Imaging and Diagnostics

Schill HM, Wolfe JM, Brady TF. Relationships between expertise and distinctiveness: Abnormal medical images lead to enhanced memory performance only in experts. Mem Cognit 2021;49(6):1067-1081.Abstract
Memories are encoded in a manner that depends on our knowledge and expectations ("schemas"). Consistent with this, expertise tends to improve memory: Experts have elaborated schemas in their domains of expertise, allowing them to efficiently represent information in this domain (e.g., chess experts have enhanced memory for realistic chess layouts). On the other hand, in most situations, people tend to remember abnormal or surprising items best-those that are also rare or out-of-the-ordinary occurrences (e.g., surprising-but not random-chess board configurations). This occurs, in part, because such images are distinctive relative to other images. In the current work, we ask how these factors interact in a particularly interesting case-the domain of radiology, where experts actively search for abnormalities. Abnormality in mammograms is typically focal but can be perceived in the global "gist" of the image. We ask whether, relative to novices, expert radiologists show improved memory for mammograms. We also test for any additional advantage for abnormal mammograms that can be thought of as unexpected or rare stimuli in screening. We find that experts have enhanced memory for focally abnormal images relative to normal images. However, radiologists showed no memory benefit for images of the breast that were not focally abnormal, but were only abnormal in their gist. Our results speak to the role of schemas and abnormality in expertise; the necessity for spatially localized abnormalities versus abnormalities in the gist in enhancing memory; and the nature of memory and decision-making in radiologists.
Rageh A, Ashraf M, Fleming A, Silva PS. Automated Microaneurysm Counts on Ultrawide Field Color and Fluorescein Angiography Images. Semin Ophthalmol 2021;36(4):315-321.Abstract
BACKGROUND: The severity and extent of microaneurysms (MAs) have been used to determine diabetic retinopathy (DR) severity and estimate the risk of DR progression over time. The recent introduction of ultrawide field (UWF) imaging has allowed ophthalmologists to readily image nearly the entire retina. Manual counting of MAs, especially on UWF images, is laborious and time-consuming, limiting its potential use in clinical settings. Automated MA counting techniques are potentially more accurate and reproducible compared to manual methods. METHOD: Review of available literature on current techniques of automated MA counting techniques on both ultrawide field (UWF) color images (CI) and fluorescein angiography (FA) images. RESULTS: Automated MA counting techniques on UWF images are still in the early phases of development with UWF-FA counts being further along. Early studies have demonstrated that these techniques are accurate and reproducible. CONCLUSION: Automated techniques may be an appropriate option for detecting and quantifying MAs on UWF images, especially in eyes with earlier DR severity. Larger studies are needed to appropriately validate these techniques and determine if they add substantially to clinical practice compared to standard DR grading.
Armstrong GW, Kalra G, De Arrigunaga S, Friedman DS, Lorch AC. Anterior Segment Imaging Devices in Ophthalmic Telemedicine. Semin Ophthalmol 2021;36(4):149-156.Abstract
Obtaining a clear assessment of the anterior segment is critical for disease diagnosis and management in ophthalmic telemedicine. The anterior segment can be imaged with slit lamp cameras, robotic remote controlled slit lamps, cell phones, cell phone adapters, digital cameras, and webcams, all of which can enable remote care. The ability of these devices to identify various ophthalmic diseases has been studied, including cataracts, as well as abnormalities of the ocular adnexa, cornea, and anterior chamber. This article reviews the current state of anterior segment imaging for the purpose of ophthalmic telemedical care.
Dahrouj M, Miller JB. Artificial Intelligence (AI) and Retinal Optical Coherence Tomography (OCT). Semin Ophthalmol 2021;36(4):341-345.Abstract
Ophthalmology has been at the forefront of medical specialties adopting artificial intelligence. This is primarily due to the "image-centric" nature of the field. Thanks to the abundance of patients' OCT scans, analysis of OCT imaging has greatly benefited from artificial intelligence to expand patient screening and facilitate clinical decision-making.In this review, we define the concepts of artificial intelligence, machine learning, and deep learning and how different artificial intelligence algorithms have been applied in OCT image analysis for disease screening, diagnosis, management, and prognosis.Finally, we address some of the challenges and limitations that might affect the incorporation of artificial intelligence in ophthalmology. These limitations mainly revolve around the quality and accuracy of datasets used in the algorithms and their generalizability, false negatives, and the cultural challenges around the adoption of the technology.
Ratanawongphaibul K, Tsikata E, Zemplenyi M, Lee H, Margeta MA, Ondeck CL, Kim J, Pan BX, Petrakos P, Coleman AL, Yu F, de Boer JF, Chen TC. Earlier Detection of Glaucoma Progression Using High-Density 3-Dimensional Spectral-Domain OCT Optic Nerve Volume Scans. Ophthalmol Glaucoma 2021;4(6):604-616.Abstract
PURPOSE: To compare onset times of glaucoma progression among different glaucoma tests: disc photography (DP), visual field (VF) testing, 2-dimensional (2D) retinal nerve fiber layer (RNFL) thickness, and 3-dimensional (3D) spectral-domain (SD) OCT neuroretinal rim measurements. DESIGN: Prospective, longitudinal cohort study. PARTICIPANTS: One hundred twenty-four eyes of 124 patients with open-angle glaucoma. METHODS: Over a 5-year period, 124 patients with open-angle glaucoma underwent yearly DP, VF testing, SD OCT RNFL thickness scans, and optic nerve volume scans (Spectralis; Heidelberg Engineering), all performed on the same day. From high-density optic nerve volume scans, custom-built software calculated the minimum distance band (MDB) thickness, a 3D neuroretinal rim parameter. Patients were classified as glaucoma progressors or nonglaucoma progressors using event-based analysis. Progression by DP and VF testing occurred when 3 masked glaucoma specialists unanimously concurred. Progression by RNFL and MDB thickness occurred if change of more than test-retest variability was observed. Kaplan-Meier curves were constructed to analyze time-to-progression data. Kappa Coefficients were used to measure agreement of progressing eyes among methods. MAIN OUTCOME MEASURES: Time to glaucoma progression among all 4 methods. RESULTS: Global MDB thickness detected glaucoma progression in the highest percentage of eyes (52.4%) compared with DP (16.1%; P < 0.001) and global RNFL thickness (15.3%; P < 0.001). Global MDB thickness detected glaucoma progression earlier than either DP (23 months vs. 44 months; P < 0.001) or global RNFL thickness (23 months vs. 33 months; P < 0.001). Among MDB progressing eyes, 46.2% were confirmed simultaneously or later by other conventional methods. Agreement of glaucoma-progressing eyes for all 4 methods in paired fashion were slight to fair (κ = 0.095-0.300). CONCLUSIONS: High-density 3D SD OCT neuroretinal rim measurements detected glaucoma progression approximately 1 to 2 years earlier compared with current clinically available structural tests (i.e., DP and 2D RNFL thickness measurements).
Wang M, Garg I, Miller JB. Wide Field Swept Source Optical Coherence Tomography Angiography for the Evaluation of Proliferative Diabetic Retinopathy and Associated Lesions: A Review. Semin Ophthalmol 2021;36(4):162-167.Abstract
Retinal imaging remains the mainstay for monitoring and grading diabetic retinopathy. The gold standard for detecting proliferative diabetic retinopathy (PDR) requiring treatment has long been the seven-field stereoscopic fundus photography and fluorescein angiography. In the past decade, ultra-wide field fluorescein angiography (UWF-FA) has become more commonly used in clinical practice for the evaluation of more advanced diabetic retinopathy. Since its invention, optical coherence tomography (OCT) has been an important tool for the assessment of diabetic macular edema; however, OCT offered little in the assessment of neovascular changes associated with PDR until OCT-A became available. More recently, swept source OCT allowed larger field of view scans to assess a variety of DR lesions with wide field swept source optical coherence tomography (WF-SS-OCTA). This paper reviews the role of WF-SS-OCTA in detecting neovascularization of the disc (NVD), and elsewhere (NVE), microaneurysms, changes of the foveal avascular zone (FAZ), intraretinal microvascular abnormalities (IRMA), and capillary non-perfusion, as well as limitations of this evolving technology.
Jabroun MN, AlWattar BK, Fulton AB. Optical Coherence Tomography Angiography in Prematurity. Semin Ophthalmol 2021;36(4):264-269.Abstract
Purpose: During normal foveal development there is a close interaction between the neurosensory and vascular elements of the fovea making it vulnerable to prematurity and retinopathy of prematurity (ROP). We aim to assess this potential effect on foveal development in preterms evaluated simultaneously with both optical coherence tomography (OCT) and OCT angiography (OCTA).Method: Unrestricted literature search in the PubMed and Cochrane library databases yielded 20 distinct citations. Fifteen were relevant and reviewed.Results: In preterms, OCTA demonstrated a significant decrease in the foveal avascular zone area and an increase in foveal vessel density. OCT showed a decrease in foveal pit depth and an increase in the thickness of the subfoveal retinal layers. Some studies correlated these changes with reduced vision.Conclusion: Changes in the vascular and neurosensory retina were found in premature children. It remains unclear whether this is related to prematurity alone or ROP and its treatment.
Ludwig CA, Moon J, Garg I, Miller JB. Ultra-Widefield Imaging for Evaluation of the Myopic Eye. Semin Ophthalmol 2021;36(4):185-190.Abstract
Topic : Ultra-widefield (UWF) imaging of the myopic eye. Clinical Relevance : Myopes, and particularly high and pathologic myopes, present a unique challenge in fundoscopic imaging. Critical pathology is often located in the anteriormost portion of the retina, variations in posterior segment contour are difficult to capture in two-dimensional images, and extremes in axial length make simply focusing imaging devices difficult. Methods: We review the evolution of modalities for ophthalmic imaging (color fundus photography [CFP], optical coherence topography [OCT], angiography, artificial intelligence [AI]) to present day UWF technology and its impact on our understanding of myopia. Results: Advances in UWF technology address many of the challenges in fundoscopic imaging of myopes, providing new insights into the structure and function of the myopic eye. UWF CFP improves our ability to detect and document anterior peripheral pathology prevalent in approximately half of all high myopes. UWF OCT better captures the staphylomatous contour of the myopic eye, providing enhanced visualization of the vitreoretinal interface and progressive development of myopic traction maculopathy. UWF angiography highlights the posterior vortex veins, thin choriocapillaris, far peripheral avascularity, and peripheral retinal capillary microaneurysms more prevalent in the myopic eye. Researchers have demonstrated the ability of AI algorithms to predict refractive error, and great potential remains in the use of AI technology for the screening and prevention of myopic disease. Conclusion: We note significant progress in our ability to capture anterior pathology and improved image quality of the posterior segment of high and pathologic myopes. The next jump forward for UWF imaging will be the ability to capture a high quality ora to ora multimodal fundoscopic image in a single scan that will allow for sensitive AI-assisted screening of myopic disease.
Armstrong GW, Kim LA, Vingopoulos F, Park J, Garg I, Kasetty M, Silverman RF, Zeng R, Douglas VP, Lopera F, Baena A, Giraldo M, Norton D, Cronin-Golomb A, Arboleda-Velasquez JF, Quiroz YT, Miller JB. Retinal Imaging Findings in Carriers With PSEN1-Associated Early-Onset Familial Alzheimer Disease Before Onset of Cognitive Symptoms. JAMA Ophthalmol 2021;139(1):49-56.Abstract
Importance: Individuals with autosomal dominant mutations for Alzheimer disease are valuable in determining biomarkers present prior to the onset of cognitive decline, improving the ability to diagnose Alzheimer disease as early as possible. Optical coherence tomography (OCT) has surfaced as a potential noninvasive technique capable of analyzing central nervous system tissues for biomarkers of Alzheimer disease. Objective: To evaluate whether OCT can detect early retinal alterations in carriers of the presenilin 1 (PSEN1 [OMIM 104311]) E280A mutation who are cognitively unimpaired. Design, Setting, and Participants: A cross-sectional imaging study conducted from July 13, 2015, to September 16, 2020, included 10 carriers of the PSEN1 E280A mutation who were cognitively unimpaired and 10 healthy noncarrier family members, all leveraged from a homogenous Colombian kindred. Statistical analysis was conducted from September 9, 2017, to September 16, 2020. Main Outcomes and Measures: Mixed-effects multiple linear regression was performed to compare the thickness values of the whole retina and individual retinal layers on OCT scans between mutation carriers and noncarriers. Simple linear-effects and mixed-effects multiple linear regression models were used to assess whether age was an effect modifier for PSEN1 mutation of amyloid β levels and retinal thickness, respectively. Fundus photographs were used to compare the number of arterial and venous branch points, arterial and venous tortuosity, and fractal dimension. Results: This study included 10 carriers of the PSEN1 E280A mutation who were cognitively unimpaired (7 women [70%]; mean [SD] age, 36.3 [8.1] years) and 10 healthy noncarrier family members (7 women [70%]; mean [SD] age, 36.4 [8.2] years). Compared with noncarrier controls, PSEN1 mutation carriers who were cognitively unimpaired had a generalized decrease in thickness of the whole retina as well as individual layers detected on OCT scans, with the inner nuclear layer (outer superior quadrant, β = -3.06; P = .007; outer inferior quadrant, β = -2.60; P = .02), outer plexiform layer (outer superior quadrant, β = -3.44; P = .03), and outer nuclear layer (central quadrant, β = -8.61; P = .03; inner nasal quadrant, β = -8.39; P = .04; inner temporal quadrant, β = -9.39; P = .02) showing the greatest amount of statistically significant thinning. Age was a significant effect modifier for the association between PSEN1 mutation and amyloid β levels in cortical regions (β = 0.03; P = .001) but not for the association between PSEN1 mutation and retinal thickness. No statistical difference was detected in any of the vascular parameters studied. Conclusions and Relevance: These findings suggest that OCT can detect functional and morphologic changes in the retina of carriers of familial Alzheimer disease who are cognitively unimpaired several years before clinical onset, suggesting that OCT findings and retinal vascular parameters may be biomarkers prior to the onset of cognitive decline.
Lu Y, Wang JC, Cui Y, Zhu Y, Zeng R, Lu ES, Katz R, Husain D, Vavvas DG, Kim LA, Miller JW, Miller JB. A quantitative comparison of four optical coherence tomography angiography devices in healthy eyes. Graefes Arch Clin Exp Ophthalmol 2021;259(6):1493-1501.Abstract
PURPOSE: Optical coherence tomography angiography (OCT-A) is a novel imaging modality for the diagnosis of chorioretinal diseases. A number of FDA-approved OCT-A devices are currently commercially available, each with unique algorithms and scanning protocols. Although several published studies have compared different combinations of OCT-A machines, there is a lack of agreement on the consistency of measurements across OCT-A devices. Therefore, we conducted a prospective quantitative comparison of four available OCT-A platforms. METHODS: Subjects were scanned on four devices: Optovue RTVue-XR, Heidelberg Spectralis OCT2 module, Zeiss Plex Elite 9000 Swept-Source OCT, and Topcon DRI-OCT Triton Swept-Source OCT. 3 mm × 3 mm images were utilized for analysis. Foveal avascular zone (FAZ) area was separately and independently measured by two investigators. Fractal dimension (FD), superficial capillary plexus (SCP), and deep capillary plexus (DCP) vessel densities (VD) were calculated from binarized images using the Fiji image processing software. SCP and DCP VD were further calculated after images were skeletonized. Repeated measures ANOVA, post hoc tests, and interclass correlation coefficient (ICC) were performed for statistical analysis. RESULTS: Sixteen healthy eyes from sixteen patients were scanned on the four devices. Images of five eyes from the Triton device were excluded due to poor image quality; thus, the authors performed two sets comparisons, one with and one without the Triton machine. FAZ area showed no significant difference across devices with an ICC of > 95%. However, there were statistically significant differences for SCP and DCP VD both before and after skeletonization (p < 0.05). Fractal analysis revealed no significant difference of FD at the SCP; however, a statistically significant difference was found for FD at the DCP layer (p < 0.05). CONCLUSIONS: The results showed that FAZ measurements were consistent across all four devices, while significant differences in VD and FD measurements existed. Therefore, we suggest that for both clinical follow-up and research studies, FAZ area is a useful parameter for OCT-A image analysis when measurements are made on different machines, while VD and FD show significant variability when measured across devices.
Cui Y, Zhu Y, Wang JC, Lu Y, Zeng R, Katz R, Vingopoulos F, Le R, Laíns I, Wu DM, Eliott D, Vavvas DG, Husain D, Miller JW, Kim LA, Miller JB. Comparison of widefield swept-source optical coherence tomography angiography with ultra-widefield colour fundus photography and fluorescein angiography for detection of lesions in diabetic retinopathy. Br J Ophthalmol 2021;105(4):577-581.Abstract
AIMS: To compare widefield swept-source optical coherence tomography angiography (WF SS-OCTA) with ultra-widefield colour fundus photography (UWF CFP) and fluorescein angiography (UWF FA) for detecting diabetic retinopathy (DR) lesions. METHODS: This prospective, observational study was conducted at Massachusetts Eye and Ear from December 2018 to October 2019. Proliferative DR, non-proliferative DR and diabetic patients with no DR were included. All patients were imaged with a WF SS-OCTA using a Montage 15×15 mm scan. UWF CFP and UWF FA were taken by a 200°, single capture retinal imaging system. Images were independently evaluated for the presence or absence of DR lesions including microaneurysms (MAs), intraretinal microvascular abnormalities (IRMAs), neovascularisation elsewhere (NVE), neovascularisation of the optic disc (NVD) and non-perfusion areas (NPAs). All statistical analyses were performed using SPSS V.25.0. RESULTS: One hundred and fifty-two eyes of 101 participants were included in the study. When compared with UWF CFP, WF SS-OCTA was found to be superior in detecting IRMAs (p<0.001) and NVE/NVD (p=0.007). The detection rates of MAs, IRMAs, NVE/NVD and NPAs in WF SS-OCTA were comparable with UWF FA images (p>0.05). Furthermore, when we compared WF SS-OCTA plus UWF CFP with UWF FA, the detection rates of MAs, IRMAs, NVE/NVD and NPAs were identical (p>0.005). Agreement (κ=0.916) between OCTA and FA in classifying DR was excellent. CONCLUSION: WF SS-OCTA is useful for identification of DR lesions. WF SS-OCTA plus UWF CFP may offer a less invasive alternative to FA for DR diagnosis.
Kras A, Celi LA, Miller JB. Accelerating ophthalmic artificial intelligence research: the role of an open access data repository. Curr Opin Ophthalmol 2020;31(5):337-350.Abstract
PURPOSE OF REVIEW: Artificial intelligence has already provided multiple clinically relevant applications in ophthalmology. Yet, the explosion of nonstandardized reporting of high-performing algorithms are rendered useless without robust and streamlined implementation guidelines. The development of protocols and checklists will accelerate the translation of research publications to impact on patient care. RECENT FINDINGS: Beyond technological scepticism, we lack uniformity in analysing algorithmic performance generalizability, and benchmarking impacts across clinical settings. No regulatory guardrails have been set to minimize bias or optimize interpretability; no consensus clinical acceptability thresholds or systematized postdeployment monitoring has been set. Moreover, stakeholders with misaligned incentives deepen the landscape complexity especially when it comes to the requisite data integration and harmonization to advance the field. Therefore, despite increasing algorithmic accuracy and commoditization, the infamous 'implementation gap' persists. Open clinical data repositories have been shown to rapidly accelerate research, minimize redundancies and disseminate the expertise and knowledge required to overcome existing barriers. Drawing upon the longstanding success of existing governance frameworks and robust data use and sharing agreements, the ophthalmic community has tremendous opportunity in ushering artificial intelligence into medicine. By collaboratively building a powerful resource of open, anonymized multimodal ophthalmic data, the next generation of clinicians can advance data-driven eye care in unprecedented ways. SUMMARY: This piece demonstrates that with readily accessible data, immense progress can be achieved clinically and methodologically to realize artificial intelligence's impact on clinical care. Exponentially progressive network effects can be seen by consolidating, curating and distributing data amongst both clinicians and data scientists.
Park EA, Tsikata E, Lee JJ, Shieh E, Braaf B, Vakoc BJ, Bouma BE, de Boer JF, Chen TC. Artifact Rates for 2D Retinal Nerve Fiber Layer Thickness Versus 3D Neuroretinal Rim Thickness Using Spectral-Domain Optical Coherence Tomography. Transl Vis Sci Technol 2020;9(10):10.Abstract
Purpose: To compare the rates of clinically significant artifacts for two-dimensional peripapillary retinal nerve fiber layer (RNFL) thickness versus three-dimensional (3D) neuroretinal rim thickness using spectral-domain optical coherence tomography (SD-OCT). Methods: Only one eye per patient was used for analysis of 120 glaucoma patients and 114 normal patients. For RNFL scans and optic nerve scans, 15 artifact types were calculated per B-scan and per eye. Neuroretinal rim tissue was quantified by the minimum distance band (MDB). Global MDB neuroretinal rim thicknesses were calculated before and after manual deletion of B-scans with artifacts and subsequent automated interpolation. A clinically significant artifact was defined as one requiring manual correction or repeat scanning. Results: Among glaucomatous eyes, artifact rates per B-scan were significantly more common in RNFL scans (61.7%, 74 of 120) compared to B-scans in neuroretinal rim volume scans (20.9%, 1423 of 6820) (95% confidence interval [CI], 31.6-50.0; < 0.0001). For clinically significant artifact rates per eye, optic nerve scans had significantly fewer artifacts (15.8% of glaucomatous eyes, 13.2% of normal eyes) compared to RNFL scans (61.7% of glaucomatous eyes, 25.4% of normal eyes) (glaucoma group: 95% CI, 34.1-57.5, < 0.0001; normal group: 95% CI, 1.3-23.3, = 0.03). Conclusions: Compared to the most commonly used RNFL thickness scans, optic nerve volume scans less frequently require manual correction or repeat scanning to obtain accurate measurements. Translational Relevance: This paper illustrates the potential for 3D OCT algorithms to improve in vivo imaging in glaucoma.

Pages