Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Transl Vis Sci Technol . 2026 Apr 6;15(4):2. doi: 10.1167/tvst.15.4.2 Search in PMC Search in PubMed View in NLM Catalog Add to search Diagnosis of High Intracranial Pressure by Non–Optic Nerve Retinal Image Features Farzan Abdolahi Farzan Abdolahi 1 Department of Ophthalmology, Keck School of Medicine of University of Southern California, Los Angeles, CA, USA Find articles by Farzan Abdolahi 1 , Kelvin Z Li Kelvin Z Li 2 Departments of Ophthalmology, Neurology and Neurological Sciences, Stanford University, Palo Alto, CA, USA 3 Department of Ophthalmology, Tan Tock Seng Hospital, Singapore, Singapore 4 Centre of AI in Medicine, Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, Singapore Find articles by Kelvin Z Li 2, 3, 4 , Yuhang Zou Yuhang Zou 1 Department of Ophthalmology, Keck School of Medicine of University of Southern California, Los Angeles, CA, USA 5 Alfred E. Mann Department of Biomedical Engineering, University of Southern California, Los Angeles, CA, USA Find articles by Yuhang Zou 1, 5 , Heather E Moss Heather E Moss 2 Departments of Ophthalmology, Neurology and Neurological Sciences, Stanford University, Palo Alto, CA, USA Find articles by Heather E Moss 2 , Mahnaz Shahidi Mahnaz Shahidi 1 Department of Ophthalmology, Keck School of Medicine of University of Southern California, Los Angeles, CA, USA 5 Alfred E. Mann Department of Biomedical Engineering, University of Southern California, Los Angeles, CA, USA Find articles by Mahnaz Shahidi 1, 5, ✉ Author information Article notes Copyright and License information 1 Department of Ophthalmology, Keck School of Medicine of University of Southern California, Los Angeles, CA, USA 2 Departments of Ophthalmology, Neurology and Neurological Sciences, Stanford University, Palo Alto, CA, USA 3 Department of Ophthalmology, Tan Tock Seng Hospital, Singapore, Singapore 4 Centre of AI in Medicine, Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, Singapore 5 Alfred E. Mann Department of Biomedical Engineering, University of Southern California, Los Angeles, CA, USA * Correspondence: Mahnaz Shahidi, Department of Ophthalmology, Keck School of Medicine of University of Southern California, 1450 San Pablo St., Los Angeles, CA 90033, USA. e-mail: [email protected] ✉ Corresponding author. Accepted 2026 Feb 24; Received 2025 Oct 29; Collection date 2026 Apr. Copyright 2026 The Authors This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. PMC Copyright notice PMCID: PMC13068025 PMID: 41940708 Abstract Purpose To assess the performance of an artificial intelligence deep learning (DL) model compared with neuro-ophthalmologists for the classification of subjects with elevated intracranial pressure (ICP) and papilledema versus control subjects, based on retinal images with and without masking the optic nerve heads (ONHs). Methods Widefield retinal images were obtained in 32 subjects (70 images) with elevated ICP and papilledema and 31 control subjects (62 images). ONH-unmasked images were generated by cropping the image to a 30° circular region. ONH-masked images were generated by masking the ONH and peripapillary region. Classification was performed using a convolutional neural network model and by two neuro-ophthalmologists (graders). Results For the ONH-unmasked images, the classification accuracies of the model, grader 1, and grader 2 were 83% (area under the receiver operating characteristic curve [AUC] = 0.94), 93%, and 86%, respectively. For the ONH-masked images, the classification accuracies of the model, grader 1, and grader 2 were 79% (AUC = 0.90), 66%, and 76%, respectively. The classification performance of the DL model and graders did not significantly differ for both datasets ( P ≥ 0.38). There was no significant effect of ONH masking on the performance of the DL model ( P = 1.0) and grader 2 ( P = 0.38), whereas the performance of grader 1 was significantly reduced ( P = 0.02). Conclusions Both expert graders and the DL models demonstrated excellent performance for classifying retinal images of subjects with elevated and normal ICP using images with or without masking the ONH. Translational Relevance Methods for assessment of non-optic nerve retinal image features have the potential to improve diagnosis and monitoring the progression and response to treatment of elevated ICP. Keywords: papilledema, idiopathic intracranial hypertension, artificial intelligence, deep learning, retinal imaging Introduction Papilledema, optic disc swelling resulting from elevated intracranial pressure (ICP), is visible on the bedside ophthalmoscopic exam and accordingly is an important marker for diagnosing and monitoring elevated ICP. 1 However, optic nerve appearance is an indirect marker of ICP, and direct ICP monitoring tools, such as lumbar puncture or intracranial pressure monitoring, are invasive with accordant risks. The current monitoring tools for papilledema are based on visual function test, patient symptoms, and optic disc swelling grading, such as Frisen score. 2 , 3 Visual field testing is critically important for management of papilledema, 4 and available objective methods of visual function testing can overcome the subjectivity of standard automated perimetry. 5 Although higher Frisen scores have been reported to be associated with increased risk of permanent vision loss, 6 – 8 this scoring system is subjective and may not be sensitive to detect subtle changes in papilledema. 9 Optical coherence tomography (OCT) offers quantitative analysis of retinal layer thickness and correlates with modified Frisen scores, 10 and it may serve as a complementary tool for detection of papilledema. 11 Other retinal features are also impacted by increased ICP and are candidates for biomarkers. Retinal vessel appearance is altered due to blood inflow and outflow variations induced by the increased ICP and compression of the vessels by the swelling of the optic disc. Previous studies have reported alterations in peripapillary vasculature in patients with elevated ICP. 12 , 13 Moss et al. 14 reported an increase in retinal venous diameter in papilledema, and a decrease in diameter after treatment, a finding also reported by Lee et al. 15 after optic nerve sheath decompression. Moss et al. 16 also reported that retinal arteriole and venule diameters decrease within an hour of reduced ICP via lumbar puncture in patients with papilledema. Using OCT angiography, Kaya et al. 17 demonstrated increased peripapillary capillary density in idiopathic intracranial hypertension (IIH) patients compared to healthy controls, a result supported by findings from Korçer et al. 18 Furthermore, peripapillary vessel density was shown to be associated with papilledema severity 19 and was lower in IIH patients with papilledema. 20 In addition, retinal vessel tortuosity was reliably measured in papilledema 21 and shown to correlate with optic nerve head (ONH) edema. 22 With the advancement of quantitative assessments of papilledema using fundus photography and OCT, a growing number of novel machine learning and artificial intelligence approaches have been applied for differential diagnosis and evaluating the severity of papilledema. 23 , 24 Notably, the Brain and Optic Nerve Study with Artificial Intelligence (BONSAI) group developed a deep learning (DL) system that achieved a high area under the curve in distinguishing papilledema due to high ICP from normal disc and other disc abnormalities, using standard color fundus photographs. 24 This model performed as well as neuro-ophthalmologists in classifying optic disc abnormalities. 25 Furthermore, when their model was extended to grade papilledema severity it demonstrated strong performance in differentiating mild to moderate from severe papilledema. 26 The model was similarly extended to distinguish papilledema from optic disc drusen. 27 More recently, a tri-branch convolutional neural network was shown to outperform human experts in classifying pediatric papilledema due to high ICP versus pseudo-papilledema. 28 These findings highlight the promising potential of artificial intelligence–based models for papilledema identification and classification. The clinical diagnosis of papilledema is likely based on the convolved observation of ONH swelling and retinal features such as vascular morphological changes. 29 Knowledge gained by differentiating these two factors may elucidate the role of features such as vasculopathy in the disease development and may offer novel diagnostic biomarkers. The purpose of the current study was to investigate the performance of a DL model compared with neuro-ophthalmologists for classification of papilledema versus healthy subjects based on retinal images with masked or unmasked ONHs. Materials and Methods Subjects The study protocol was reviewed and approved by an Institutional Review Board of Stanford University. The study adhered to the tenets of the Declaration of Helsinki and was approved with a waiver of consent by the Stanford Institutional Review Board. Adult subjects with a diagnosis of IIH and retinal photographs seen at Byers Eye Institute at Stanford were identified using the Stanford Research Repository (STARR). Subjects who met diagnostic criteria for IIH using modified Dandy criteria and had widefield retinal imaging demonstrating papilledema prior to treatment or when partially treated were included. Control adult subjects seen at Byers Eye Institute at Stanford for hydroxychloroquine retinal screening who had widefield retinal imaging were identified using STARR. Exclusion criteria for both groups included poor quality of images, presence of other retinal pathologies (e.g., panretinal photocoagulation scars), absence of corresponding consultation notes, or unavailability of Digital Imaging and Communications in Medicine (DICOM) format image exports. Seventy deidentified images from 32 IIH subjects were included: first visit prior to treatment ( n = 3), during treatment ( n = 25), or recurrence after previously completing a course of treatment ( n = 4). Three subjects contributed images in two categories (one subject had images prior to treatment and during treatment; two subjects had images during treatment and recurrence). Sixty-two deidentified images from 31 control subjects were included. Imaging and Processing Clinical retinal images were obtained using a widefield scanning laser ophthalmoscope (Optos, Dunfermline, UK) encompassing up to a 200° field of view centered on the macula. To compensate for variations in image magnification and field of view among eyes, each retinal image was scaled and cropped according to the measured distance between the ONH center and the foveal center. ONH-unmasked retinal images were generated by cropping the image to a circular region with an approximately 30° field of view, centered approximately 5° temporally from the ONH center. ONH-masked retinal images were generated by masking the papillary region, using a circle with a diameter approximately twice the average size of the ONH diameters in images of subjects with papilledema to keep the ONH-mask the same across images. The Figure shows examples of ONH-unmasked and ONH-masked retinal images in control and papilledema subjects. Figure. Open in a new tab ( a , b ) Examples of ONH-unmasked and ONH-masked retinal images in a control subject. ( c , d ) ONH-unmasked and ONH-masked retinal images in a subject with papilledema. Deep Learning Model We utilized the EfficientNet-B0 model, a convolutional neural network, pretrained on ImageNet to benefit from transfer learning. The classification layer was replaced with a linear layer suitable for binary classification. Because of the limited number of images available, the images were randomly split into train, validation, and test datasets with a distribution of 70%, 10%, and 20%, respectively. To prevent information leakage, data splitting was done to ensure that images from both eyes of a subject were assigned to a single dataset (train, validation, or test). Data augmentation was applied to the training dataset to improve model generalization. Data augmentation employed included random resized cropping to 224 × 224 pixels, random horizontal flipping, random rotation within a range of ±15°, and color jittering (brightness and contrast). Images in all datasets were normalized using the ImageNet mean and standard deviation. Training was done with a learning rate of 1e-4, for 50 epochs, with a batch size of 16. Two models were trained independently on ONH-unmasked and ONH-masked retinal images to classify images as papilledema or control by thresholding the probability output at 0.5. The best-performing model based on validation loss was selected and evaluated on the test dataset. Neuro-Ophthalmologists’ Classification Two fellowship-trained neuro-ophthalmologists (graders) who were masked to the clinical diagnosis and DL model results independently classified test images (ONH-unmasked and ONH-masked retinal images) as papilledema or control. Classification was based solely on retinal images for comparison with the DL models. The ground truth for clinical diagnosis was made by the managing neuro-ophthalmologist based on comprehensive clinical examination including history, ophthalmic and magnetic resonance imaging, and lumbar puncture results. Statistical Analysis Differences in demographic and systemic parameters between control and papilledema groups were evaluated using the χ 2 test for categorical variables and either unpaired t -tests or Mann–Whitney U tests depending on the normality of variable distribution for continuous variables. The DL model and grader results were compared to clinical diagnosis using 29 test images. For the DL models, the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity were calculated. The accuracy, sensitivity, and specificity were calculated for each of the two graders. The classification performance (accuracy) between DL models and graders and the performance between ONH-unmasked and ONH-masked retinal images were compared using McNemar's tests. Statistical significance was accepted at P < 0.05. Results Characteristics of Participants Table 1 shows the demographics and systemic parameters of the study participants. The study included 132 images from control (62 images) and papilledema (70 images) subjects. For one subject, images were available in only one eye, and, as noted above, three subjects contributed more than two images. The test dataset included 29 images (15 papilledema, 14 control), and papilledema images had Frisen scores between 1 and 3 (5 images per score). In the test dataset, there were no statistically significant differences in age ( P = 0.30), body mass index ( P = 0.09), or systolic or diastolic blood pressure ( P ≥ 0.8) between groups. Table 1. Characteristics of Study Participants Variable Control Papilledema P Subjects, n 31 32 Images, n (%) 62 (47) 70 (53) Female sex, n (%) 24 (90) 27 (84) 0.7 a Age (y), mean ± SD 46 ± 18 30 ± 8 <0.001 b BMI, median (interquartile range) 26.4 (24.1–30.0) 34.4 (28.0–42.4) <0.001 c SBP (mm Hg), median (interquartile range) 115.0 (108.0–127.0) 122.5 (114.0–133.5) 0.09 c DBP (mm Hg), mean ± SD 67.3 ± 11.7 76.6 ± 14.2 0.01 b MAP (mm Hg), median (interquartile range) 81.7 (76.7–91.7) 91.3 (82.2–100.4) 0.02 c Frisen score, n (%) — 1 — 19 (27.1) 2 — 29 (41.4) 3 — 17 (24.3) 4 — 5 (7.1) Open in a new tab BMI, body mass index; DBP, diastolic blood pressure; MAP, mean arterial pressure; SBP, systolic blood pressure. a χ 2 test. b Unpaired t -test. c Mann–Whitney U test. Classification Performance The AUCs of the DL models for classification of papilledema versus control using ONH-unmasked and ONH-masked test retinal images were 0.94 and 0.90, respectively. The accuracy, sensitivity, and specificity of the DL models and the two graders using the ONH-unmasked and ONH-masked retinal images are listed in Table 2 . For the ONH-unmasked retinal images, the classification accuracies of the DL model and the two graders were 83%, 93%, and 86%, respectively. The classification performance of the DL model and graders did not differ significantly ( P ≥ 0.38). For the ONH-masked retinal images, the classification accuracies of the DL model and two graders were 79%, 66%, and 76%, respectively. The classification performance of the DL model and the graders did not differ significantly ( P ≥ 0.29). There was no significant effect of ONH masking on the performance of the DL model ( P = 1.0) or grader 2 ( P = 0.38), but the performance of grader 1 was significantly reduced by masking of the ONH ( P = 0.02). Table 2. Classification Performance of the DL Models and Graders on the Test Image Set Accuracy (%) Sensitivity (%) Specificity (%) ONH-unmasked Model 83 87 79 Grader 1 93 93 93 Grader 2 86 73 100 ONH-masked Model 79 80 79 Grader 1 66 40 92 Grader 2 76 67 86 Open in a new tab Discussion High intracranial pressure is a medical emergency due to the possibility of a secondary cause requiring emergent treatment and the potential for irreversible vision loss from papilledema. The ophthalmoscopic exam is critically important for early detection of papilledema with swelling of the ONH being an important sign. However, a sudden increase in ICP may not initially present with papilledema, because swelling of the optic disc is a gradual process. Other ophthalmic changes may also be relevant for detection of high ICP, including retinal vessel shape and size and peripapillary wrinkles and chorioretinal folds. 12 , 30 Notably, chorioretinal folds have been shown to be present in patients with high ICP even in the absence of papilledema. 31 In this study, we sought to assess the diagnostic potential of non-ONH changes in differentiating retinal images of patients with high ICP using widefield fundus photographs. The classification performance of the DL models and neuro-ophthalmologists was similar for the dataset of images with and without ONH ablation, confirming the diagnostic importance of the extra-optic nerve ophthalmic findings. Our results build on those of DL models with excellent performance for detection and classification of optic disc edema based on ONH imaging alone, as well as reports of cross-sectional and longitudinal differences in non-optic nerve retinal features between high and normal ICP states. Our results show successful classification of ICP state using widefield retinal images even with ONH ablation by both DL models and neuro-ophthalmologists. This finding demonstrates the relevance of posterior pole changes other than the ONH for diagnosis of high ICP. This confirms and advances beyond demonstration of between- and within-subject differences in features including retinal vessels and retinal wrinkles. In fact, ablating the ONH resulted in no significant reduction in classification performance for the DL model and one of the neuro-ophthalmologists, demonstrating that the non-ONH retina appearance was sufficient for classification of ICP state. This is striking given that attention maps analyses of previously published DL algorithms for ICP classification consistently highlight the optic nerve as the region contributing the most to the prediction of the model. 24 , 32 The results support further development and evaluation of diagnostic and classification models for high ICP using widefield fundus photography in addition to those focusing on the ONH, particularly in cases of acute ICP elevation when papilledema may not have developed. The observations also have implications for understanding the pathophysiological impact of high ICP on the globe beyond the optic nerve which may contribute to visual symptoms and impairment. To fully elucidate the contribution of retinal image features for classification, future studies are warranted to add images with masked retina in addition to images with masked and unmasked ONH. Interestingly, the classification performance of one neuro-ophthalmologist was reduced when the ONH was masked from the images. Physician diagnosis in the clinical setting considers multimodal inputs including patient history, examination, and imaging. In contrast, the task in this study utilized a single modality. The reduction in performance for one neuro-ophthalmologist but not the other likely reflects individual differences in diagnostic approach—specifically, their approach to image interpretation. 33 , 34 The use of clinically acquired images defined by the ICP state rather than by the prominence of papilledema is a strength of the study but also a source of limitations. Hydroxychloroquine treatment facilitated identification of clinically obtained fundus photographs in a young population likely without high ICP, although it is possible that drug treatment or comorbidity differences from the high ICP patients may be associated with retinal differences other than those attributed to ICP. There were statistically significant differences in age and systemic parameters between groups, which may have had confounding effects on retinal images. However, in the test dataset, these differences did not reach statistical significance. The single-center nature and use of clinically acquired images limited the sample size and likely performance of the DL models; however, the AUC of our model with the optic disc is only marginally lower than those reported by previous studies with larger sample sizes. 19 , 20 The small sample size may have contributed to the finding of a lack of significant difference in classification using unmasked and masked images, requiring further studies. Additionally, because of the small sample size, images from both eyes of subjects were included which may have affected the performance of the DL models and the neuro-ophthalmologists (who had access to the full image dataset). However, the effect on the DL model was minimized by including images from both eyes of the same subject together in the training, validation, or test groups. Our dataset contained only images of patients with papilledema and normal subjects. Retinal images with other disorders affecting the ONH and retina were not present in our dataset, which does not reflect the diversity of images encountered in clinical practice. This might have simplified the classification task, giving our DL models and neuro-ophthalmologists an advantage. In conclusion, both DL models and neuro-ophthalmologists were able to correctly classify widefield fundus images of subjects with high ICP and normal ICP, both with and without masking of the ONH. Future directions include determining the relevance of extraoptic nerve fundus image changes for detecting ICP treatment response and differentiating between causes of ONH edema or pallor. Acknowledgments Supported by grants from the National Eye Institute, National Institutes of Health (EY030115, EY029220, EY026877, EY031726), and by unrestricted departmental awards (University of Southern California and Stanford University) from Research to Prevent Blindness. Disclosure: F. Abdolahi , None; K.Z. Li, None; Y. Zou , None; H.E. Moss , None; M. Shahidi , None References 1. Rigi M, Almarzouqi SJ, Morgan ML, Lee AG. Papilledema: epidemiology, etiology, and clinical management. Eye Brain . 2015; 7: 47–57. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. Frisen L. Swelling of the optic nerve head: a staging scheme. J Neurol Neurosurg Psychiatry . 1982; 45: 13–18. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Frisen L. Swelling of the optic nerve head: a backstage view of a staging scheme. J Neuroophthalmol . 2017; 37: 3–6. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Wall M. The importance of visual field testing in idiopathic intracranial hypertension. Continuum (Minneap Minn) . 2014; 20: 1067–1074. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Moss HE. Objective measures of visual function in papilledema. Adv Ophthalmol Optom . 2016; 1: 231–247. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Behbehani R, Ali A, Al-Moosa A. Course and predictors of visual outcome of idiopathic intracranial hypertension. Neuroophthalmology . 2022; 46: 80–84. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Gospe SM 3rd, Bhatti MT, El-Dairi MA. Anatomic and visual function outcomes in paediatric idiopathic intracranial hypertension. Br J Ophthalmol . 2016; 100: 505–509. [ DOI ] [ PubMed ] [ Google Scholar ] 8. Wall M, White WN 2nd. Asymmetric papilledema in idiopathic intracranial hypertension: prospective interocular comparison of sensory visual function. Invest Ophthalmol Vis Sci . 1998; 39: 134–142. [ PubMed ] [ Google Scholar ] 9. Sheils CR, Fischer WS, Hollar RA, Blanchard LM, Feldon SE, NORDIC Idiopathic Intracranial Hypertension Study Group. The relationship between optic disc volume, area, and Frisen score in patients with idiopathic intracranial hypertension. Am J Ophthalmol . 2018; 195: 101–109. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Scott CJ, Kardon RH, Lee AG, Frisen L, Wall M. Diagnosis and grading of papilledema in patients with raised intracranial pressure using optical coherence tomography vs clinical expert assessment using a clinical staging scale. Arch Ophthalmol . 2010; 128: 705–711. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Costello F, Hamann S. Advantages and pitfalls of the use of optical coherence tomography for papilledema. Curr Neurol Neurosci Rep . 2024; 24: 55–64. [ DOI ] [ PubMed ] [ Google Scholar ] 12. Moss HE. Retinal vein changes as a biomarker to guide diagnosis and management of elevated intracranial pressure. Front Neurol . 2021; 12: 751370. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Moss HE, Treadwell G, Wanek J, DeLeon S, Shahidi M. Retinal vessel diameter assessment in papilledema by semi-automated analysis of SLO images: feasibility and reliability. Invest Ophthalmol Vis Sci . 2014; 55: 2049–2054. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Moss HE, Hollar RA, Fischer WS, Feldon SE. Retinal vessel diameter changes after 6 months of treatment in the Idiopathic Intracranial Hypertension Treatment Trial. Br J Ophthalmol . 2020; 104: 1430–1434. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Lee SY, Shin DH, Spoor TC, Kim C, McCarty B, Kim D. Bilateral retinal venous caliber decrease following unilateral optic nerve sheath decompression. Ophthalmic Surg . 1995; 26: 25–28. [ PubMed ] [ Google Scholar ] 16. Moss HE, Vangipuram G, Shirazi Z, Shahidi M. Retinal vessel diameters change within 1 hour of intracranial pressure lowering. Transl Vis Sci Technol . 2018; 7: 6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Kaya FS, Sonbahar O, Acar PA, Ozbas M, Yigit FU. Evaluating peripapillary vessel density in regressed papilledema in idiopathic intracranial hypertension patients. Photodiagnosis Photodyn Ther . 2021; 36: 102551. [ DOI ] [ PubMed ] [ Google Scholar ] 18. Korçer AM, Yilmaz A, Atesoglu HI, Turkay M, Citirik M. Evaluation of retinal vascular and structural alterations in pediatric idiopathic intracranial hypertension patients without papilledema. Photodiagnosis Photodyn Ther . 2023; 42: 103608. [ DOI ] [ PubMed ] [ Google Scholar ] 19. Rodriguez Torres Y, Lee P, Mihlstin M, Tomsak RL. Correlation between optic disc peripapillary capillary network and papilledema grading in patients with idiopathic intracranial hypertension: a study of optical coherence tomography angiography. J Neuroophthalmol . 2021; 41: 48–53. [ DOI ] [ PubMed ] [ Google Scholar ] 20. Tuntas Bilen F, Atilla H. Peripapillary vessel density measured by optical coherence tomography angiography in idiopathic intracranial hypertension. J Neuroophthalmol . 2019; 39: 319–323. [ DOI ] [ PubMed ] [ Google Scholar ] 21. Moss HE, Cao J, Wasi M, Feldon SE, Shahidi M. Variability of retinal vessel tortuosity measurements using a semiautomated method applied to fundus images in subjects with papilledema. Transl Vis Sci Technol . 2021; 10: 32. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Buickians D, Dahshan D, Shahidi M, Nankali A, Feldon SE, Moss H. Retinal vessel tortuosity indices (VTIs) as a potential biomarker of idiopathic intracranial hypertension (IIH): baseline analysis of the IIH treatment trial. Invest Ophthalmol Vis Sci . 2023; 64: 4101. [ Google Scholar ] 23. Akbar S, Akram MU, Sharif M, Tariq A, Yasin UU. Decision support system for detection of papilledema through fundus retinal images. J Med Syst . 2017; 41: 66. [ DOI ] [ PubMed ] [ Google Scholar ] 24. Milea D, Najjar RP, Zhubo J, et al.. Artificial intelligence to detect papilledema from ocular fundus photographs. N Engl J Med . 2020; 382: 1687–1695. [ DOI ] [ PubMed ] [ Google Scholar ] 25. Biousse V, Newman NJ, Najjar RP, et al.. Optic disc classification by deep learning versus expert neuro-ophthalmologists. Ann Neurol . 2020; 88: 785–795. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Vasseneix C, Najjar RP, Xu X, et al.. Accuracy of a deep learning system for classification of papilledema severity on ocular fundus photographs. Neurology . 2021; 97: e369–e377. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Sathianvichitr K, Najjar RP, Zhiqun T, et al.. A deep learning approach for accurate discrimination between optic disc drusen and papilledema on fundus photographs. J Neuroophthalmol . 2024; 44: 454–461. [ DOI ] [ PubMed ] [ Google Scholar ] 28. Chang MY, Heidary G, Beres S, et al.. Artificial intelligence to differentiate pediatric pseudopapilledema and true papilledema on fundus photographs. Ophthalmol Sci . 2024; 4: 100496. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Fatima KN, Hassan T, Akram MU, Akhtar M, Butt WH. Fully automated diagnosis of papilledema through robust extraction of vascular patterns and ocular pathology from fundus photographs. Biomed Opt Express . 2017; 8: 1005–1024. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Sibony PA, Kupersmith MJ, Feldon SE, Wang JK, Garvin M, OCT Substudy Group for the NORDIC Idiopathic Intracranial Hypertension Treatment Trial. Retinal and choroidal folds in papilledema. Invest Ophthalmol Vis Sci . 2015; 56: 5670–5680. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Griebel SR, Kosmorsky GS. Choroidal folds associated with increased intracranial pressure. Am J Ophthalmol . 2000; 129: 513–516. [ DOI ] [ PubMed ] [ Google Scholar ] 32. Szanto D, Erekat A, Woods B, et al.. Deep learning approach readily differentiates papilledema, non-arteritic anterior ischemic optic neuropathy, and healthy eyes. Am J Ophthalmol . 2025; 276: 99–108. [ DOI ] [ PubMed ] [ Google Scholar ] 33. Berneshawi AR, Patwardhan A, Yu M, Liao YJ. Expert vs. novice: characterizing visuo-motor behavior during static perimetry interpretation. Invest Ophthalmol Vis Sci . 2023; 64: 4121. [ Google Scholar ] 34. Patwardhan AC, Berneshawi AR, Yu M, Liao YJ. Expert vs. novice review of ophthalmic data: an oculography study of optic neuropathies. Invest Ophthalmol Vis Sci . 2023; 64: 4109. [ Google Scholar ] Articles from Translational Vision Science & Technology are provided here courtesy of Association for Research in Vision and Ophthalmology ACTIONS View on publisher site PDF (1.2 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top