 18 patients, 9 were male and 9 were female. The mean age was 41 ± 13.5years. The follow-up period was 38.5 ± 24.7 (7-92) months. Thirteen patients were myopic. FCE appeared from slight pigment irregularity to yellowish white plaque. FFA revealed window defects corresponding to the lesions. Early-phase ICGA images showed hypofluorescence and medium to large choroidal blood vessels. Plaque- or dot-like hyperfluorescence was seen on late-phase images. OCT scans showed that 11 patients had unilateral FCE and 7 patients were bilaterally affected. Twenty-three eyes had a single FCE and the other two eyes had 2 excavations. FCE presented as confirming type in 21 eyes and non-confirming type in 3 eyes. Discontinuity of the interdigitation zone, ellipsoid zone, and retinal pigment epithelium/Bruch's complex zone was found. The average subfoveal choroidal thickness was 312.06 ± 66.12μm in FCE eyes. During the follow-up, choroidal neovascularization developed in 4 patients and retinal pigment epithelium discontinuity was aggravated in one eye. No remarkable changes were found in the other eyes on B scan of the OCT images, but the volume of FCE might have changed a little. FCE, both confirming type and non-confirming type, may keep stable during a long period of time. Choroidal neovascularization may develop at the site of FCE.

Vessel Optimal Transport for Automated Alignment of Retinal Fundus Images.

Optimal transport has emerged as a promising and useful tool for supporting modern image processing applications such as medical imaging and scientific visualization. Indeed, the optimal transport theory enables great flexibility in modeling problems related to image registration, as different optimization resources can be successfully used as well as the choice of suitable matching models to align the images. In this paper, we introduce an automated framework for fundus image registration which unifies optimal transport theory, image processing tools, and graph matching schemes into a functional and concise methodology. Given two ocular fundus images, we construct representative graphs which embed in their structures spatial and topological information from the eye's blood vessels. The graphs produced are then used as input by our optimal transport model in order to establish a correspondence between their sets of nodes. Finally, geometric transformations are performed between the images so as to accomplish the registration task properly. Our formulation relies on the solid mathematical foundation of optimal transport as a constrained optimization problem, being also robust when dealing with outliers created during the matching stage. We demonstrate the accuracy and effectiveness of the present framework throughout a comprehensive set of qualitative and quantitative comparisons against several influential state-of-the-art methods on various fundus image databases.

Current status and future trends of clinical diagnoses via image-based deep learning.

With the recent developments in deep learning technologies, artificial intelligence (AI) has gradually been transformed from cutting-edge technology into practical applications. AI plays an important role in disease diagnosis and treatment, health management, drug research and development, and precision medicine. Interdisciplinary collaborations will be crucial to develop new AI algorithms for medical applications. In this paper, we review the basic workflow for building an AI model, identify publicly available databases of ocular fundus images, and summarize over 60 papers contributing to the field of AI development.

Ability of MultiColor scanning laser ophthalmoscope to detect non-glaucomatous retinal nerve fiber layer defects in eyes with retinal diseases

PurposeTo compare the ability of ocular fundus images obtained by Spectralis MultiColor scanning laser ophthalmoscope (MC-SLO) to that obtained by conventional color fundus images (CF) in detecting non-glaucomatous nerve fiber layer defects (NFLDs).MethodsA cross-sectional, retrospective study. Patients with retinal diseases who had ocular examination with both the MC-SLO and CF instruments at the Kagoshima University from December 2016 to February 2017 were studied. Eyes that had NFLDs with non-glaucomatous optic discs were analyzed. The visibility of the NFLDs was classified into three grades: grade 0, not visible; grade 1, barely visible; and grade 2, clearly visible. The NFLD grade for blue, green, and red scanning lights of the MC-SLO, merged images with three wavelengths and the color and red-free images were determined by two ophthalmologists. These scores were compared by Steel-Dwass tests.ResultsThirty-one eyes of 26 patients with a mean age of 63.1 ± 11.2 years were studied. There were 14 eyes with diabetic retinopathy, 11 eyes with age-related macular degeneration, 3 eyes with a branch retinal vein occlusion, and 3 eyes with an epiretinal membrane/macular hole. Both the intra-rater (0.631–0.790) and inter-rater (0.637–0.733) agreements were good. NFLDs were detected by the blue wavelength in all cases and by green wavelength and merged wavelengths in 90.3% of the images. The mean NFLD grade was 1.58 ± 0.49 for blue light images, 1.13 ± 0.54 for green light images, 0.07 ± 0.24 for red light images, and 1.16 ± 0.56 for merged images. The NFLD score for blue wavelength was significantly higher than that for green and red wavelength images (P < 0.05 and P < 0.01) but not significantly higher than that for the merged images. NFLDs were detected in 12 eyes (38.7%) in the color images and 16 eyes (51.6%) in the red-free images. The NFLD score for the CF and the red-free image was 0.41 ± 0.55 and 0.70 ± 0.67 which is significantly lower than that of blue MC-SLO images.ConclusionThe images obtained by MC-SLO are superior to that obtained by CF in detecting NFLDs in eyes with retinal diseases. We recommend MC-SLO imaging to screen for NFLDs in eyes with retinal diseases.

Application of remote screening system for diabetic retinopathy in Beijing Desheng Community

Objective To develop a remote diabetic retinopathy (DR) screening system and to evaluate the effectiveness of the screening system in community. Methods A cross-sectional study was carried out under the informed consent of subjects in Peking University People's Hospital and Beijing Xicheng District Desheng Community Health Service Center from June 2015 to December 2016.A remote DR screening system was established in Peking University People's Hospital and Beijing Xicheng District Desheng Community Health Service Center during June 2015 to December 2016.Based on non-mydriatic digital eye fundus camera photography and the internet transmission technology, anterior ocular segment and fundus images of 2 473 eyes from 1 355 community subjects with type 2 diabetes mellitus were transmitted from Beijing Xicheng District Desheng Community Health Service Center to the reading center of Peking University People's Hospital, and the results were provided to the subjects after analysis, including visual examination, diagnosis and follow-up rate of the subjects, the agreement between remote screening system and conventional screening method was analyzed and compared. Results The visual acuities of the 2 473 eyes of 1 355 subjects were obtained by trained community physician, and the visual acuity was ≤0.05 in 103 eyes (4.2%), >0.05-0.3 in 780 eyes (31.5%), >0.3 in 1 590 eyes (64.3%). A good consistency was found in the diagnosis and grading of DR (Kappa value=0.895) and in diagnosis of macular disorder (Kappa value=0.763) between the remote screening system and conventional screening method.In addition, the diagnosis results of retinal photocoagulation were consistent between the two methods (Kappa value=1.000). The mean duration of the remote screening system for one subject was 10 minutes, which was shorter than 23 minutes of conventional screening method.The follow-up rate of remote screening system was 75.2%. Conclusions There is a high consistency in the DR diagnosis and evaluation between the remote non-mydriatic screening system and conventional screening method.The screening program with follow-up requests has a satisfying follow-up rate, which could meet the demand of DR screening. Key words: Diabetic retinopathy/diagnosis; Mass screening; Health care services accessibility; Telemedicine; Cross-sectional studies

Ophthalmoscopy in the 21st century

Although the usefulness of viewing the ocular fundus is well-recognized, ophthalmoscopy is infrequently and poorly performed by most nonophthalmologist physicians, including neurologists. Barriers to the practice of ophthalmoscopy by nonophthalmologists include not only the technical difficulty related to direct ophthalmoscopy, but also lack of adequate training and discouragement by preceptors. Recent studies have shown that digital retinal fundus photographs with electronic transmission and remote interpretation of images by an ophthalmologist are an efficient and reliable way to allow examination of the ocular fundus in patients with systemic disorders such as diabetes mellitus. Ocular fundus photographs obtained without pharmacologic dilation of the pupil using nonmydriatic fundus cameras could be of great value in emergency departments (EDs) and neurologic settings. The Fundus Photography vs Ophthalmoscopy Trial Outcomes in the Emergency Department (FOTO-ED) study showed that ED providers consistently failed to correctly identify relevant ocular funduscopic findings using the direct ophthalmoscope, and that nonmydriatic fundus photography was an effective alternate way of providing access to the ocular fundus in the ED. Extrapolating these results to headache clinics, outpatient neurology clinics, and adult and pediatric primary care settings seems self-evident. As technology advances, nonmydriatic ocular fundus imaging systems will be of higher quality and more portable and affordable, thereby circumventing the need to master the use of the ophthalmoscope. Visualizing the ocular fundus is more important than the method used. Ocular fundus photography facilitates nonophthalmologists' performance of this essential part of the physical examination, thus helping to reestablish the value of doing so.

Scanning laser ophthalmoscopy – basic optical principles

SummaryScanning Laser Ophthalmoscopy (SLO) produces an ocular fundus image by moving a focused laser beam via scanning mirrors over the retina in a grid pattern and registering the reflected light from each scanned point. In confocal SLO, a pinhole is placed in front of the detector to cut off scattered or defocused light coming from outside the point of interest, which otherwise can blur the image. This results in a focused, high‐contrast image of a single tissue layer located at the focal plane. Tomographic information can be extracted by moving the plane of the pinhole. The use of various wavelengths allows for different applications, such as fluorescein angiography, indocyanine green angiography, and autofluorescence imaging. This lecture will present the basic principles of SLO, as well as discuss notable applications and variants of the technology.

Stimulation of μ-opioid receptors dilates retinal arterioles by neuronal nitric oxide synthase-derived nitric oxide in rats

Opioids contribute to the regulation of cerebral vascular tone. The purpose of this study was to examine the effects of herkinorin, a non-opioid μ-opioid receptor agonist derived from salvinorin A, on blood vessels in the rat retina and to investigate the mechanism underlying the herkinorin-induced retinal vasodilatory response. Ocular fundus images were captured using an original high-resolution digital fundus camera in vivo. The retinal vascular responses were evaluated by measuring the diameter of retinal arterioles in the fundus images. Both systemic blood pressure and heart rate were continuously recorded. Herkinorin increased the retinal arteriolar diameter without significantly changing mean blood pressure and heart rate. The retinal vasodilator response to herkinorin was almost completely prevented following treatment with naloxone, a nonselective opioid receptor antagonist and H-D-Phe-Cys-Tyr-D-Trp-Orn-Thr-Pen-Thr-NH2 (CTOP), a selective μ-opioid receptor antagonist. Nω-nitro-L-arginine methyl ester, a nonselective nitric oxide (NO) synthase inhibitor, or indomethacin, a cyclooxygenase inhibitor, significantly attenuated the herkinorin-induced retinal vasodilator responses. In addition, Nω-propyl-L-arginine, an inhibitor of neuronal NO synthase, diminished the herkinorin-induced retinal vasodilator responses. Seven days after an intravitreal injection of N-methyl-D-aspartic acid, loss of inner retinal neurons and abolishment of the retinal vasodilator response to herkinorin were observed. These results suggest that herkinorin dilates rat retinal arterioles through stimulation of retinal μ-opioid receptors. The μ-opioid receptor-mediated retinal vasodilator response is likely mediated by NO generated by neuronal NO synthase. Retinal neurons play an important role in the retinal vasodilator mechanism involving μ-opioid receptors in rats.

Stimulation of β1- and β2-adrenoceptors dilates retinal blood vessels in rats

Our previous studies have demonstrated that adrenaline dilates rat retinal arterioles by stimulating propranolol-sensitive β-adrenoceptors and β3-adrenoceptors, and selective stimulation of β2- or β3-adrenoceptors causes retinal vasodilator responses. In the present study, we compared the effects of β1- and β2-adrenoceptor stimulation on rat retinal arterioles in vivo. Rat ocular fundus images were captured using an original high-resolution digital fundus camera. Diameters of retinal arterioles contained in the images were measured. Systemic blood pressure and heart rate were recorded continuously. Denopamine, a β1-adrenoceptor agonist, increased the diameter of retinal arterioles and heart rate, and produced a small but statistically insignificant decrease in mean arterial pressure. CGP20712A, a β1-adrenoceptor antagonist, but not ICI118551, a β2-adrenoceptor antagonist, significantly prevented denopamine-induced retinal vasodilator and heart rate responses. Salbutamol, a β2-adrenoceptor agonist, increased the diameter of retinal arterioles and decreased mean arterial pressure without significantly changing heart rate. The effects of salbutamol were significantly prevented by ICI118551, but not by CGP20712A. These results suggest that stimulation of β1- and β2-adrenoceptors dilates retinal blood vessels and indicate that all three β-adrenoceptor subtypes (β1, β2, and β3) may be involved in the retinal vasodilator response to adrenaline in rats.

Anti-diabetic drug metformin dilates retinal blood vessels through activation of AMP-activated protein kinase in rats

The aim of this study was to examine whether metformin, a biguanide anti-hyperglycemic drug, dilates retinal blood vessels in rats. Ocular fundus images were captured with an original high-resolution digital fundus camera in vivo and diameters of retinal blood vessels were measured. Both systemic blood pressure and heart rate were continuously recorded. Metformin (0.01–0.3mg/kg/min) increased diameters of retinal blood vessels in a dose-dependent manner. This retinal vasodilator effect of metformin was abolished by compound C, an inhibitor of AMP-activated protein kinase (AMPK), and NG-nitro-L-arginine methyl ester, an inhibitor of nitric oxide (NO) synthase. Similar results were obtained with the AMPK activator 5-aminoimidazole-4-carboxamide-1-β-D-ribonucleoside (AICAR, 0.01–1mg/kg/min). Neither metformin nor AICAR exerted significant effect on mean blood pressure and heart rate. However, a significant pressor response to AICAR was observed upon inhibition of NO synthase. These results suggest that metformin dilates retinal blood vessels through activation of AMPK, and NO plays an important role in the retinal vasodilator response following AMPK activation.

A smart feature selection technique for object localization in ocular fundus images with the aid of color subspaces

We propose a technique for selecting effective features for object localization in ocular fundus images. The technique has made it possible to conduct smart feature analysis with the aid of color subspaces when solving a problem of selecting the areas of interest. The relevance of the problem is associated with enhancing the effectiveness of laser coagulation surgery. The proposed technique enables not only the informative features to be extracted in particular color spaces but also the most informative color subspace to be identified. The technique allows an effective feature for separating two particular classes to be identified at a definite size of the fragmentation block thanks to the use of various feature selection rules. The technique also makes it possible to find a universal feature using which two particular initial classes can be separated with a minimal clustering error in all color subspaces, also finding a color-specific in-formative feature enabling the majority of classes under study to be separated. The most informative color subspace was defined.

Scanning laser ophthalmoscopy – basic optical principles

SummaryScanning Laser Ophthalmoscopy (SLO) produces an ocular fundus image by moving a focused laser beam via scanning mirrors over the retina in a grid pattern and registering the reflected light from each scanned point. In confocal SLO, a pinhole is placed in front of the detector to cut off scattered or defocused light coming from outside the point of interest, which otherwise can blur the image. This results in a focused, high‐contrast image of a single tissue layer located at the focal plane. Tomographic information can be extracted by moving the plane of the pinhole. The use of various wavelengths allows for different applications, such as fluorescein angiography, indocyanine green angiography, and autofluorescence imaging. This lecture will present the basic principles of SLO, as well as discuss notable applications and variants of the technology.

Effect of small head tilt on ocular fundus image: Consideration of proper head positioning for ocular fundus scanning.

Head tilt and resultant ocular cyclotorsion can influence the results of ophthalmologic examinations. Thus, proper head positioning during fundus scanning has been emphasized. However, there is no perfect method to control the head tilt and little is known about the effect of small head tilts. In this study, we investigated the effect of minimal head tilt on the ocular cyclotorsion which we cannot easily detect.Forty-seven participants without ophthalmologic or vestibular abnormalities were recruited as normal subjects. Their faces were positioned at the desired head tilt using a customized adjustable head tilter and facial and fundus photographs of both the left and right eyes were taken in the upright neutral position; as well as at rightward and leftward head tilts of 2°, 4°, and 6°. The actual head tilt was determined using the facial photographs by measuring the slope of a line that intersected the corneal reflexes of both eyes. Rotational changes in the fundus images were recorded and the correlation of these changes with the degree of head tilt was determined.The degree of head tilt was significantly correlated with rotational changes in the fundus images from both the right and left eyes (P < 0.001; right eye: R2 = 0.897, left eye: R2 = 0.899). The mean relative compensations for head tilt, mediated by the ocular counterrolling reflex, were 0.376 ± 0.255 in the right eye (range: −0.02 to 1.0), and 0.350 ± 0.263 in the left eye (range: −0.03 to 1.0), and exhibited a significant negative correlation with head tilt (P < 0.05). The mean relative compensation of the right eye did not differ significantly from that of the left eye (P = 0.380), but the value did vary widely among individuals and within individuals.Even very small head tilt was partially and variably compensated for, and caused significant rotation in the fundus image. We concluded that proper head positioning does not guarantee the minimal ocular cyclotorsion change of the eyes and image-adjusting technique would be a better solution for minimizing errors from ocular cyclotorsion changes.