Adaptive optics optical coherence tomography (AO-OCT) has transformed retinal imaging, enabling in vivo visualization of cellular structures such as photoreceptors, capillaries, and ganglion cells. However, one persistent limitation has constrained its broader clinical utility: a shallow depth of focus. High numerical aperture systems deliver exquisite lateral resolution but restrict the focal plane to just tens of microns, meaning multiple acquisitions at different depths are typically required to capture the full retinal thickness.
Now, a new Biocybernetics and Biomedical Engineering study proposes an elegant solution – combining hardware AO-OCT with computational aberration correction (CAC) to digitally extend depth of focus from a single acquisition.
From hardware complexity to computational efficiency
Conventionally, clinicians and researchers rely on sequential refocusing – acquiring multiple volumes at different retinal layers and stitching them together. But focusing on one layer (e.g., ganglion cell layer) leads to significant blurring in others, such as the photoreceptor layer. This approach increases acquisition time and susceptibility to motion artifacts.
The proposed method instead applies CAC as a post-processing step. By modelling and correcting depth-dependent defocus computationally, the system restores sharpness across all retinal layers without additional hardware adjustments. The workflow uses Fourier-domain processing and Zernike-based phase correction to iteratively optimize image sharpness.
Importantly, the study demonstrates that correcting defocus alone is sufficient. Higher-order aberration correction provided minimal additional benefit – reducing residual phase error by just a little while dramatically increasing computational burden.
Clinical-quality imaging from a single focus
Using this hybrid approach, the authors achieved cellular-level resolution across the full retinal thickness from a single focus setting – optimally placed at the inner plexiform layer (IPL). This choice balances signal-to-noise ratio (SNR) across layers, which is critical for successful computational correction. The study found that CAC performs reliably when layer SNR exceeds ~25 dB, reinforcing the importance of adequate signal acquisition. If SNR was below this level, the authors observed, the algorithm would fail to produce a meaningful phase correction.
The results are compelling. In healthy eyes, CAC enabled clear visualization of the photoreceptor mosaic even at 1.5° from the fovea – an area where cones are densely packed and difficult to resolve. The computationally refocused images closely matched “ground truth” images acquired with hardware focus placed directly on the photoreceptor layer.
Beyond photoreceptors, the technique enabled simultaneous visualization of multiple retinal layers – including nerve fiber layer, ganglion cells, plexiform layers, and outer retina – in a single volumetric dataset.
Applications in disease imaging
The translational potential is highlighted in a 53-year-old patient with multiple sclerosis and prior optic neuritis. CAC-enhanced imaging revealed ganglion cell loss alongside improved delineation of cone photoreceptors, with a reported 278% increase in cone contrast compared to uncorrected images. Such multi-layer insights could prove valuable in diseases where pathology spans both the inner and outer retina.
Implications for clinical practice
By reducing the need for multiple focus acquisitions, this approach could shorten imaging sessions by up to threefold while simplifying workflows. For clinicians, this translates into improved patient comfort, reduced motion artifacts, and more practical integration of AO-OCT into routine care.
The study authors note that CAC enhances image sharpness rather than signal itself – meaning good acquisition quality of course remains essential. Computational demands also remain a consideration, though defocus-only correction can be performed in seconds and could be accelerated further with GPU implementation.
Looking ahead
As AO-OCT continues to evolve, hybrid hardware–software solutions like CAC may be key to bridging the gap between research-grade imaging and clinical adoption. By enabling full-thickness, cellular-resolution imaging in a single acquisition, this approach opens new possibilities for diagnosing and monitoring retinal disease – and for understanding the complex interplay between retinal layers in health and pathology.
