11 Segmentation and Visualization of Drusen …
329
to localize the RPE. Using manually annotating pathologic features in a full SDOCT stack images is time-consuming and costly in a clinical setting, which required
quick and precise results. However, the methods by Stopa [29] lacks information
about drusen thickness, which is needed for characterizing drusen. Georczynska
[38] presented a more robust approach using each single OCT cube by selectively
summing different retinal depth levels to generate a series of projection OCT fundus
images, which enhanced contrast and visualized outer retinal pathology that are
not visible with standard fundus imaging or OCT fundus imaging techniques. This
technique separated drusen into several-projected fundus images summed at different
retinal depth levels, hence this cannot be directly visualized in a single image. The
evolution of the GA that causes a deformation in the main retinal layer is still not very
clear; either the RPE, choriocapillaris, or photoreceptors (PR) layers that can resulted
in layer deformities in patients with GA [74, 75]. Recently, new histopathological
findings suggest that the first place in which the GA appearance can be confirm
is the RPE cell loss, then with ensuing PR cell death and choriocapillaris atrophy
[76–79]. Bearelly [75] researched on the PR-RPE interface in GA using SD-OCT in
an effort to test in vivo whether SD-OCT provides enough resolution for reproducible
measurement of the PR layer at the margins of GA, and if the relationship between
PR layer and RPE at those margins could be delineated successfully. The research
work emphasized the direct association between GA and cell loss or “thinning” of PR
and RPE as seen in SD-OCT images. These GA can also be viewed in en face SVP
fundus images as a bright and more uniform delimited region, due to the mentioned
cell loss and consequent increased penetration of light into the choroid coat, combine
with the constant high reflection of light from the choroid coat [72]. In addition, there
are some specific cases in which the highly reflective retinal layers, above the RPE
complex complicate and obscure GA visualization, making the use of the SVP fundus
imaging technique for GA inspection suboptimal.
In this section, we present a new combined method for drusen and GA visualization that enhances the conspicuity of GA lesion by utilizing RPE loss and increase of
reflections from the choroid coat. Mores so, a false color fusion technique by combining drusen and GA projection images is presented to accurately and effectively
display drusen and GA in a single fundus image.
11.3.4.1 Drusen Visualization
Recently we presented the RSVP method [35], which is a fully automated technique
with no user input such as indicating a seed point. This approach restricted the
projected volume to the sub-volume in the vicinity of the RPE layer of a 3D SDOCT to create an en face voxel projection image. Figure 11.1a shows an example of
the vicinity region which is projected in the RSVP method.
The presence of drusen was taken into account in order to determine the location
of the RPE layers. A bilateral filter [61] was first applied to smoothen the SD-OCT
retinal images. After which, a thresholding method was applied to detect the margin
of the vitreous, which is used in estimating the location of the RNFL. The highly
329
to localize the RPE. Using manually annotating pathologic features in a full SDOCT stack images is time-consuming and costly in a clinical setting, which required
quick and precise results. However, the methods by Stopa [29] lacks information
about drusen thickness, which is needed for characterizing drusen. Georczynska
[38] presented a more robust approach using each single OCT cube by selectively
summing different retinal depth levels to generate a series of projection OCT fundus
images, which enhanced contrast and visualized outer retinal pathology that are
not visible with standard fundus imaging or OCT fundus imaging techniques. This
technique separated drusen into several-projected fundus images summed at different
retinal depth levels, hence this cannot be directly visualized in a single image. The
evolution of the GA that causes a deformation in the main retinal layer is still not very
clear; either the RPE, choriocapillaris, or photoreceptors (PR) layers that can resulted
in layer deformities in patients with GA [74, 75]. Recently, new histopathological
findings suggest that the first place in which the GA appearance can be confirm
is the RPE cell loss, then with ensuing PR cell death and choriocapillaris atrophy
[76–79]. Bearelly [75] researched on the PR-RPE interface in GA using SD-OCT in
an effort to test in vivo whether SD-OCT provides enough resolution for reproducible
measurement of the PR layer at the margins of GA, and if the relationship between
PR layer and RPE at those margins could be delineated successfully. The research
work emphasized the direct association between GA and cell loss or “thinning” of PR
and RPE as seen in SD-OCT images. These GA can also be viewed in en face SVP
fundus images as a bright and more uniform delimited region, due to the mentioned
cell loss and consequent increased penetration of light into the choroid coat, combine
with the constant high reflection of light from the choroid coat [72]. In addition, there
are some specific cases in which the highly reflective retinal layers, above the RPE
complex complicate and obscure GA visualization, making the use of the SVP fundus
imaging technique for GA inspection suboptimal.
In this section, we present a new combined method for drusen and GA visualization that enhances the conspicuity of GA lesion by utilizing RPE loss and increase of
reflections from the choroid coat. Mores so, a false color fusion technique by combining drusen and GA projection images is presented to accurately and effectively
display drusen and GA in a single fundus image.
11.3.4.1 Drusen Visualization
Recently we presented the RSVP method [35], which is a fully automated technique
with no user input such as indicating a seed point. This approach restricted the
projected volume to the sub-volume in the vicinity of the RPE layer of a 3D SDOCT to create an en face voxel projection image. Figure 11.1a shows an example of
the vicinity region which is projected in the RSVP method.
The presence of drusen was taken into account in order to determine the location
of the RPE layers. A bilateral filter [61] was first applied to smoothen the SD-OCT
retinal images. After which, a thresholding method was applied to detect the margin
of the vitreous, which is used in estimating the location of the RNFL. The highly