Assessment of area and structural irregularity of retinal layers in diabetic retinopathy using machine learning and image processing techniques

Abstract

Diabetes retinopathy prevention necessitates early detection, monitoring, and treatment. Non-invasive optical coherence tomography (OCT) shows structural changes in the retinal layer. OCT image evaluation necessitates retinal layer segmentation. The ability of our automated retinal layer segmentation to distinguish between normal, non-proliferative (NPDR), and proliferative diabetic retinopathy (PDR) was investigated in this study using quantifiable biomarkers such as retina layer smoothness index (SI) and area (S) in horizontal and vertical OCT images for each zone (fovea, superior, inferior, nasal, and temporal). This research includes 84 eyes from 57 individuals. The study shows a significant difference in the Area (S) of inner nuclear layer (INL) and outer nuclear layer (ONL) in the horizontal foveal zone across the three groups (p < 0.001). In the horizontal scan, there is a significant difference in the smoothness index (SI) of the inner plexiform layer (IPL) and the upper border of the outer plexiform layer (OPL) among three groups (p < 0.05). There is also a significant difference in the area (S) of the OPL in the foveal zone among the three groups (p = 0.003). The area (S) of the INL in the foveal region of horizontal slabs performed best for distinguishing diabetic patients (NPDR and PDR) from normal individuals, with an accuracy of 87.6%. The smoothness index (SI) of IPL in the nasal zone of horizontal foveal slabs was the most accurate at 97.2% in distinguishing PDR from NPDR. The smoothness index of the top border of the OPL in the nasal zone of horizontal slabs was 84.1% accurate in distinguishing NPDR from PDR. Smoothness index of IPL in the temporal zone of horizontal slabs was 89.8% accurate in identifying NPDR from PDR patients. In conclusion, optical coherence tomography can assess the smoothness index and irregularity of the inner and outer plexiform layers, particularly in the nasal and temporal regions of horizontal foveal slabs, to distinguish non-proliferative from proliferative diabetic retinopathy. The evolution of diabetic retinopathy throughout severity levels and its effects on retinal layer irregularity need more study.

Figure 1

The provided illustration depicts the cross-sectional sections of the fovea and their respective subdivisions. The blue arrow indicates a vertical slab, while the yellow arrow indicates a horizontal slab. Both the horizontal and vertical slabs are subdivided into three zones. The horizontal cuts are labeled as N for nasal, F for fovea, and T for temporal. Similarly, the vertical cuts are labeled as S for superior, F for fovea, and I for inferior.

Figure 2

(A) illustrates the zones and areas that have been assessed in the present investigation. The white dashed arrows depict areas (S) of the inner nuclear layer (INL) and outer nuclear layer (ONL). The inner plexiform layer (IPL) is depicted as a linear representation by the green arrow. The Blue band represents the outer plexiform layer (OPL). The upper and lower boundaries of the outer plexiform layer (OPL) have been assessed individually, and the measurement of the distance between these boundaries has been designated as the OPL area. Moreover, the orange line represents the ellipsoid zone (EZ). (B) In this section, we have presented the calculation of the smoothness index (SI). The line length, shown by the yellow dashed line, represents the Euclidean distance between the starting and ending points of the IPL. On the other hand, the curve length, depicted by the green curve, represents the actual length of the IPL.

Figure 3

The receiver operating characteristic curves (ROC curves). (A) To differentiate between diabetic patients (NPDR and PDR) and normal patients, the study found that the area (S) of the inner nuclear layer (INL) in the foveal zone of horizontal slabs exhibited the most effective performance. This measure achieved an accuracy rate of 87.6% (confidence interval 0.798–0.954). (B) The accuracy of discriminating between individuals with normal health conditions and those diagnosed with Diabetes, based on the area of the outer nuclear layer (ONL) in the foveal zone of horizontal slabs, was found to be 77.3% (confidence interval 0.675–0.871).

Figure 4

The receiver operating characteristic curves (ROC curves). (A) In the context of discriminating between patients with PDR and NPDR, the smoothness index (SI) of IPL in the nasal zone of horizontal foveal slabs achieved the highest level of performance, with an accuracy of 97.2% (confidence interval 0.934–1.00). (B) Demonstrates that the accuracy of distinguishing between patients with NPDR and patients with PDR based on the smoothness index of the upper border of the OPL in the nasal zone of horizontal slabs was 84.1% (CI 0.716–0.967). (C) Demonstrates that the smoothness index of IPL in the temporal zone of horizontal slabs distinguished between patients with NPDR and those with PDR with an accuracy of 89.8% (CI 0.805–0.992).

Figure 5

Illustrates the segmentation of IPL (shown by a green line), OPL (represented by a blue region), and ELM (shown by an orange line) in three groups of patients (normal, NPDR, and PDR) in both horizontal (A) and vertical (B) OCT slabs.