Distortion-Guided Quadtree Adaptivity Within the Systematic Modeling by Predefined Envelope and Signature Sequence Framework for Deterministic Compression of Four-Dimensional Lung Computed Tomography Images

Abstract

Four-dimensional (4D) lung computed tomography (CT) provides valuable spatiotemporal information for respiratory motion analysis but introduces substantial challenges related to data volume, motion-induced heterogeneity, and spatially varying entropy. Although strong temporal correlations exist between respiratory phases, uniform exploitation of temporal redundancy may obscure localized representation effects that are important for preserving anatomical fidelity in motion-affected regions. This study presents a deterministic compression framework within the Systematic Modeling by Predefined Envelope and Signature Sequences paradigm, extended with distortion-guided quadtree adaptivity for respiratory-resolved 4D lung CT data. The proposed method builds upon the Classified Energy and Pattern Blocks representation and integrates overlap-aware block processing with hierarchical spatial partitioning controlled directly by reconstruction distortion rather than variance- or entropy-based signal statistics. Consequently, spatial refinement is applied selectively only in regions where representation error exceeds predefined thresholds. Experimental evaluation on publicly available 4D lung CT datasets shows that the proposed framework expands the achievable rate–distortion operating range compared with fixed-block and overlap-only configurations. Quantitative results indicate improved compression efficiency while maintaining competitive reconstruction quality. Spatial analyses further demonstrate localized distortion reduction in motion-intensive anatomical regions such as the diaphragm and chest wall boundaries. These findings suggest that distortion-guided adaptive representation provides a transparent and deterministic strategy for compression of motion-resolved medical imaging data and establishes a methodological foundation for future comparisons with conventional compression standards and task-based clinical validation.

Cite this article as: I. Z. Gökbay and B. S. B. Yarman, “Distortion-guided quadtree adaptivity within the Systematic Modeling by Predefined Envelope and Signature Sequences framework for deterministic compression of four-dimensional lung computed tomography images,” Electrica, 26, 0433, 2026. doi: 10.5152/electrica.2026.25433.