Chest CT Pneumonia/COVID Severity Classification (PWDR) L1-514

Medical ImagingLung tissue density reconstruction with pneumonia/COVID severity-grade categorical readoutδ=3 · standardL_DAG = 5.7📋 Stub — not mineable

📋

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.

Read the contribution guide →Open a claim issue →

Forward model E

S t a g e 1 (L 1 - 029) : y_{p} r o j ec t i o n (t h e t a, s) \to m u_{a} tt e n u a t i o n (r) . S t a g e 2 (T S S / C O - R A D S) : GG O_{f} r a c t i o n_{l} o b e = v o l u m e (m u in GG O_{H} U_{r} an g e) / v o l u m e_{l} o b e; T S S_{l} o b e = s t e p (GG O_{f} r a c t i o n_{l} o b e a tt a u_{t} h r es h o l d s); T S S_{t} o t a l = l \sum o b e T S S_{l} o b e; C O - R A D S = m or p h o l o g i c a l_{c} l a ss i f i er (GG O_{d} i s t r ib u t i o n, co n so l i d a t i o n_{d} i s t r ib u t i o n, cr a z y_{p} a v in g_{p} r ese n ce, p er i p h er a l_{v} s_{c} e n t r a l_{r} a t i o) .

Chest CT Severity Classification (PWDR): wraps L1-029 CT analytical core with the international Pan/Francone Total Severity Score and Dutch CO-RADS clinical-grading rules. Stage 1 (analytical, from L1-029): reconstruct lung tissue density mu_attenuation(r) [HU] from CT projections; segment lobes and identify abnormal regions per Hounsfield-density thresholds (-700 to -400 HU = ground-glass opacity GGO; -100 to 50 HU = consolidation; -700 to -800 + interlobular septal thickening = crazy-paving). Stage 2 (deterministic threshold): aggregate per-lobe involvement fractions to per-lobe TSS sub-scores; sum across 5 lobes for total TSS in [0, 25]; apply morphological-pattern rules to yield CO-RADS likelihood category. The threshold function is piecewise-constant on per-lobe involvement fractions and locally Lipschitz-continuous on the underlying density reconstruction. Difficulty tier delta = 3 inherited from L1-029. Mismatch parameters: kV_calibration_drift, slice_thickness_anisotropy, motion_blur_breath_hold, contrast_enhancement_state, lobe_segmentation_error, threshold_calibration_uncertainty.

L-DAG

L.xray_source -> L.attenuation_projection -> L.filtered_back_projection -> L.lobe_segmentation -> L.density_thresholding -> L.tss_classifier -> int.spatial

L.xray_sourceL.attenuation_projectionL.filtered_back_projectionL.lobe_segmentationL.density_thresholdingL.tss_classifierint.spatial

Well-posedness W

Existence:: true
Uniqueness:: conditional
Stability:: conditional
κ:: 50

Existence and uniqueness inherited from L1-029 CT. Stability inherits L1-029's well-conditioned reconstruction plus a small additive contribution from threshold_calibration_uncertainty. Joint Hadamard well-posedness for the coupled CT-reconstruction + TSS/CO-RADS forward established by Pan F et al. (2020, foundational TSS), Francone M et al. (2020 TSS validation), Prokop M et al. (2020 CO-RADS Dutch consensus), Inui S et al. (2020 multicenter TSS), Liu Z et al. (2021 deep-learning TSS automation), and Lessmann N et al. (2021 multicentre CO-RADS validation).

Solvability C

Solver class:: linear-operator + convex optimisation [L1-029 FBP + lobe segmentation] + categorical-readout [TSS/CO-RADS rule] | end-to-end deep neural [DeepSARS, COVNet, COVID-Net] with explicit physics-informed lobe-segmentation regularization
Convergence rate q:: 2
Complexity:: O(H * W * Z * log(N)) for stage 1 (inherited from L1-029); O(N_lobes * N_HU_classes) for stage 2 thresholding; total stage-1-dominated

Specs (0)

No L2 specs registered yet for this principle.