Quantitative characterization of cigarette meso-structure and its relationship with carbon monoxide emission based on x-ray micro-CT

Carbon monoxide (CO) is one of the major harmful constituents in cigarette smoke; however, the influence of cigarette meso-structure on CO emission remains insufficiently understood. Conventional macroscopic physical indicators fail to adequately characterize the role of internal porous structures in smoke generation and transport. In this study, high-resolution x-ray micro-computed tomography (μCT) was employed to perform helical scanning of 110 cigarette samples representing seven brands and specifications provided by China Tobacco Hunan Industrial Co., Ltd, enabling three-dimensional characterization of their internal meso-structures. A total of ten structural parameters including porosity statistics, specific surface area, permeability, fractal dimension, pore–throat ratio, mean coordination number, and three-dimensional shape factor were extracted. Random forest regression and a least absolute shrinkage and selection operator–backpropagation neural network regression model were then constructed to analyze the relationships between these meso-structural features and unit CO emission. The results demonstrate that cigarette meso-structure, particularly porosity distribution characteristics, has a significant influence on CO emission and exhibits a dominant role in both modeling frameworks. The combination of μCT-based structural parameters with machine learning methods provides a novel approach for exploring the statistical associations between meso-structural features and the emission of harmful smoke constituents and offers a scientific basis for harm-reduction strategies through structural optimization.