Artificial Intelligence Computer Vision for Pulmonary Tuberculosis Screening Under Variable Chest Radiograph Grid Misalignment: A Geometric Self-Calibration Evaluation of Healthcare Diagnostic Concordance

Chest radiograph grid misalignment is a routine acquisition fault in tuberculosis screening programmes, yet its effect on the concordance of artificial intelligence computer vision classifiers is rarely quantified in a way that separates geometric distortion from the classifier's own decision behaviour. No observational data are used or claimed: the analysis characterises the design under a declared variance structure with between-cluster SD = 6.0, residual SD = 4.5 and intraclass correlation = 0.640, across 2000 replications per design cell for cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. The design requires 52 clusters to detect a 5 percentage-point concordance difference at 80% power. The framework proposes that grid-line geometry can be recovered without a phantom, that concordance loss is estimable as a separable interaction, and that the before-after-control-impact estimator is the appropriate inferential vehicle for programme-level evaluation. The contribution is a formalised calibration model and a power-analytic characterisation of the design, not a report of field performance.