Fully Automated Robust System to Detect Retinal Edema, Central Serous Chorioretinopathy, and Age Related Macular Degeneration from Optical Coherence Tomography Images

Abstract

Maculopathy is the excessive damage to macula that leads to blindness. It mostly occurs due to retinal edema (RE), central serous chorioretinopathy (CSCR), or age related macular degeneration (ARMD). Optical coherence tomography (OCT) imaging is the latest eye testing technique that can detect these syndromes in early stages. Many researchers have used OCT images to detect retinal abnormalities. However, to the best of our knowledge, no research that presents a fully automated system to detect all of these macular syndromes is reported. This paper presents the world's first ever decision support system to automatically detect RE, CSCR, and ARMD retinal pathologies and healthy retina from OCT images. The automated disease diagnosis in our proposed system is based on multilayered support vector machines (SVM) classifier trained on 40 labeled OCT scans (10 healthy, 10 RE, 10 CSCR, and 10 ARMD). After training, SVM forms an accurate decision about the type of retinal pathology using 9 extracted features. We have tested our proposed system on 2819 OCT scans (1437 healthy, 640 RE, and 742 CSCR) of 502 patients from two different datasets and our proposed system correctly diagnosed 2817/2819 subjects with the accuracy, sensitivity, and specificity ratings of 99.92%, 100%, and 99.86%, respectively.

Figure 1

Images of (a) healthy eye vision, (b) RE affected vision, (c) CSCR affected vision, and (d) ARMD affected vision.

Figure 2

Clinical macular analysis using OCT images: (a) healthy OCT scan, (b) CSCR affected OCT scan, (c) OCT scan with RE symptoms, and (d) OCT scan with ARMD symptoms (drusen).

Figure 3

Detailed step-by-step flow diagram of the proposed system.

Figure 4

Second-order structure tensor grid: (a) tensor computed through the dot product of horizontal gradient, (b) tensor computed through the dot product of horizontal and vertical gradients, (c) tensor computed through the dot product of vertical and horizontal gradients, and (d) tensor computed through the dot product of vertical gradient.

Figure 5

Segmented retinal and choroidal layers: (a) highly coherent 2D structure tensor I_C(s, t), (b) binary map I_B′(s, t) of highly coherent tensor, (c) canny edge detection of retinal and choroid layer, and (d) segmented retinal layers.

Figure 6

Segmented retinal layers in one B-scan: ILM, retinal nerve fiber layer (RNFL), inner plexiform layer (IPL), inner nuclear layer (INL), outer nuclear layer (ONL), outer plexiform layer (OPL), outer segment (OS), inner segment (IS), RPE, Bruch's membrane (BM), and choroid.

Figure 7

OCT scans structure. A-scan after the 10th interval is highlighted by white color. ILM is shown in red color, RPE is shown in yellow color, and choroid is highlighted in green color. A-scans were fused together to form a single B-scan.

Figure 8

(a) Segmented ILM, choroid, and intraretinal cyst pathology mapped onto the candidate scan; (b) retinal thickness profile.

Figure 9

Cyst fluid detection: (a) I_B′(s, t); (b) R_M(s, t) generated through ILM and choroid; (c) Cyst(s, t); (d) extracted cyst fluid mapped onto the candidate scan. It is shown in yellow color.

Figure 10

Drusen detection through atrophic analysis of RPE. Segmented ILM is shown in red color. RPE is highlighted in yellow color. Choroid is shown in green color.

Figure 11

Training phase of SVM.

Figure 12

Flowchart of classification algorithm.

Figure 13

Unlabeled AFIO dataset: (a) RE classified scans; (b) CSCR classified scans; (c) healthy classified scan.

Figure 14

Unlabeled Duke dataset: (a) healthy classified scans; (b) ARMD classified scans.

Figure 15

Early self-diagnosis of macular syndromes: (a) early symptoms of RPE atrophy due to the presence of drusen; (b) early formation of cyst fluid within macular pathology. These scans are correctly identified by our proposed system.