Zhou et al. examined the estimation of corneal elastic properties through the utilization of shear wave optical coherence elastography (OCE), more specifically via the departure from conventional time-of-flight (TOF) processing and towards a DenseNet-based deep learning pipeline. In order to do so, Zhou et al. first addressed the low repeatability, precision, and reproducibility of conventional TOF processing. Then, Zhou et al. were able to design and train a two-dimension plus time (2D + t) convolutional neural network (CNN) called DenseNet. This concentration prediction network was then trained alongside agar phantoms, which have highly reproducible elastic properties. When applied to both phantom and porcine cornea data, Zhou et al. were able to showcase a nonlinear relationship between agar concentration and Young’s modulus (YM), a finding that corroborated prior research despite yielding higher absolute values given the differences in strain. Specifically, the porcine samples demonstrated smaller standard deviations compared to the TOF processing, essentially indicating greater repeatability of the deep-learning approach. For example, the CNN was able to predict agar concentration while maintaining a mean absolute error (MAE) of 0.028 ± 0.036 (training) and 0.036 ± 0.024 (testing), with each data piece representing percentage-point units. In these same porcine samples, the elasticity maps had also highlighted a higher stiffness located at the corneal vertex as opposed to the periphery, hence further reflecting known collagen architecture. However, Zhou et al. also provides a holistic review of the deep-learning approach and notes how the reliable relativity is also balanced with indirect predictions of absolute YM thus requiring much more positional data in order to mitigate this challenge.