Purpose
Functionality of a filtering bleb is often affected by post-operative tissue alteration including scarring and fibrosis. In order to assess the integrity of blebs, birefringence imaging is performed using polarization sensitive optical coherence tomography (PS-OCT). A diagnostic algorithm is developed to classify the blebs using birefringence property.
Methods
A custom built Jones matrix (JM)-OCT with 1.3-µm probe wavelength specifically designed for anterior eye imaging was used. 12 × 12-mm JM-OCT volumes of filtering blebs from 12 subjects and 3 × 3-mm volumes of anterior sclera from 10 control subjects were used. The volumes were processed with a Bayesian numerical estimation algorithm. It provides fully quantitative birefringence tomographies. The birefringence tomographies of the control eyes were used to determine the best birefringence threshold which separates the normal sclera and conjunctiva. The blebs were classified based on IOP and medication; functional bleb if IOP < 18 without medication or nonfunctional if IOP ≥ 18 mmHg or with medication. A parameter called fibrosis score (FS) was computed for each bleb. FS is the occupancy of the pixels with higher birefringence in conjunctiva than the formerly described threshold. It relies on the fact that collagen in the fibrosis has high birefringence. An average FS was calculated from the 10 frames for each case. The clinical diagnostic capability of FS was evaluated for bleb functionality using area under receiver operating characteristics curve (AROC).
Results
The subjects include 6 subjects of primary open angle glaucoma, 4 pseudo-exfoliative glaucoma, 1 rubeotic glaucoma, and 1 primary angle closure glaucoma. The mean age of patients was 73 ± 9 years. The mean time period between trabeculectomy and PS-OCT imaging was 3.4 ± 4.5 years. Figure 1 shows examples of functional bleb [(a) and (b)] with low FS (2.3%) and nonfunctioning bleb [(c) and (d)] with high FS (27.1%). (a) and (c) show birefringence maps and (b) and (d) highlight highly birefringent pixels. Mean FS of the two groups were 9.2% (functional) and 19.5% (nonfunctional) (non-significant, p = 0.07). High AROC of 0.78 is obtained for the classification of bleb functionality. At the best performance, the specificity was 100% and the sensitivity was 67%.
Conclusions
A new diagnosis method based on birefringence successfully assessed the functionality of blebs.