Predicting ophthalmoscopic visibility of retinal photocoagulation lesions by high-speed OCT: an animal study in rabbits

Abstract

Purpose: Optical tomography (OCT) was successfully used to investigate tissue changes after photocoagulation. It is able to grade the strength and spatial extension of the lesions with high accuracy. We developed a high-speed OCT (840 nm, 200,000 A-scans/s) and a slitlamp-based OCT (1060 nm, 90,000 A-scans/s), to visualizes effects of the laser irradiation during laser exposure. Based on ex-vivo results changes in tissue scattering and, by use of the Doppler signal, tissue motion caused by thermal expansion and coagulation-induced tissue contraction were measured in rabbits. Aim was to develop algorithms for quantitative detection of visible photocoagulation lesions in-vivo.

Methods: OCT image series (up to 1000 B-Scans/s) were recorded in Chinchilla Grey rabbits during laser exposure. Lesions were applied with different spot size and exposure times. Irradiance was varied from below threshold until strong lesions were visible in the fundus. Using cross-correlation of images, motion due to heart beat was corrected. More than 200 lesions having sufficient image quality with respect to noise and motion artifacts were evaluated. Tissue scattering and axial movement of the neuronal retina was automatically determined from the images sequence and 26 different parameters were calculated which were then correlated with the ophthalmoscopic visibility of the lesions. Sensitivity, specificity and receiver operating characteristics (ROC) were calculated.

Results: During coagulation an increase of scattering was observed near the retinal pigment epithelium. It predicted the ophthalmoscopic visibility of lesions with 80% sensitivity and 78% specificity. Area under ROC curve (AUC) was 0.78, Youden index 0.58. Axial tissue motion predicted lesions only with a Youden index of 0.45; sensitivity and specificity were 63% and 80%, respectively. The AUC was 0.74. Considerably better results were obtained when the local tissue expansion was calculated. This parameter is sensitive to local changes of the tissue morphology and corrects for global tissue motion. AUC reached 0.82 with 71% sensitivity, 91% specificity and a Youden index of 0.62.

Conclusions: In conclusion, high-speed OCT recording during photocoagulation was able to predict the visibility of the lesion with good reliability. It does provide information on tissue effects of photocoagulation and may be used for an online dosimetry.