The work, published in Frontiers in Bioengineering and Biotechnology, comes from a team spanning ophthalmology and mechanical engineering. Qi Ren, Zhaozhe Deng, Zhe Chu, Zeyu Pan and Pengxia Wan constructed a deliberately simple geometric model rather than a computationally heavy simulation. Their starting point is an axisymmetric two-sphere construction. Before the air puff strikes, the anterior corneal surface is approximated as a sphere whose radius is the clinically measured corneal curvature radius, tangent to the corneal apex. At the moment of highest concavity, when the cornea has been pushed furthest inward, the central deflected surface is fitted by a second, inverse-curvature sphere whose apex sits posteriorly by the measured deflection amplitude. The overlap region between these two spheres, revolved around the optical axis, defines the deflection volume.

What makes the approach attractive is its closed-form character. The distance between the two sphere centers follows directly from the deflection amplitude, and the overlap volume of two partially intersecting spheres has a known analytic expression. That means DeflVol can be computed from just three inputs that are already available in routine clinical imaging: the anterior corneal curvature radius, the inverse curvature radius reported by the Corvis ST tonometer at highest concavity, and the deflection amplitude itself. No finite-element meshing, no iterative fitting, no proprietary software. The authors even released an R package, CorDeflVol, so that other researchers can reproduce the calculation from standard device outputs.

Of course, a simple model is only useful if it tracks reality closely enough. To test it, the team assembled paired measurements from 1,351 young emmetropic or myopic participants examined at the First Affiliated Hospital of Sun Yat-sen University, using the OCULUS Pentacam AXL for corneal topography and the Corvis ST for dynamic response. Crucially, they validated the model against Corvis ST outputs that were not used as model inputs. The predicted two-dimensional deflection area, obtained by integrating the overlap chord width along the sagittal section, closely tracked the device-measured deflection area, with a coefficient of determination of 0.99 and a regression slope of 0.96. The modeled overlap chord, in turn, closely tracked the device-reported peak distance, with a coefficient of determination of 0.84 and a slope of 0.95.

The agreement was not perfect, and the authors were candid about the residual discrepancy. Predicted deflection area ran about 5.5 percent, on average, below the measured value, corresponding to a mean bias of minus 0.181 square millimeters. Digging into the geometry, they found that the modeled arc length was systematically shorter than the device-reported deflection length, and that this arc-length difference correlated with the area shortfall. The likely explanation is that the real deflected region extends peripherally beyond the neat boundary where the two idealized spheres intersect, while the model truncates the region at that analytic intersection. Differences in the baseline geometry, image sampling, contour processing and whole-eye movement correction may also contribute to the gap.

To quantify how much that two-dimensional shortfall matters in three dimensions, the researchers ran a deterministic area-matching sensitivity analysis. For each eye, they adjusted the deflection amplitude while holding the two curvature radii fixed until the predicted area matched the measured area, then recomputed the volume. The result was a model-implied DeflVol increase of 7.8 percent, plus or minus 2.2 percent, or roughly 0.7 microliters against a mean volume of about 9.5 microliters. An elasticity analysis showed that within the model, a small relative change in predicted area produces about 1.3 times that relative change in volume, so modest area differences propagate to modest, bounded volume differences rather than exploding.

The team also stress-tested the two geometric assumptions that make the whole construction possible: that the corneal surfaces are spherical and rotationally symmetric. Real corneas are aspheric, with a conic constant typically around minus 0.25 to minus 0.30, and many carry regular astigmatism, meaning the anterior surface has distinct flat and steep meridians. When physiologically representative asphericity was introduced into the sag profiles, DeflVol changed by less than 1 percent on average, and even a conservative extreme of minus 0.50 shifted the mean by under 1.2 percent. At the representative value of minus 0.25, the predicted area changed by only about 0.003 square millimeters, less than 5 percent of the observed area bias. Regular anterior toricity, evaluated by averaging the volume over 72 meridional orientations using each eye’s measured flat and steep radii, changed DeflVol by only about 0.017 percent on average.

Across the full cohort, DeflVol came out at 9.48 plus or minus 1.72 microliters, and it correlated strongly with the established Corvis ST metrics it should logically track: a Pearson correlation of 0.94 with deflection amplitude and 0.99 with deflection area, both highly significant. That tight coupling is reassuring for internal consistency, but it also raises the practical question of what a new parameter adds when it moves almost in lockstep with the old ones. The authors position DeflVol as a complement rather than a replacement: a spatially integrated representation of deflection magnitude that may better characterize corneal deformability under mechanical loading, a property relevant to habitual stresses such as eye rubbing or forceful eyelid closure, and potentially to individual susceptibility to biomechanics-related ocular disorders.

The study’s limitations are clearly drawn. DeflVol describes deformation at a single instant, the highest concavity moment, and cannot capture transient instabilities such as wobbling or high-frequency oscillations that high-speed imaging sometimes reveals during corneal buckling. The cohort consisted exclusively of young emmetropic or myopic eyes with strict quality filters on central corneal thickness, curvature, astigmatism and intraocular pressure, so ectatic corneas like those in keratoconus, or corneas altered by refractive surgery, remain untested and may depart from the model’s assumptions in larger or less predictable ways. The authors call for test-retest repeatability studies, evaluation in structurally irregular corneas, and comparison against volumetric estimates from elevation-map integration or finite-element modeling.

Even with those caveats, the paper makes a case that a transparent geometric model can extract a genuinely three-dimensional descriptor from equipment already sitting in clinics worldwide. Where finite-element simulations offer mechanistic depth at the cost of computational intensity and calibration, DeflVol offers a one-line calculation with quantified error bars: a known area bias, a bounded volumetric correction of roughly 0.7 microliters, and demonstrated insensitivity to the asphericity and toricity that real corneas exhibit. For a field that has spent years squeezing biomechanical insight out of one-dimensional amplitudes and two-dimensional areas, having a computationally cheap volumetric companion metric, verified against 1,351 paired clinical measurements, is a small but concrete step toward describing corneal deformation in the full dimensionality it actually occupies.

Subject of Research: A model-derived volumetric descriptor of corneal deflection under Corvis ST air-puff loading

Article Title: DeflVol: a model-derived volumetric descriptor of corneal deflection at highest concavity under Corvis ST air-puff loading

Article References: Ren, Q., Deng, Z., Chu, Z., Pan, Z., & Wan, P. (2026). DeflVol: a model-derived volumetric descriptor of corneal deflection at highest concavity under Corvis ST air-puff loading. Frontiers in Bioengineering and Biotechnology, 14, Article 1933733. https://doi.org/10.3389/fbioe.2026.1933733

Image Credits: AI Generated

DOI: 10.3389/fbioe.2026.1933733

Keywords: corneal biomechanics, DeflVol, Corvis ST, air-puff tonometry, Scheimpflug imaging, dynamic corneal response, deflection volume, highest concavity, two-sphere model, Pentacam AXL, keratoconus, ophthalmic diagnostics