Changes in entropy on polarized-sensitive optical coherence tomography images after therapeutic subthreshold micropulse laser for diabetic macular edema: A pilot study

Abstract

Purpose: To investigate the dynamics of the healing process after therapeutic subthreshold micropulse laser (SMPL) for diabetic macular edema (DME) using polarization-sensitive optical coherence tomography (PS-OCT).

Methods: Patients with treatment-native or previously-treated DME were prospectively imaged using PS-OCT at baseline, 1, 2, 3, and 6 months. The following outcomes were evaluated: changes in the entropy value per unit area (pixel2) in the retinal pigment epithelium (RPE) on the B-scan image; changes in the entropy value in each stratified layer (retina, RPE, choroid) based on the ETDRS grid circle overlaid with en face entropy mapping, not only the whole ETDRS grid area but also a sector irradiated by the SMPL; and the relationship between edema reduction and entropy changes.

Results: A total of 11 eyes of 11 consecutive DME patients were enrolled. No visible signs of SMPL treatment were detected on PS-OCT images. The entropy value per unit area (pixel2) in the RPE tended to decrease at 3 and 6 months from baseline (35.8 ± 17.0 vs 26.1 ± 9.8, P = 0.14; vs 28.2 ± 18.3, P = 0.14). Based on the en face entropy mapping, the overall entropy value did not change in each layer in the whole ETDRS grid; however, decrease of entropy in the RPE was observed at 2, 3, and 6 months post-treatment within the SMPL-irradiated sectors (P < 0.01, each). There was a positive correlation between the change rate of retinal thickness and that of entropy in the RPE within the SMPL-irradiated sector at 6 months (r2 = 0.19, P = 0.039).

Conclusion: Entropy measured using PS-OCT may be a new parameter that facilitates objective monitoring of SMPL-induced functional changes in the RPE that could not previously be assessed directly. This may contribute to a more promising therapeutic evaluation of DME.

Clinical trial: This clinical study was registered in UMIN-CTR (ID: UMIN000042420).

Fig 1. CONSORT 2010 flow diagram.

(DME patients). This study included 11 eyes of 11 subjects.

Fig 2. Representative SD-OCT (A), ETDRS grid (B), and PS-OCT images (C-F).

(A) En face macular map with the ETDRS grid obtained using SD-OCT. (B) The ETDRS grid was used to demarcate nine zones delimited by solid circles with diameters 1 mm, 3 mm, and 6 mm meter centered on the fovea. (C) Intensity B-scan image [color scale: = 45 to 80 dB]. (D) Entropy B-scan image [color scale: entropy of 0 to 1]. Pixels below the detection threshold of the OCT intensity are displayed in black. (E) Overlapped with signalized entropy B-scan image (set as > 0.1) on intensity OCT image. (F) En face entropy map of each segmented layer (512 × 512 pixels, each) and the numeral tables, which present entropy values at each pixel location, are shown. Abbreviations in (B) are shown in the following order; TI, temporal inner; II, inferior inner; NI, nasal inner; SI, superior inner; TO, temporal outer; NO, nasal outer; SO, superior outer.

Fig 3. Assessment of entropy signals per unit area of RPE (pixel²).

The linear scan image obtained by scanning through the fovea within a 1500 μm radius centered on the fovea was selected (delimited with two vertical markers traced at established distances; upper) and its imaging was binarized using Image J software (middle). Unit area of RPE was defined as 3 pixels above/below the line from automatic RPE segmentation (yellow lines), and the percentage of RPE entropy signals (red color) was calculated (bottom).

Fig 4. Evaluation of the number of entropy signals along with HRF.

Intensity images on PS-OCT B-scan (upper) and overlapped image with signalized entropy (set as > 0.1) on its intensity OCT image (lower) of DME patient are shown. Each image obtained by scanning through the fovea within a 1500μm radius centered on the fovea was selected (delimited with two vertical markers traced at established distances, left side of each). The numbers of HRF and dots of entropy signals in the retina were evaluated. The pictures on the right side of each picture are a partial enlarged image and yellow and green arrows indicate HRS and entropy signals, respectively. HRF, hyperreflective foci; DME, diabetic macular edema.

Fig 5. Comparison of the mean entropy value of each en face segmentation layers according to ETDRS grid between healthy subjects (n = 11) and patients with DME (n = 11).

Mean entropy value in the choroid layer within the ETDRS circle with the diameter of 1 mm was a significantly lower in patients with DME than in healthy subjects (0.16 ± 0.064 vs 0.23 ± 0.037, P = 0.010), whereas in other circle ranges and strata, no difference was observed between patients with DME and healthy subjects. The ETDRS grid was used to demarcate nine zones delimited by solid circles with diameters of 1 mm, 3 mm, and 6 mm centered on the fovea; radial lines were projected onto the fundus to divide the OCT map into nine sectors. The results are the means ± standard deviation.

Fig 6. Changes in the number of HRF and entropy signals after SMPL as determined using PS-OCT B-scans (n = 11).

After treatment with SMPL, there was a significant increase the number of HRF at 3-month timepoint relative to that at baseline (P = 0.019), whereas the number of entropy signals, HRF, and entropy signals overlaid with HRF did not change significantly during the follow-up. SMPL, subthreshold micropulse laser; HRF, hyperreflective foci. The results are the means ± standard deviation.

Fig 7. Changes in mean entropy values on SMPL-irradiated sectors according to ETDRS grid in each en face layer (n = 11).

The mean entropy of the RPE layer tended to be lower at the 1-month timepoint (P = 0.24) and was significantly lower at the 2-month, 3-month, and 6-month timepoint after SMPL than at baseline value (P < 0.01, each). The results are the means ± standard deviation.

Fig 8. Relationship between the rate of change of retinal thickness and that of entropy value in SMPL-irradiated sectors.

There was a positive correlation between the change rate of retinal thickness and that of entropy in the RPE at 6 months after SMPL (n = 9; r² = 0.19, P = 0.039). On the contrary, there was a negative correlation between the change the rate of retinal thickness and that of entropy in the choroid at 3 months after SMPL (n = 11; r² = 0.28, P = 0.002).