Discriminating Between Papilledema and Optic Disc Drusen Using 3D Structural Analysis of the Optic Nerve Head

BACKGROUND AND OBJECTIVES

The distinction of papilledema from other optic nerve head (ONH) lesions mimicking papilledema, such as optic disc drusen (ODD), can be difficult in clinical practice. We aimed the following: (1) to develop a deep learning algorithm to automatically identify major structures of the ONH in 3-dimensional (3D) optical coherence tomography (OCT) scans and (2) to exploit such information to robustly differentiate among ODD, papilledema, and healthy ONHs.

METHODS

This was a cross-sectional comparative study of patients from 3 sites (Singapore, Denmark, and Australia) with confirmed ODD, those with papilledema due to raised intracranial pressure, and healthy controls. Raster scans of the ONH were acquired using OCT imaging and then processed to improve deep-tissue visibility. First, a deep learning algorithm was developed to identify major ONH tissues and ODD regions. The performance of our algorithm was assessed using the Dice coefficient. Second, a classification algorithm (random forest) was designed to perform 3-class classifications (1: ODD, 2: papilledema, and 3: healthy ONHs) strictly from their drusen and prelamina swelling scores (calculated from the segmentations). To assess performance, we reported the area under the receiver operating characteristic curve for each class.

RESULTS

A total of 241 patients (256 imaged ONHs, including 105 ODD, 51 papilledema, and 100 healthy ONHs) were retrospectively included in this study. Using OCT images of the ONH, our segmentation algorithm was able to isolate neural and connective tissues and ODD regions/conglomerates whenever present. This was confirmed by an averaged Dice coefficient of 0.93 ± 0.03 on the test set, corresponding to good segmentation performance. Classification was achieved with high AUCs, that is, 0.99 ± 0.001 for the detection of ODD, 0.99 ± 0.005 for the detection of papilledema, and 0.98 ± 0.01 for the detection of healthy ONHs.

DISCUSSION

Our artificial intelligence approach can discriminate ODD from papilledema, strictly using a single OCT scan of the ONH. Our classification performance was very good in the studied population, with the caveat that validation in a much larger population is warranted. Our approach may have the potential to establish OCT imaging as one of the mainstays of diagnostic imaging for ONH disorders in neuro-ophthalmology, in addition to fundus photography.

Longitudinal assessment of optic nerve head changes using optical coherence tomography in a primate microbead model of ocular hypertension

In humans, the longitudinal characterisation of early optic nerve head (ONH) damage in ocular hypertension (OHT) is difficult as patients with glaucoma usually have structural ONH damage at the time of diagnosis. Previous studies assessed glaucomatous ONH cupping by measuring the anterior lamina cribrosa depth (LCD) and minimal rim width (MRW) using optical coherence tomography (OCT). In this study, we induced OHT by repeated intracameral microbead injections in 16 cynomolgus primates (10 unilateral; 6 bilateral) and assessed the structural changes of the ONH longitudinally to observe early changes. Elevated intraocular pressure (IOP) in OHT eyes was maintained for 7 months and serial OCT measurements were performed during this period. The mean IOP was significantly elevated in OHT eyes when compared to baseline and compared to the control eyes. Thinner MRW and deeper LCD values from baseline were observed in OHT eyes with the greatest changes seen between month 1 and month 2 of OHT. Both the mean and maximum IOP values were significant predictors of MRW and LCD changes, although the maximum IOP was a slightly better predictor. We believe that this model could be useful to study IOP-induced early ONH structural damage which is important for understanding glaucoma pathogenesis.

DeshadowGAN: A Deep Learning Approach to Remove Shadows from Optical Coherence Tomography Images

Purpose

To remove blood vessel shadows from optical coherence tomography (OCT) images of the optic nerve head (ONH).

Methods

Volume scans consisting of 97 horizontal B-scans were acquired through the center of the ONH using a commercial OCT device for both eyes of 13 subjects. A custom generative adversarial network (named DeshadowGAN) was designed and trained with 2328 B-scans in order to remove blood vessel shadows in unseen B-scans. Image quality was assessed qualitatively (for artifacts) and quantitatively using the intralayer contrast—a measure of shadow visibility ranging from 0 (shadow-free) to 1 (strong shadow). This was computed in the retinal nerve fiber layer (RNFL), the inner plexiform layer (IPL), the photoreceptor (PR) layer, and the retinal pigment epithelium (RPE) layer. The performance of DeshadowGAN was also compared with that of compensation, the standard for shadow removal.

Results

DeshadowGAN decreased the intralayer contrast in all tissue layers. On average, the intralayer contrast decreased by 33.7 ± 6.81%, 28.8 ± 10.4%, 35.9 ± 13.0%, and 43.0 ± 19.5% for the RNFL, IPL, PR layer, and RPE layer, respectively, indicating successful shadow removal across all depths. Output images were also free from artifacts commonly observed with compensation.

Conclusions

DeshadowGAN significantly corrected blood vessel shadows in OCT images of the ONH. Our algorithm may be considered as a preprocessing step to improve the performance of a wide range of algorithms including those currently being used for OCT segmentation, denoising, and classification.

Translational Relevance

DeshadowGAN could be integrated to existing OCT devices to improve the diagnosis and prognosis of ocular pathologies.

Corneal elevation changes after forced eyelid closure in healthy participants and in patients with keratoconus

BACKGROUND

To assess pointwise corneal elevation changes after forced eyelid closure test (FECT) in the eyes of healthy subjects and in eyes with keratoconus.

METHODS

Twenty-nine subjects with keratoconus and 31 healthy volunteers were evaluated. Patients with keratoconus who had corneal hydrops, apical scarring, corneal thickness ≤ 400 μm, ocular surface disease, contact lens wear on the examination day and a history of corneal cross-linking were excluded. Exclusion criteria for healthy participants were spherical error > +3.00 D and < -3.00 D, corneal astigmatism > 1.50 D, corneal curvature > 47 D, ocular allergy, clinical findings and family history of keratoconus. Pentacam was performed before and after 20 seconds of FECT and raw data were extracted from the built-in software. Pointwise anterior and posterior elevation changes in the central 8 mm cornea were assessed using paired samples t-test and heat maps were constructed to reflect mean changes and statistically significant data points. Statistical significance was assumed at p < 0.01.

RESULTS

Age and gender were similar between healthy subjects (24.5 ± 1.6 years, 46.4 per cent female) and subjects with keratoconus (28.6 ± 9.2 years, 46.4 per cent female, p = 0.19, 0.61, respectively). Healthy eyes displayed posterior depression clustering in the inferotemporal and inferonasal areas (mean change: -4.5 ± 7.8 μm and -5.2 ± 9.8 μm, respectively, all p < 0.01). In contrast, keratoconus eyes exhibited a wider area of posterior elevation clustering in the inferior cornea (mean change: 8.1 ± 14.5 μm, all p < 0.01) with a small extension in the inferotemporal cornea (mean change: 12.1 ± 22.3 μm, all p < 0.01).

CONCLUSION

FECT elicits corneal elevation changes mainly in the inferior cornea with the change being more pronounced and wider in eyes with keratoconus.

In Vivo 3-Dimensional Strain Mapping Confirms Large Optic Nerve Head Deformations Following Horizontal Eye Movements

Purpose

To measure lamina cribrosa (LC) strains (deformations) following abduction and adduction in healthy subjects and to compare them with those resulting from a relatively high acute intraocular pressure (IOP) elevation.

Methods

A total of 16 eyes from 8 healthy subjects were included. Among the 16 eyes, 11 had peripapillary atrophy (PPA). For each subject, both optic nerve heads (ONHs) were imaged using optical coherence tomography (OCT) at baseline (twice), in different gaze positions (adduction and abduction of 20°) and following an acute IOP elevation of approximately 20 mm Hg from baseline (via ophthalmodynamometry). Strains of LC for all loading scenarios were mapped using a three-dimensional tracking algorithm.

Results

In all 16 eyes, LC strains induced by adduction and abduction were 5.83% ± 3.78% and 3.93% ± 2.57%, respectively, and both significantly higher than the control strains measured from the repeated baseline acquisitions (P < 0.01). Strains of LC in adduction were on average higher than those in abduction, but the difference was not statistically significant (P = 0.07). Strains of LC induced by IOP elevations (on average 21.13 ± 7.61 mm Hg) were 6.41% ± 3.21% and significantly higher than the control strains (P < 0.0005). Gaze-induced LC strains in the PPA group were on average larger than those in the non-PPA group; however, the relationship was not statistically significant.

Conclusions

Our results confirm that horizontal eye movements generate significant ONH strains, which is consistent with our previous estimations using finite element analysis. Further studies are needed to explore a possible link between ONH strains induced by eye movements and axonal loss in optic neuropathies.

Automatic measurements of choroidal thickness in EDI-OCT images

Enhanced Depth Imaging (EDI) optical coherence tomography (OCT) provides high-definition cross-sectional images of the choroid in vivo, and hence is used in many clinical studies. However, measurement of choroidal thickness depends on the manual labeling, which is tedious and subjective of inter-observer differences. In this paper, we propose a fast and accurate algorithm that could measure the choroidal thickness automatically. The lower boundary of the choroid is detected by searching the biggest gradient value above the retinal pigment epithelium (RPE) and the upper boundary is formed by finding the shortest path of the graph formed by valley pixels using dynamic programming. The average of Dice's Coefficient on 10 EDI-OCT images is 94.3%, which shows good consistency of the algorithm with the manual labeling. The processing time for each image is about 2 seconds.

Automatic segmentation of the choroid in enhanced depth imaging optical coherence tomography images

Enhanced Depth Imaging (EDI) optical coherence tomography (OCT) provides high-definition cross-sectional images of the choroid in vivo, and hence is used in many clinical studies. However, the quantification of the choroid depends on the manual labelings of two boundaries, Bruch's membrane and the choroidal-scleral interface. This labeling process is tedious and subjective of inter-observer differences, hence, automatic segmentation of the choroid layer is highly desirable. In this paper, we present a fast and accurate algorithm that could segment the choroid automatically. Bruch's membrane is detected by searching the pixel with the biggest gradient value above the retinal pigment epithelium (RPE) and the choroidal-scleral interface is delineated by finding the shortest path of the graph formed by valley pixels using Dijkstra's algorithm. The experiments comparing automatic segmentation results with the manual labelings are conducted on 45 EDI-OCT images and the average of Dice's Coefficient is 90.5%, which shows good consistency of the algorithm with the manual labelings. The processing time for each image is about 1.25 seconds.