Abstract
Purpose :
Ultrafast ophthalmic adaptive optics (AO) outperforms conventional AO in correcting ocular aberrations and improves AO clinical utility [1]. However, the higher loop rate necessitates shorter wavefront sensor exposure durations that are comparable with deformable mirror (DM) response time. Thus, DM actuation may corrupt the sensor measurement and degrade the AO performance. Here, we study this effect and develop a way to minimize it.
Methods :
Our AO system [1] consists of (1) a Shack-Hartmann wavefront sensor (SHWS) with 300 lenslets that sample a 6.7 mm eye pupil and a rolling-shutter camera (0.1 ms exposure time) that captures the SHWS images, and (2) a DM (ALPAO high-speed DM97-15) with a 1 ms rise time followed by ringing as specified by the manufacturer. To study the DM actuation effect, we controlled the delay between sending control commands to the DM and start of the SHWS exposure. We compared the power rejection curves (for pink noise applied to the DM) and step responses to fixed aberrations for exposure delays of 0, 1 and 6 ms. 0-ms delay exposes the SHWS to almost the entire DM response and 6-ms delay avoids it entirely. The rows of lenslets affected by DM actuation were determined by comparing the lenslet spot displacements, from which we determined the optimal delay.
Results :
Overshoot of the step response to a large aberration (2.7 µm RMS wavefront error) occurred for the 0-ms exposure delay, while no overshoot occurred for the 1- and 6-ms delays. DM actuation affected the top two rows of SHWS lenslets (out of 20) for the 0-ms exposure delay and no rows for the 1-ms delay. Because the rolling shutter exposes rows from top to bottom in 3 ms, 0.3 ms was therefore determined as the optimal exposure delay. Reducing exposure delay from 1 ms to 0.3 ms, AO loop rate increased 17% to 233 Hz; closed-loop bandwidth increased 14% to 37.2 Hz; convergence time to diffraction limit reduced 14% to 4.3 ms; and power rejection curve peak magnitude remained the same. For smaller aberrations (0.32 µm RMS wavefront error, comparable to ocular aberrations after spectacle correction), no overshoot was observed in the step response for 0-ms delay, and closed-loop bandwidth increased 21% to 39.3 Hz compared with 1-ms delay.
Conclusions :
DM actuation reduces AO performance when aberrations are large, but this decrease can be greatly reduced with the right exposure delay.
[1] Liu et al., IOVS, 62(8):16 (2021).
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.