This is a 2D field companion to the 3D vessel model: instead of an explicit capillary tube mesh, the retina is discretized into a grid of capillary tissue units. Each unit carries three pericyte "slots" that die off under the same first-order survival law driven by hyperglycemic exposure:
S(t) = exp( -k · t ), k = k0 · max(0, HbA1c − 5) / 5
S(t) = fraction of pericytes still alive at time t (years)
k0 = baseline loss-rate constant
Where local coverage falls below a structural-support threshold, a unit converts to a microaneurysm or an occluded (non-perfused) cell exactly as in the 3D model. But two mechanisms here are genuinely 2D-native and not present in the vessel-graph version: VEGF is solved as a real reaction-diffusion field over the grid,
∂V/∂t = D·∇²V + γ·(V_target − V), V_target = 3.6 in occluded cells, 1.0 elsewhere
so ischemic hotspots visibly bleed VEGF into neighbouring tissue rather than a single lumped scalar. Blood flow is a Darcy pressure solve, ∇·(k(x)∇P)=0 with high-pressure/low-pressure boundaries and near-zero permeability k(x) inside occluded cells, relaxed by Jacobi iteration each update — tracer particles are then advected along −k∇P, so flow visibly reroutes around and stalls inside blocked tissue instead of just halting on a fixed edge.
- Duration / HbA1c — same exposure and glycemic-control drivers as the 3D model; pericyte death is irreversible once a slot's random threshold is crossed.
- Run time — animates duration forward continuously.
- Blood flow — toggles the pressure-driven tracer field.
- VEGF field — overlays the diffusing VEGF concentration as a green heat map; its spatial mean matches the same 1 + 2.6·(ischemic fraction) target used in the 3D scalar model.