Abstract
Purpose :
To evaluate and select a suitable photodetector for collecting near infrared autofluorescence (IRAF) and indocyanine green (ICG) fluorescent light in a custom-built adaptive optics scanning light ophthalmoscope and then characterize the performance of the system.
Methods :
One photomultiplier tube (PMT, Hamamatsu H7422-50) and one avalanche photodiode (APD, Excelitas C30659-900-R8AH with a custom amplifier circuit provided by Boston Micromachines) were inserted into the same fluorescence detection channel of the AOSLO and used to detect the weaker IRAF and stronger ICG signals, which share similar emission bands (805-840 nm) but with 2-4x difference in signal strength. Readout electronics were identical except for the amplifiers. Images from an ICG model eye consisting of ICG-stained paper and from a human subject were used to select the better-performing detector, which was further tested on human subjects to evaluate the impact of detector gain and confocal pinhole (PH) size. Signal-to-noise ratio (SNR) was used to evaluate detector performance on the ICG model eye. Images from subjects were normalized to a simultaneously acquired nonfluorescent channel.
Results :
Although the APD had higher quantum efficiency (QE) with the capability to detect higher near-infrared wavelengths, images of the ICG model eye acquired by the APD were grainier and noisier, with SNR 8x and 12x lower than PMT images for 2.5 and 7.5 Airy Disk Diameter (ADD) PHs, respectively. This was consistent with images from a human eye, where retinal pigment epithelial cells could be seen in IRAF and ICG images using the PMT but only faintly (for ICG) or not at all (for IRAF) using the APD. Thus, the PMT was selected. The standard deviation of IRAF and ICG PMT images increased monotonically with gain (favorable for distributing signal over a broader range of usable bits). However, near the upper gain control voltage limit, the images become very noisy. When increasing the detector PH from 1 to 7.5ADD, the throughput increased by about 10X compared to a theoretical 1.25X for a diffraction-limited system, likely due to the presence of residual aberrations.
Conclusions :
For near-infrared fluorescence imaging in the eye, superior performance was achieved using a PMT when compared to an APD. Improved images can be captured by increasing the internal gain up to (but not beyond) the upper limit and by increasing the size of the PH.
This abstract was presented at the 2019 ARVO Imaging in the Eye Conference, held in Vancouver, Canada, April 26-27, 2019.