7 Quantitative Analysis of Retinal Layers’ Optical Intensities …
171
they also evaluated the effects of age, sex, height, weight, refractive status, axial
length, image quality, optic disc area and rim/disc area ratio on optical intensity.
In the third study [27], the optical intensity of each retinal layer in central retinal
artery occlusion (CRAO) was quantitatively investigated and compared with the same
measurements obtained from normal controls.
7.2 Automatic Layer Segmentation in OCT Images
The segmentation of retinal layers in OCT scans has been an important work since we
want to establish the distribution and variations of the intensity in the retinal layers.
Eleven surfaces were automatically segmented using a validated 3D graph search
approach [20–23]. The workflow included preprocessing and layer segmentation. The
preprocessing part was a denoising step. OCT data speckle noise was first reduced by
a curvature anisotropic diffusion filter. Retinal boundaries were later automatically
detected by finding an optimal closed set in a vertex-weighted graph. The RNFL,
ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL),
outer plexiform layer (OPL), outer nuclear layer (ONL), photoreceptor layer and
retinal pigment epithelium (RPE) were identified (Fig. 7.2). Every B-scan image was
visually inspected by an ophthalmologist (B.C.) and excluded if any misidentification
of boundaries between retinal layers occurred. Choroid was defined as the region
within 25 pixels below the RPE (Fig. 7.1).
7.3 The Optical Intensity of Retinal Layers of Normal
Subjects
7.3.1 Data Acquisition
Forty normal subjects were included in the first study. Spectral domain OCT scanning
was obtained using the Topcon 3D OCT-1000 (Topcon Corporation, Tokyo, Japan).
6 × 6 mm macular-centered OCT volumes were obtained, which consisted of 64
B scans. Axial and transverse resolution was 6 and 20 μm respectively. A fundus
photograph was obtained at the same time. The image was 512 × 64 × 480 voxels,
with each voxel corresponding to 11.72 × 93.75 × 3.50 μm
3 . The image quality index
was given by the OCT software and ranged from 0 to 100. Raw data were exported
with intensity ranged from 0 to 65,535. Intensity was expressed in arbitrary units
(AU).
171
they also evaluated the effects of age, sex, height, weight, refractive status, axial
length, image quality, optic disc area and rim/disc area ratio on optical intensity.
In the third study [27], the optical intensity of each retinal layer in central retinal
artery occlusion (CRAO) was quantitatively investigated and compared with the same
measurements obtained from normal controls.
7.2 Automatic Layer Segmentation in OCT Images
The segmentation of retinal layers in OCT scans has been an important work since we
want to establish the distribution and variations of the intensity in the retinal layers.
Eleven surfaces were automatically segmented using a validated 3D graph search
approach [20–23]. The workflow included preprocessing and layer segmentation. The
preprocessing part was a denoising step. OCT data speckle noise was first reduced by
a curvature anisotropic diffusion filter. Retinal boundaries were later automatically
detected by finding an optimal closed set in a vertex-weighted graph. The RNFL,
ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL),
outer plexiform layer (OPL), outer nuclear layer (ONL), photoreceptor layer and
retinal pigment epithelium (RPE) were identified (Fig. 7.2). Every B-scan image was
visually inspected by an ophthalmologist (B.C.) and excluded if any misidentification
of boundaries between retinal layers occurred. Choroid was defined as the region
within 25 pixels below the RPE (Fig. 7.1).
7.3 The Optical Intensity of Retinal Layers of Normal
Subjects
7.3.1 Data Acquisition
Forty normal subjects were included in the first study. Spectral domain OCT scanning
was obtained using the Topcon 3D OCT-1000 (Topcon Corporation, Tokyo, Japan).
6 × 6 mm macular-centered OCT volumes were obtained, which consisted of 64
B scans. Axial and transverse resolution was 6 and 20 μm respectively. A fundus
photograph was obtained at the same time. The image was 512 × 64 × 480 voxels,
with each voxel corresponding to 11.72 × 93.75 × 3.50 μm
3 . The image quality index
was given by the OCT software and ranged from 0 to 100. Raw data were exported
with intensity ranged from 0 to 65,535. Intensity was expressed in arbitrary units
(AU).