Geometry-Constrained Multi-Modal Medical Imaging Using a Projected Bansal-Kalman Algorithm on a Stratified Manifold

This paper presents a geometry-constrained framework for multi-modal medical imaging based on a projected Bansal-Kalman algorithm operating on a stratified manifold. Imaging states are modeled on the Bansal manifold, which admits compatible Riemannian, Clifford, and Lie-group structures to enforce geometry-admissible evolution and estimation. The proposed framework provides a unified formulation across CT, MRI, PET, and whole-slide imaging without modality-specific redesign. Analytical stability and convergence properties are established, and the approach supports interpretable and regulator-aligned imaging pipelines.