Optical coherence tomography (OCT) manufacturers graphically present circumpapillary retinal nerve fiber layer thickness (cpRNFLT) together with normative limits to support clinicians in diagnosing ophthalmic diseases. The impact of age on cpRNFLT is typically implemented by linear models. cpRNFLT is strongly location-specific, whereas previously published norms are typically restricted to coarse sectors and based on small populations. Furthermore, OCT devices neglect impacts of lens or eye size on the diameter of the cpRNFLT scan circle so that the diameter substantially varies over different eyes. We investigate the impact of age and scan diameter reported by Spectralis spectral-domain OCT on cpRNFLT in 5646 subjects with healthy eyes. We provide cpRNFLT by age and diameter at 768 angular locations. Age/diameter were significantly related to cpRNFLT on 89%/92% of the circle, respectively (pointwise linear regression), and to shifts in cpRNFLT peak locations. For subjects from age 42.1 onward but not below, increasing age significantly decreased scan diameter (r=−0.28, p<0.001), which suggests that pathological cpRNFLT thinning over time may be underestimated in elderly compared to younger subjects, as scan diameter decrease correlated with cpRNFLT increase. Our detailed numerical results may help to generate various correction models to improve diagnosing and monitoring optic neuropathies.
Clinicians use retinal nerve fiber layer thickness (RNFLT) measured by optical coherence tomography (OCT) as an adjunct to glaucoma diagnosis. Ametropia is accompanied by changes to the optic nerve head (ONH), which may affect how OCT machines mark RNFLT measurements as abnormal. These changes in abnormality patterns may bias glaucoma diagnosis. Here, we investigate the relationship between OCT abnormality patterns and the following ONH-related and ametropia-associated parameters on 421 eyes of glaucoma patients: optic disc tilt and torsion, central retinal vessel trunk location (CRVTL), and nasal and temporal retinal curvature adjacent to ONH, quantified as nasal/temporal slopes of the inner limiting membrane. We applied multivariate logistic regression with abnormality marks as regressands to 40,401 locations of the peripapillary region and generated spatial maps of locations of false positive/negative abnormality marks independent of glaucoma severity. Effects of torsion and temporal slope were negligible. The effect of tilt could be explained by covariation with ametropia. For CRVTL/nasal slope, abnormality pattern shifts at 7.2%/23.5% of the peripapillary region were detected, respectively, independent of glaucoma severity and ametropia. Therefore, CRVTL and nasal curvature should be included in OCT RNFLT norms. Our spatial location maps may aid clinicians to improve diagnostic accuracy.
Retinal nerve fiber layer thickness (RNFLT) measured by optical coherence tomography (OCT) is widely used in clinical practice to support glaucoma diagnosis. Clinicians frequently interpret peripapillary RNFLT areas marked as abnormal by OCT machines. However, presently, clinical OCT machines do not take individual retinal anatomy variation into account, and according diagnostic biases have been shown particularly for patients with ametropia. The angle between the two major temporal retinal arteries (interartery angle, IAA) is considered a fundamental retinal ametropia marker. Here, we analyze peripapillary spectral domain OCT RNFLT scans of 691 glaucoma patients and apply multivariate logistic regression to quantitatively compare the diagnostic bias of spherical equivalent (SE) of refractive error and IAA and to identify the precise retinal locations of false-positive/negative abnormality marks. Independent of glaucoma severity (visual field mean deviation), IAA/SE variations biased abnormality marks on OCT RNFLT printouts at 36.7%/22.9% of the peripapillary area, respectively. 17.2% of the biases due to SE are not explained by IAA variation, particularly in inferonasal areas. To conclude, the inclusion of SE and IAA in OCT RNFLT norms would help to increase diagnostic accuracy. Our detailed location maps may help clinicians to reduce diagnostic bias while interpreting retinal OCT scans.
Purpose: To evaluate the effects of four anatomical parameters (angle between superior and inferior temporal retinal
arteries [inter-artery angle, IAA], optic disc [OD] rotation, retinal curvature, and central retinal vessel trunk entry point
location [CRVTL]) on retinal nerve fiber layer thickness (RNFLT) abnormality marks by OCT machines.
Methods: Cirrus OCT circumpapillary RNFLT measurements and Humphrey visual fields (HVF 24-2) of 421 patients
from a large glaucoma clinic were included. Ellipses were fitted to the OD borders. Ellipse rotation relative to the
vertical axis defined OD rotation. CRVTL was manually marked on the horizontal axis of the ellipse on the OCT fundus
image. IAA was calculated between manually marked retinal artery locations at the 1.73mm radius around OD. Retinal
curvature was determined by the inner limiting membrane on the horizontal B-scan closest to the OD center. For each
location on the circumpapillary scanning area, logistic regression was used to determine if each of the four parameters
had a significant impact on RNFLT abnormality marks independent of disease severity. The results are presented on
spatial maps of the entire scanning area.
Results: Variations in IAA significantly influenced abnormality marks on 38.8% of the total scanning area, followed by
CRVTL (19.2%) and retinal curvature (18.7%). The effect of OD rotation was negligible (<1%).
Conclusions: A natural variation in IAA, retinal curvature, and CRVTL can affect OCT abnormality ratings, which
may bias clinical diagnosis. Our spatial maps may help OCT manufacturers to introduce location specific norms to
ensure that abnormality marks indicate ocular disease instead of variations in eye anatomy.
Purpose: To assess whether modeling of central vision loss (CVL) due to glaucoma by optical coherence tomography (OCT) retinal nerve fiber (RNF) layer thickness (RNFLT) can be improved by including the location of the major inferior temporal retinal artery (ITA), a known correlate of individual RNF geometry. Methods: Pat- tern deviations of the two locations of the Humphrey 24-2 visual field (VF) known to be specifically vulnerable to glaucomatous CVL and OCT RNFLT on the corresponding circumpapillary sector around the optic nerve head within the radius of 1.73mm were retrospectively selected from 428 eyes of 428 patients of a large clinical glaucoma service. ITA was marked on the 1.73mm circle by a trained observer. Linear regression models were fitted with CVL as dependent variable and VF mean deviation (MD) plus either of (1) RNFLT, (2) ITA, and
(3) their combination, respectively, as regressors. To assess CVL over all levels of glaucoma severity, the three models were compared to a null model containing only MD. A Baysian model comparison was performed with the Bayes Factor (BF) as measure of strength of evidence (BF<3: no evidence, 3-20: positive evidence, >20: strong evidence over null model). Results: Neither RNFLT (BF=0.9) nor ITA (BF=1.4) alone provided positive evidence over the null model, but their combination resulted in a model with strong evidence (BF=21.4). Conclusion: While the established circumpapillary RNFLT sector, based on population statistics, could not satisfactorily model CVL, the inclusion of a retinal parameter related to individual eye anatomy yielded a strong structure-function model.
Purpose: Optic disc tilt defined over 3D optic disc morphology has been shown to be associated with the location of
initial glaucomatous damages. In this work, we study the impact of optic cup depth (OCD) on spatial patterns of visual
field loss in glaucoma.
Methods: Pairs of reliable Cirrus OCT scans around optic disc and Humphrey visual fields of glaucoma patients without
visually significant cataract and age-related macular degeneration were selected. The most recent visit of a randomly
selected eye of each patient was chosen. The OCD was automatically calculated on the superior-inferior cross sectional
image passing through the optic disc center. The correlations between the mean pattern deviation (PD) of each sector in
glaucoma hemifield test (GHT) and Garway-Heath scheme and OCD were evaluated for all severities glaucoma and mild
glaucoma (mean deviation ≥ -5 dB), respectively.
Results: 424 eyes of 424 patients passed the data reliability criteria with 346 mild glaucoma patients. For all severities
glaucoma, there was no significant correlation between the mean sector PD and OCD. For mild glaucoma, OCD was
uniquely correlated to the mean PD of the inferior pericentral sector (r=-0.18, p=0.01) in GHT, which was independent
of mean deviation and retinal nerve fiber layer thickness (p<0.001 for both).
Conclusion: OCD was uniquely correlated to the vision loss of the inferior pericentral sector in GHT and Garway-
Health scheme for mild glaucoma. Future advancement of OCT imaging techniques may provide better clinical diagnosis
for early glaucoma by focusing on 3D morphological variation of the optic disc.
KEYWORDS: Error analysis, Digital breast tomosynthesis, Breast, Digital imaging, Modeling, Visibility, Tissues, Mammography, Data acquisition, Imaging systems, Radiology, Image segmentation, Performance modeling, Medical imaging, Medicine
Digital breast tomosynthesis (DBT) can improve lesion visibility by eliminating the issue of overlapping breast tissue present in mammography. However, this new modality likely requires new approaches to training. The issue of training in DBT is not well explored. We propose a computer-aided educational approach for DBT training. Our hypothesis is that the trainees’ educational outcomes will improve if they are presented with cases individually selected to address their weaknesses. In this study, we focus on the question of how to select such cases. Specifically, we propose an algorithm that based on previously acquired reading data predicts which lesions will be missed by the trainee for future cases (i.e., we focus on false negative error). A logistic regression classifier was used to predict the likelihood of trainee error and computer-extracted features were used as the predictors. Reader data from 3 expert breast imagers was used to establish the ground truth and reader data from 5 radiology trainees was used to evaluate the algorithm performance with repeated holdout cross validation. Receiver operating characteristic (ROC) analysis was applied to measure the performance of the proposed individual trainee models. The preliminary experimental results for 5 trainees showed the individual trainee models were able to distinguish the lesions that would be detected from those that would be missed with the average area under the ROC curve of 0.639 (95% CI, 0.580-0.698). The proposed algorithm can be used to identify difficult cases for individual trainees.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.