Abstract
Purpose: :
Recent volumetric imaging of the in vivo human retina with spectral domain optical coherence tomography (SD–OCT) has yielded unprecedented detail of intra–retinal structures owing to SD–OCT’s high axial resolution and sensitivity to weak reflections. Extraction of microscopic features at the cellular level, however, has largely failed owing to the poor lateral resolution of the eye and involuntary eye motion that corrupts the 3D image. To address these shortcomings, we have developed a SD–OCT instrument that incorporates adaptive optics for correcting the aberrations of the eye (which increases lateral resolution) and high speed imaging for reduced motion artifacts.
Methods: :
A fiber–based, high–resolution AO SD–OCT retina camera was developed for high–speed imaging of small volumes of retina in vivo. The OCT subsystem included a superluminescent diode and a high speed 512 pixel linescan detector. The AO sub–system included a 36–actuator AOptix bimorph mirror (placed between the scanners and eye) and a Shack–Hartmann wavefront sensor that operated at closed loop up to 20 Hz and corrected the ocular aberration across a 6.6 mm pupil. A bite bar and forehead rest stabilized the subject’s head. Individual A–scans were acquired at 73,000 per second and combined to form volume images. Speckle noise was reduced by registering and averaging consecutive 3D images.
Results: :
Transverse and axial resolution of the AO SD–OCT instrument were measured at ∼3 and ∼5 micron, respectively. Sensitivity was 85.3 dB for a 8.5 micro–sec exposure time. The centroid RMS error of the AO system was reduced by a factor of 3. Small volumetric images of the living retina were successfully acquired within 10 msec, a time duration short enough to preclude eye motion at the cellular level. All major layers of retina were clearly observed in the 3D images. AO correction of the most significant ocular aberrations produced a 7.5 dB increase in signal–to–noise for the reflection from the inner/outer segment junction when placed within the instrument’s depth of focus. Early results suggest that speckle contrast was reduced after the averaging of multiple 3D images.
Conclusions: :
A fast acquisition AO SD–OCT retina camera was developed and successfully acquired volumetric images of small patches of retina in vivo. Image quality was greatly enhanced by eliminating motion artifacts and reducing ocular aberration blur.
Keywords: retina • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • image processing