Dosimetry of selective retina therapy using optical coherence tomography: first results from the CENTAURUS study

Purpose : Assessment of the SPECTRALIS CENTAURUS device for real-time feedback-controlled dosimetry (RFD) of selective retina therapy (SRT). Evaluation of optical coherence tomography (OCT) data for future real-time dosimetry applications.
Methods : A test pattern (single pulse) for damage threshold determination as well as a specified laser treatment pattern for RFD simulation (ramped up pulse sequence, maximal 30 pulses, 100 Hz) were applied in 3 patients with diabetic retinopathy using the SPECTRALIS CENTAURUS device (HuCE-optoLab, BFH-TI, CH). This device consists of a modified imaging platform (SPECTRALIS HRA+OCT, Heidelberg Engineering, DE) extended with a prototype laser (Meridian Medical; wavelength: 532 nm; peak power: 30 W) for SRT. Laser pulses of 8 µs duration with a maximum pulse energy of 170 µJ were used. OCT M-scan data recorded during treatment was analyzed post-treatment to investigate OCT-RFD. The patterns were evaluated binary for retinal pigment epithelium (RPE) damage based on fluorescence angiography (FA), overtreatment based on color fundus photography (CFP), as well as OCT M-scans fringe washouts (visually / dedicated algorithm). Probit analysis was used to calculate the ED-values (median-effective dose), the therapeutic window (TW = ED50-CFP / ED50-FA), and the safety range (SR = ED16-CFP / ED84-FA) for SRT.
Results : Probit analysis for the test pattern shows an RPE damage threshold of 57 µJ (ED50-FA) and a threshold towards overtreatment of 95 µJ (ED50-CFP). Furthermore, a TW of 1.7 and a SR of 1.2 for SRT was found. For all lesions, OCT M-scan feedback (detectable fringe washouts) is guaranteed with 50% probability from 47 µJ on (visual evaluation), respectively 53 µJ (algorithm). At the lower SR limit (ED84-FA = 75 µJ), visually evaluated dosimetry could provide feedback with 98% and the algorithm with 95% probability. At the upper SR limit (ED16-CFP = 92 µJ), both methods could provide feedback with 99% probability. A statistical evaluation of the RFD simulation pattern shows that over all a sensitivity of 83% and a specificity of 93% was achieved for the dosimetry algorithm.
Conclusions : OCT-RFD of SRT using the SPECTRALIS CENTAURUS device is feasible in patients with diabetic retinopathy, indicating that controlled retinal treatment is possible. Applying the algorithm in real-time would have reduced the applied energy dose for certain lesions by 30%.