SD-OCT to Differentiate Traumatic Submacular Hemorrhage Types Using Automatic Three-Dimensional Segmentation Analysis

Introduction

Submacular hemorrhage is a vision-threatening condition in which immediate and appropriate management may restore useful vision in select patients. Exact anatomical location of the submacular hemorrhage may be between the neurosensory retina and retinal pigment epithelium (RPE) or below the RPE. Although both types of submacular hemorrhage may present with significant reduction in vision, their prognosis and management would differ significantly. It is often difficult to differentiate the type of submacular hemorrhage based on clinical examination and fluorescein angiography because they can appear similar. We present the high-definition spectral-domain optical coherence tomography (SD-OCT) features to differentiate the two types of subretinal hemorrhages using the automatic three-dimensional segmentation of layers in two patients with traumatic submacular hemorrhage. Sequential SD-OCT changes during observation in one patient and following pneumatic displacement in the other are described. Use of en face C-scan imaging to diagnose choroidal rupture is also discussed.

Case Reports

Case 1

A 32-year-old man presented following right eye trauma with a tennis ball. His vision was hand motions with accurate light projection. The patient had dispersed hyphema and the fundus details were hazy. Digital intraocular pressure (IOP) was high. The patient was prescribed topical lubricant drops and antiglaucoma medications along with 1 mg/kg of oral prednisolone. The next day, his best-corrected visual acuity (BCVA) improved to counting fingers at 1 m with normal IOP. Indirect ophthalmoscopy revealed a grayish elevation at the macula. SD-OCT raster scan with Optovue RTvue-100 (Optovue, Fremont, CA) showed subretinal elevation that was below the RPE and involved the foveal center (Fig. 1). Because the hemorrhage was sub-RPE, we decided to observe the patient. Three days later, BCVA improved to 20/200, hyphema had reduced, and the subretinal hemorrhage was well delineated. A 512 × 128 macular cube scan (Cirrus HD-OCT model 4000; Carl Zeiss Meditec, Dublin, CA) was performed. Advanced analysis using a ray-traced three-dimensional rendering system showed an elevation of the RPE layer with a corresponding elevation of the internal limiting membrane (ILM), whereas the ILM–RPE thickness was normal, indicating that the hemorrhage was below the RPE (Fig. 2). Sequential three-dimensional segmentation maps and high-definition raster scans clearly showed the reducing height of the sub-RPE bleed, which correlated with improving vision (Figs. 3 and 4). On follow-up his visual acuity improved to 20/120 at 7 days, 20/60 at 17 days, and 20/40 at 35 days after injury. Subretinal hemorrhage resolved after 3 weeks, revealing the choroidal rupture. At 3 months, his BCVA was 20/25 with normal IOP without any medication.

Figure 1. Spectral Domain Optical Coherence Tomography Raster Scan Shows Submacular Elevation Involving the Fovea. High Backscattering of the Retinal Pigment Epithelium Is Clearly Seen Nasal to the Fovea with Preservation of Retinal Architecture. The Arrow Indicates the Direction of the Line Scan (Temporal to Nasal).

Figure 2. Case 1. (A) Fundus Photograph 3 Days After Trauma Showing the Extent of the Submacular Hemorrhage. The Temporal Edge of the Hemorrhage Appeared to Be Clearing, Indicating the Previous Extent of the Hemorrhage (white Arrows), Whereas the Arrow Head Shows the New Edge. (B) Advanced Three-Dimensional Segmentation Analysis of the 512 × 128 Macular Cube Scan Shows Elevation of Both the Internal Limiting Membrane (ILM) and Retinal Pigment Epithelium (RPE) Due to the Submacular Hemorrhage. (C) ILM Segmentation Map. (D) RPE Segmentation Map. (E) ILM–RPE Thickness Map Shows Normal Retinal Thickness in the Area Corresponding to the Submacular Bleed.

Figure 3. Case 1. Sequential Fundus Photographs and Corresponding Three-Dimensional Retinal Pigment Epithelium Segmentation Maps Show the Reducing Height of the Submacular Bleed in the First Case. (A) Three Days After Trauma. (B) Seven Days After Trauma. (C) Seventeen Days After Trauma. (D) Thirty-Five Days After Trauma.

Figure 4. Case 1. Sequential High-Definition Raster Scans Show the Decreasing Height of the Submacular Bleed at Each Follow-Up. (A) Three Days After Trauma (height = 353 μm). (B) Seven Days After Trauma (height = 238 μm). (C) Seventeen Days After Trauma (height =101 μm).(D) Thirty-Five Days After Trauma (height = 36 μm).

Case 2

A 16-year-old boy presented for a second opinion 7 days after left eye trauma with a tennis ball. On initial presentation to another center, his visual acuity was noted to be counting fingers and he was noted to have iritis with Berlin’s edema. The patient was prescribed oral steroids (40 mg daily) with a topical antibiotic steroid eye drop combination administered every 5 hours. The patient sought a second opinion because there was no improvement in visual acuity. On examination, his vision was counting fingers close to face and his IOP was 29 mm Hg. Anterior chamber showed 2+ cells, whereas fundus revealed a reddish elevation at the macula involving the fovea that was suggestive of subretinal hemorrhage (Fig. 5). On three-dimensional segmentation analysis of 512 × 128 macular cube scan, an elevation of the ILM with a flat RPE layer was noted (Fig. 5). ILM–RPE thickness was increased, indicating the blood was between the neurosensory retina and the RPE layer (Fig. 5). The patient underwent 0.5-cc intravitreal SF₆ gas injection with prone positioning. Serial three-dimensional OCT imaging at 6 hours, 36 hours, 7 days, and 12 days after injection revealed displacement of the bleed, evidenced by shifting of the elevation and subsequent flattening (Figs. 6 and 7). Fundus examination 1 week after injection revealed choroidal rupture. The site of the choroidal rupture corresponded to the peripapillary RPE elevation seen on the RPE segmentation map (Fig. 5). Three months after pneumatic displacement, his BCVA was 20/50.

Figure 5. Case 2. (A) Fundus Photograph 7 Days After Trauma Showing the Extent of the Submacular Hemorrhage. (B) Advanced Three-Dimensional Segmentation Analysis of the 512 × 128 Macular Cube Scan Shows Elevation of the Internal Limiting Membrane (ILM), Whereas the Retinal Pigment Epithelium (RPE) Is Flat Except for Small Elevation (oval Dashed Line). (C) ILM Segmentation Map. (D) RPE Segmentation Map Showing Small Elevation (oval Dashed Line) that Corresponded to Choroidal Rupture Detected Later. (E) ILM–RPE Thickness Map Shows Increased Retinal Thickness in the Area Corresponding to the Submacular Bleed.

Figure 6. Case 2. Sequential Fundus Photographs Show Displacement of Submacular Bleed Following Gas Injection. Progressive Displaced Edge of the Bleed at Each Visit (white Arrows) and Unmasking of Choroidal Rupture Are Clearly Seen. (A) Seven Days After Injury. (B) Thirty-Six Hours After Gas Injection. (C) Seven Days After Gas Injection. (D) Twelve Days After Gas Injection.

Figure 7. Case 2. Sequential High-Definition Raster Scans Show the Decreasing Height of the Submacular Bleed at Each Follow-Up. Corresponding Internal Limiting Membrane Segmentation Maps Are Shown in the Insert on the Right Side. (A) Seven Days After Trauma (height = 104 μm). (B) Six Hours After Gas Injection (height = 148 μm). (C) Thirty-Six Hours After Gas Injection (height = 72 μm). (D) Twelve Days After Gas Injection (height = 33 μm).

Discussion

Submacular hemorrhage is a condition that threatens vision. Natural history and visual prognosis depend on the cause of hemorrhage, the duration of submacular hemorrhage prior to treatment, and the location and thickness of the subretinal bleed.

In a retrospective review of natural history and visual prognosis in 29 eyes with subretinal hemorrhage, Bennett et al.¹ reported a mean final BCVA of 20/480 with an average 3-year follow-up. Eyes with age-related macular degeneration had a poor final BCVA of 20/1700, whereas eyes with subretinal hemorrhage from trauma-induced choroidal rupture had a better BCVA of 20/35.

The pretreatment duration of the subretinal hemorrhage correlates inversely with final visual acuity. In an interventional study, Hattenbach et al.² found that final visual acuity was better in eyes with subretinal hemorrhage of less than 2 weeks’ duration before treatment. Similar findings were reported by Lewis³ after surgical evacuation and the final BCVA was worse in patients with hemorrhage lasting longer than 7 days. Toxicity due to iron could be partially responsible for retinal damage apart from trauma itself, as shown in experimental studies.⁴

Prognosis also depends on the thickness and location of the bleed. Increased thickness of subretinal hemorrhage has been correlated with poorer final visual acuity.^1,5,6 This is because a larger separation between the photoreceptors and the RPE correlates with a greater diffusion barrier, therefore resulting in a higher chance of permanent photoreceptor damage. The larger the separation between the photoreceptors and the RPE, the greater the diffusion barrier will be, resulting in a higher chance of permanent photoreceptor damage.

Although studies have looked at both the location (foveal involvement) and the clinical assessment of hemorrhage thickness to prognosticate outcome, the exact anatomical location has not been investigated directly with ancillary testing. Submacular hemorrhage may be present between the photoreceptors and RPE or below the RPE. Clinical examination or fluorescein angiography may not differentiate the type of submacular hemorrhage with certainty. In our cases, we were able to accurately diagnose the exact anatomical location of the subretinal bleed using three-dimensional segmentation analysis of Cirrus HD-OCT. This differentiation is important because both types of hemorrhage may significantly reduce vision; however, prognosis and approach to their management would differ. In patients with hemorrhage between the neurosensory retina and RPE, immediate management, including either surgical evacuation or gas injection to displace the blood away from the fovea, needs to be initiated. In patients with a sub-RPE bleed, it may be difficult to displace the blood because it is a closed space. Furthermore, sub-RPE blood cannot be removed with surgery without creating iatrogenic damage to the RPE with resultant poor vision. Hence, observation would be prudent in such cases.

Both our patients had similar clinical presentation. The maximum foveal elevation on high-definition raster scan was higher in the first case (353 microns) compared with the second case (104 microns), and should therefore have had a poorer prognosis without therapeutic measures to displace the blood. However, the visual recovery started as early as 3 days after injury and continued to improve, which correlated with a decrease in the thickness of the sub-RPE bleed. Chaudhry et al.⁷ reported a similar case of spontaneous resolution of large traumatic submacular hemorrhage. The case reported by Chaudhry et al.⁷ and our first case defy our logic based on animal studies⁴ that have revealed irreversible iron-related damage happening within 24 hours of the onset of submacular hemorrhage. One possible explanation is that the photoreceptors were not directly exposed to the iron-induced damage because the blood was sub-RPE.

To the best of our knowledge, use of SD-OCT with automatic three-dimensional segmentation analysis to differentiate submacular hemorrhage types and its serial use to objectively monitor hemorrhage resolution following observation/gas injection has not yet been reported. Apart from establishing the anatomical location of the bleed, utility of an RPE-fitted slab to visualize retinochoroidal interface was demonstrated. This feature allowed en face visualization (C-scan) of the choroidal rupture in both cases that was hidden to clinical evaluation due to hemorrhage (Fig. 8A). C-scan imaging of the outer segment/inner segment junction was also done at the final visit, which revealed reduced density of photoreceptors in the area immediately above the hemorrhage. C-scan of the RPE and sub-RPE level revealed a relatively intact RPE with altered sub-RPE in case 1, whereas the RPE uniformity was altered but the sub-RPE appeared intact in the second case (Figs. 8B, 8C, and 8D).

Figure 8. Advanced Three-Dimensional En Face Visualization (upper Images = Case 1; Lower Images = Case 2). (A) Choroidal Rupture Seen as a Hyperreflective Area on a 20-μm Retinal Pigment Epithelium (RPE) Fitted Slab Placed Just Below the RPE (Case 1: 3 Days After Injury; Case 2: 7 Days After Injury). (B) Inner Segment/outer Segment Photoreceptor Junction 3 Months After Injury Shows Reduced Density of Photoreceptors in Both Cases in the Area of Subretinal Hemorrhage. (C) RPE at 3 Months After Injury Is Relatively Normal in Case 1, Whereas It Shows Areas of Obliteration in Case 2. (D) Sub-RPE 3 Months After Injury Shows a Well-Defined Area of Reduced Reflectivity from Sub-RPE Structures Corresponding to the Area of Hemorrhage in Case 1, Whereas It Shows Minimal Patchy Areas of Reduced Reflectivity in Case 2.

OCT was also used to guide the duration of face-down positioning in the second case. Prone positioning was maintained for 15 days until the subretinal elevation was nearly flat on OCT.

Based on our experience in these two cases, we infer that SD-OCT has an important role in managing submacular hemorrhage in patients. Combined with the ability for automatic segmentation, SD-OCT clearly improves the clinical evaluation of chorioretinal disease by adding relevant new parameters to diagnose the exact anatomical location of lesions and to monitor the patient’s progress toward treatment. Further experience with a larger number of patients is desirable.