Participants

Patients presenting macular diseases with impairment of the photoreceptor layer observed on spectral-domain OCT cross-sectional B-scan were studied. All patients underwent comprehensive ophthalmologic examination including best-corrected visual acuity measurement and slit-lamp fundus examination. Ocular axial length was measured using an IOLMaster 500 (Carl Zeiss Meditec, Dublin, California, USA); spherical equivalent refractive errors were measured using the TONOREF II autorefractometer (Nidek, Aichi, Japan).

All patients underwent spectral-domain OCT imaging (Spectralis; Heidelberg Engineering, Heidelberg, Germany). The protocol consisted of acquiring a retinal volume of 21 consecutive horizontal 20-degree (length) lines with 1024 A-scans/B-scans. At least 9 B-scans per image were averaged by the Tracking ART (Automatic Real-Time) software. Only scans with a signal-to-noise ratio above 30 dB were kept for analysis.

Exclusion criteria for this study aimed to obtain the best quality of retinal images for analysis. They included any condition impairing the media clarity, myopia (≥3 diopters spherical equivalent), astigmatism known to produce high-order aberrations, patients presenting a significant tear film disorder (tear breakup time <10 s), insufficient pharmacologically induced mydriasis (pupil diameter <5 mm), uncontrolled movements of the eyes or the head, and poor fixation. Also excluded were patients presenting intraretinal or subretinal fluid, massive exudates, hemorrhage, and pigment changes masking the outer retinal structures.

Quantitative Analysis of Inner Segment Ellipsoid Band Reflectivity and Thickness

Images were analyzed using ImageJ (National Institutes of Health, Bethesda, Maryland, USA). Scans were exported through the built-in export function of the OCT software with an 8-bit color depth (256 grayscale values). Images were used as exported for analysis and not changed in brightness or contrast. Both parameters were preset by the OCT software. Four distinct peaks within the longitudinal reflectivity profiles were detected and the one corresponding to the inner segment ellipsoid band was analyzed. Inner segment ellipsoid band reflectivity was created by averaging pixels that corresponded to the 0.3-degree portion of the inner segment ellipsoid band in the region of interest across each level.

Measurements were taken at 1.5 degrees eccentricity from the fovea ( Figure 1 ) in nasal, temporal, superior, and inferior areas along the horizontal and vertical axis crossing the fovea. For each area, 3 measurements were taken in 3 adjacent B-scans. The data presented correspond to the mean of these 3 measurements for each area.

Reflectivity of the outer retina on spectral-domain optical coherence tomography imaging. Retinal longitudinal reflectivity profile in a healthy eye (Top) and in an eye with hydroxychloroquine-related maculopathy (Bottom). Dotted lines correspond to the inner segment ellipsoid band reflectivity, which is dramatically decreased in the diseased retina. The 90-μm-width vertical band corresponds to the region of interest, where the measurements were performed.

In addition, relative reflectivity was measured. The latter corresponded to the ratio of inner segment ellipsoid band reflectivity on overall retinal reflectivity.

For the inner segment ellipsoid band thickness measurement, images were evaluated independently by 2 readers (M.S., M.F.), both unaware of clinical information. When the evaluations did not agree (difference >2 μm), the opinion of a third author (B.D.) was requested. In practice, on the Spectralis viewing module, the calipers were manually moved to the borders of the inner segment ellipsoid band. The thinnest point of the band in the area studied was determined. The accuracy of this procedure was assessed in a previous paper.

Adaptive Optics Imaging

Eyes were imaged using a flood illumination adaptive optics camera (RTX-1; Imagine Eyes, Orsay, France). This commercial clinical prototype, approved for research use only in Europe, is based on a noncoherent flood-illuminated design with an 850 nm central illumination wavelength allowing imaging of macular photoreceptors with a lateral resolution of 2 μm. Its low-noise charge-coupled-device camera has a pixel resolution of 1.6 μm and a frame rate of 9.5 fps.

Image Acquisition and Processing Procedures

Adaptive optics imaging sessions were conducted after the pupils were dilated with 1 drop each of 0.5% tropicamide. The retinal area imaged was centered on the fovea, resulting in an area 4 degrees square. The imaging depth was chosen within the range of 0–80 μm in which the live retina images appeared to be the sharpest. On the device control panel, a numerical value informed the operator about the adaptive optics correction level. The lower the value was, the better the correction, providing better acquisition quality. During the approximately 30-s acquisition, 40 live high-resolution images of the retina were automatically averaged by the device.

Image Processing

Each series of 40 images acquired by the adaptive optics camera was processed using software programs provided by the system’s manufacturer (CK v0.1 and AOdetect v0.1; Imagine Eyes). These images were registered using a cross-correlation method and averaged to produce a final image with an improved signal-to-noise ratio. The raw images that showed artifacts due to eye blinking and saccades were automatically eliminated before averaging. For display and printing purposes, the background of the resulting image was subtracted using a Gaussian filter and the histogram was stretched over a 16-bit range of gray levels. The positions of photoreceptors were computed by automatically detecting the central coordinates of small circular spots whose brightness was higher than the surrounding background level. First, the averaged image, as obtained before background removal and histogram stretching, was further processed using adaptive and multiple-scale digital filters. Then the local maxima of the resulting filtered image were detected and their pixel coordinates were recorded. The area of analysis corresponded to 4 squares (0.3 × 0.3 degree) placed at a fixed distance from the fovea (1.5 degrees), respectively superiorly, inferiorly, nasally, and temporally along the horizontal and vertical axes crossing the fovea of the eye studied ( Figure 2 ). The measurement protocol was previously described and was shown to be reproducible in normal eyes. The estimates of cone density were verified by 3 investigators (P.T., M.S., M.M.) to minimize any potential cone under- or oversampling made by the automated software. The cone density presented corresponds to the mean of these 3 measurements.

Cone counting performed with adaptive optics in the right eye of a healthy volunteer. Adaptive optics counts corresponded to 4 0.3-degree squares, placed at a fixed distance (1.5 degrees) superiorly, inferiorly, nasally, and temporally from the fovea. (Left) Raw adaptive optics image (4 × 4 degrees); (Right) Cone density obtained after image processing using the manufacturer’s software with the corresponding colored scale.

It is important to note that all the processing and detection steps were fully automated ( Figures 3 and 4 ). The only significant intervention of the grader was the placement of the square, which determines the measurement area.

Comparison of cone counting with adaptive optics and measurement of optical coherence tomography outer retinal reflectivity in a 58-year-old male patient (Patient #4) with chronic central serous chorioretinopathy in the right eye (left images) and in the healthy left eye of a 31-year-old female patient (Patient #1) (right images). (Top) Adaptive optics images centered on the fovea. Squares represent the regions of interest where cones were measured with the corresponding cone densities represented on colored scale. (Bottom) Horizontal spectral-domain OCT scans passing through the fovea showing inner segment ellipsoid band hyporeflectivity in the impaired eye compared with the normal eye.

Illustration of the differences in cone densities measured with adaptive optics and inner segment ellipsoid band reflectivities on spectral-domain optical coherence tomography imaging. (Top) Central serous chorioretinopathy with a persisting pigment epithelium detachment (cone density = 18 567/mm ² ). The area studied was located at the top of the pigment epithelium detachment (107 μm) in a relatively flat section. (Middle) Another case with a smaller pigment epithelium detachment (67 μm) (cone density = 22 846/mm ² ). The area studied was located on the slight slope of the pigment epithelium detachment. (Bottom) Healthy eye (cone density = 29 076/mm ² ).

Statistical Analysis

All values represent the mean ± standard deviation. Correlations with calculation of the Pearson r and linear regression were performed.

Statistics were calculated using a commercially available software program (SPSS, ver. 20.0; SPSS Inc, Chicago, Illinois, USA). P < .05 was considered statistically significant.