284
Q. Chen et al.
the location of the retinal nerve fiber layer (RNFL) was estimated and then the RPE
layer was localized. By enforcing a local convexity condition and fitting second or
fourth order polynomials to the possibly unhealthy (abnormal) RPE curve, the healthy
(normal) shape of the RPE layer is estimated. The area between the estimated normal
and the segmented RPE outlines was marked as possible drusen. Yi [28] utilized a
similar algorithm to automatically segment drusen. The main difference between
Farsiu’s method and Yi’s method was the extraction of RPE layers. While the prior
work on automated drusen segmentation is a step in the direction of quantitation, there
are unsolved challenges. First, drusen may obscure portions of the image needed for
accurate estimation of RPE layers. Second, RPE layer segmentation is difficult even
in normal patients because the inner segment/outer segment (IS/OS) retinal layers
are often contiguous with the RPE and there is abundant noise in low signal-to-noise
ratio (SNR) OCT images. To counteract these potential problems, [26] provided a
manual correction using a software interface. According to one study [29], drusen
detection in CFP and spectral domain optical coherence tomography (SD-OCT)
images has good concordance, and each imaging modality has its own advantages.
Our algorithm utilizes the projection image to verify and refine the segmentation
results from SD-OCT images, which reduces the influence of RPE estimation error
and improves the robustness of drusen segmentation. Gregori [30] and Iwama [31]
also segmented drusen based on the distance between the abnormal RPE and the
normal RPE floor. Recently, many researchers proposed several other methods [15,
32, 33].
In this section, we tackle the above challenges, and present several novel automated drusen segmentation method in SD-OCT images. Our proposed methods
include: (a) Automated Drusen Segmentation and Quantification in SD-OCT Images
[34], (b) An improved OCT-Derived Fundus Projection Image for Drusen Visualization [35]. In addition, we also provide a means of generating a high-quality projection
image based on RPE layer estimation.
11.2.1 Automated Drusen Segmentation and Quantification
in SD-OCT Images
11.2.1.1 Overview of Proposed Method
We briefly introduce the algorithm, for full detail see [34]. A flowchart of our algorithm is shown in Fig. 11.2, which comprises the following operations on the input
SD-OCT image:
1. Image denoising: A modified bilateral filtering algorithm is used to reduce noise
in order to facilitate the subsequent estimation of the RNFL and RPE retinal
layer.
2. RNFL complex removal: The RNFL complex, defined as the region between
the inner limiting membrane and the outer plexiform layer (indicated by the
Q. Chen et al.
the location of the retinal nerve fiber layer (RNFL) was estimated and then the RPE
layer was localized. By enforcing a local convexity condition and fitting second or
fourth order polynomials to the possibly unhealthy (abnormal) RPE curve, the healthy
(normal) shape of the RPE layer is estimated. The area between the estimated normal
and the segmented RPE outlines was marked as possible drusen. Yi [28] utilized a
similar algorithm to automatically segment drusen. The main difference between
Farsiu’s method and Yi’s method was the extraction of RPE layers. While the prior
work on automated drusen segmentation is a step in the direction of quantitation, there
are unsolved challenges. First, drusen may obscure portions of the image needed for
accurate estimation of RPE layers. Second, RPE layer segmentation is difficult even
in normal patients because the inner segment/outer segment (IS/OS) retinal layers
are often contiguous with the RPE and there is abundant noise in low signal-to-noise
ratio (SNR) OCT images. To counteract these potential problems, [26] provided a
manual correction using a software interface. According to one study [29], drusen
detection in CFP and spectral domain optical coherence tomography (SD-OCT)
images has good concordance, and each imaging modality has its own advantages.
Our algorithm utilizes the projection image to verify and refine the segmentation
results from SD-OCT images, which reduces the influence of RPE estimation error
and improves the robustness of drusen segmentation. Gregori [30] and Iwama [31]
also segmented drusen based on the distance between the abnormal RPE and the
normal RPE floor. Recently, many researchers proposed several other methods [15,
32, 33].
In this section, we tackle the above challenges, and present several novel automated drusen segmentation method in SD-OCT images. Our proposed methods
include: (a) Automated Drusen Segmentation and Quantification in SD-OCT Images
[34], (b) An improved OCT-Derived Fundus Projection Image for Drusen Visualization [35]. In addition, we also provide a means of generating a high-quality projection
image based on RPE layer estimation.
11.2.1 Automated Drusen Segmentation and Quantification
in SD-OCT Images
11.2.1.1 Overview of Proposed Method
We briefly introduce the algorithm, for full detail see [34]. A flowchart of our algorithm is shown in Fig. 11.2, which comprises the following operations on the input
SD-OCT image:
1. Image denoising: A modified bilateral filtering algorithm is used to reduce noise
in order to facilitate the subsequent estimation of the RNFL and RPE retinal
layer.
2. RNFL complex removal: The RNFL complex, defined as the region between
the inner limiting membrane and the outer plexiform layer (indicated by the