Decorrelation Signal of Diabetic Hyperreflective Foci on Optical Coherence Tomography Angiography

Abstract

Diabetic hyperreflective foci in the outer retinal layers are a clinically relevant finding on optical coherence tomography (OCT) images, although their characteristics remain to be elucidated. Here we investigated the decorrelation signal around hyperreflective foci on OCT angiography (OCTA) images in diabetic retinopathy (DR). We retrospectively reviewed sufficient quality OCTA images from 102 eyes of 66 patients that were obtained using split-spectrum amplitude-decorrelation angiography algorithm. Most confluent hyperreflective foci were randomly deposited or appeared in a radiating array on the en-face structural OCT images in the inner nuclear layer (INL) or Henle's fiber layer (HFL), respectively. Within the INL, hyperreflective foci were not accompanied by decorrelation signals and attached to capillaries on OCTA images. Decorrelation signals were sometimes delineated in hyperreflective foci in the HFL and other times appeared to be pseudopod-like or wrapping around hyperreflective foci, referred to as reflectance-decorrelated foci. The decorrelation signal intensity of hyperreflective foci in the HFL was associated with logMAR VA (R = 0.553, P < 0.001) and central subfield thickness (R = 0.408, P < 0.001) but not with DR severity. These data suggest that reflectance-decorrelated foci on OCTA images are clinically relevant as well as shed lights on the properties in diabetic hyperreflective foci.

Figure 1

The distribution of hyperreflective foci on en-face OCT images in the INL and HFL in a 42-year-old patient with severe NPDR. (A) Hard exudates on fundus color photography. (B) The late-phase fluorescein angiography image shows hyperfluorescence in the macula. (C,D) The 10-μm-thick segmentation in the INL (C, between green lines) and HFL (D, between red lines) on B-scan images with decorrelation signals along the arrows in panels G and J, respectively. En-face OCTA (E,H) and structural OCT (F,I) and merged images (G,J; grayscale = structural OCT, red = OCTA) in the green rectangle in panel A. (E–G) The 10-µm-thick en-face image in the INL between the green lines in panel C. The structural OCT image delineates dot-like or spot-like hyperreflective foci. (H–J) The 10-µm-thick en-face image in the HFL between the red lines in panel D. The OCT image reveals that most hyperreflective foci are deposited in a radiating fashion and are partly colocalized to the decorrelation signal on the OCTA image. Scale bar = 500 μm.

Figure 2

Hyperreflective foci attached to capillaries in the INL in a 67-year-old patient with moderate NPDR. The 10-μm-thick en-face OCTA (A), structural OCT (B), and merged images (C; grayscale = structural OCT, red = OCTA) within a central 3 × 3 mm square in the INL between the green lines in panel D. (D) The B-scan image with decorrelation signals along the green arrow in panel C. The magnified images of OCTA (E), structural OCT (F), and merged images (G) within the green square in panel C show the attachment of hyperreflective foci to the capillaries in this layer. Scale bar = 500 μm.

Figure 3

Reflectance-decorrelated foci correspond to parts of hyperreflective foci in the HFL in a 75-year-old patient with PDR. (A) En-face OCTA image in the superficial capillary layer. The 10-μm-thick en-face OCTA (B) and structural OCT (C) images. (D,E) The B-scan images without and with decorrelation signals along the green arrow in panel G. (F) Projection artifacts were removed from the OCTA slab in the HFL using the subtraction function in ImageJ. (G) The merged image of structural OCT and OCTA images after removal of projection artifacts. (H–J) The magnified images within the square in panel G. Decorrelation signals appear to be pseudopod-like or are wrapped around dot-like hyperreflective foci (arrows) and are partly colocalized to hyperreflective foci (arrowheads), referred to as reflectance-decorrelated foci. Scale bar = 500 μm.

Figure 4

Projection artifacts in confluent hyperreflective foci within the HFL in a 60-year-old patient with moderate NPDR. The 10-μm-thick en-face OCTA (A) and structural OCT (B) images in the HFL between the red lines in panel D. (C) OCTA image in the superficial layer. (D,E) B-scan images without and with decorrelation signals along the green arrow in panel F. (F) The merged image (grayscale = structural OCT, red = OCTA) within a central 3 × 3 mm in the HFL. (G) The magnified image within the square in panel F. The decorrelation signal appears to be a meshwork-like capillary network. (H) The merged image of OCTA in the superficial layer (green) and HFL (red). (I) The magnified image within the square in panel H shows that the decorrelation signals in hyperreflective foci correspond to projection artifacts. Scale bar = 500 μm.

Figure 5

Quantification of the decorrelation signal intensity of hyperreflective foci in the HFL in a 63-year-old patient with moderate NPDR. (A) En-face OCTA image in the superficial layer. The 10-μm-thick structural OCT (B) and OCTA (C) slabs in the HFL between the red lines in panel D. (D,E) B-scan images without and with decorrelation signals along the green arrow in panel B. (F,G) Binary images of panels A and B, respectively, obtained using the global threshold function in ImageJ. (H) The areas of hyperreflective foci (white) after the exclusion of projection artifacts from the vessels in the superficial layer (blue). (I) The decorrelation signals (red) within highlighted areas in panel H. The signal levels in individual pixels were measured and averaged to calculate the decorrelation signal intensity. Scale bar = 500 μm.

Figure 6

Association between logMAR VA and the decorrelation signal intensity of hyperreflective foci within the HFL. Correlation between logMAR VA and the decorrelation signal intensity of hyperreflective foci within the HFL in all 102 eyes (A), in 42 eyes with center-involved DME (C), and in 60 eyes without DME (D). (B) The association between CSF thickness and the decorrelation signal intensity.

Figure 7

The decorrelation signal intensity of hyperreflective foci within the HFL in individual DR severity grades in all 102 eyes (A), 60 eyes without DME (B), and 42 eyes with DME (C).