Cardiac 4D-flow MRI with Hemodynamic Biomechanics L1-510
Unclaimed Principle — open for contribution
This Principle is declared in the catalog but has no reference solver, no pinned dataset, and is not registered on-chain. There is no reward pool. Submitting a cert against this Principle today will record the cert for reproducibility but pay zero PWM.
To claim it as a Bounty #7 contribution: open a PR adding (1) a reference solver, (2) ≥1 dataset pinned to IPFS, (3) updates to the L3 manifest with dataset CIDs. After verifier-agent triple-review, the founders' 3-of-5 multisig signs PWMRegistry.register() and the Principle becomes mineable.
Forward model E
Cardiac 4D-flow MRI with Hemodynamic Biomechanics: joint multi-physics forward couples (i) 4D phase-contrast MR acquisition where the phase difference Delta-phi(r, t) = gamma * TE * venc * v(r, t) accumulates over a flow-encoding gradient pair, encoding all three velocity components (v_x, v_y, v_z) into the MR signal phase across a 3D spatial grid + cardiac cycle phase; (ii) incompressible Navier-Stokes equations governing the velocity-pressure (v, p) coupling: rho * (dv/dt + (v . grad) v) = -grad p + mu * Laplacian(v) + f_body, with incompressibility constraint div(v) = 0; (iii) physiologically-consistent boundary conditions (no-slip at vessel walls, prescribed inflow / outflow waveforms at chamber inlets and outlets, periodic-cardiac-cycle constraint). The forward DAG has 8 primitives with two coupling constraints (n_c = 2): (i) MR-phase-to-velocity linear coupling (analytic, well-conditioned across vessel lumen but compromised at high-velocity-gradient walls); (ii) NS-equation coupling between velocity and pressure (nonlinear convective term + incompressibility, the standard incompressible-NS multi-physics constraint that regularizes noisy MR-derived velocities and recovers pressure that is otherwise unobservable directly). Recovery is posed as the joint inverse problem that recovers (v(r, t), p(r, t)) from measured Delta-phi(r, t) under NS constraint and BC. Difficulty tier delta = 5 with raw condition number kappa ~ 220 (limited by venc-aliasing and SNR per velocity component) and effective kappa_eff ~ 30 after NS-regularized variational reconstruction. Mismatch parameters: venc_aliasing, partial_volume_at_walls, breath_hold_motion, eddy_current_offset, segmentation_error_at_chambers, viscosity_uncertainty. Additive Gaussian thermal noise sets the data-fidelity floor. See forward_model field for the closed-form joint imaging equation.
L-DAG
Well-posedness W
- Existence:
- true
- Uniqueness:
- conditional
- Stability:
- conditional
- κ:
- 220
Existence of joint (v(r, t), p(r, t)) is guaranteed within the declared Omega bounds and chamber geometry. Uniqueness holds under physiologically-consistent boundary conditions (no-slip walls, prescribed inflow/outflow, periodic-cardiac-cycle) and adequate venc/SNR; pressure recovery is conditionally unique up to a constant (Helmholtz-Hodge decomposition) — typical practice fixes pressure at one anatomical reference. Stability is moderately conditioned (kappa_eff ~ 30 after NS-regularized variational reconstruction) — venc_aliasing dominates aliasing-induced velocity errors; partial_volume_at_walls dominates wall-shear-stress bias; eddy_current_offset contributes a systematic background-velocity bias; viscosity_uncertainty contributes a scaling factor for shear-related quantities. Joint Hadamard well-posedness for the coupled MR-PC + NS forward is established by Markl et al. 2012 (foundational 4D-flow review), Stankovic et al. 2014 (clinical 4D-flow), Bertoglio-Caiazzo 2015 (NS-regularized 4D-flow inverse), Pereira et al. 2016 (assimilation), Garcia et al. 2018 (NS-regularized clinical 4D-flow), and Soulat et al. 2020 (4D-flow benchmarking).
Solvability C
- Solver class:
- linear-operator + variational [4D-PC with divergence-free regularization; NS-PINN with Galerkin projection; sequential PCMRI + CFD assimilation] | iterative variational data assimilation [4D-Var, ensemble Kalman filter] | linear-operator + deep neural prior [4DFlowNet, NS-PINN]
- Convergence rate q:
- 2
- Complexity:
- O(H * W * Z * N_cardiac_phases * (NS_iterations)^(2/3)) per outer iteration via SIMPLE / projection / FEM-NS forward + adjoint; learned variants O(H W Z N_cardiac_phases * F_theta_cost) per forward pass