Comprehensive approach to pulmonary image analysis

Results of the ongoing medical image analysis project leading towards development of a comprehensive suite of methods and tools for quantitative pulmonary image analysis from multi-detector CT image data are presented. This includes methodology for segmentation of lung, lung lobe, bronchial tree and vascular tree, together with methods for quantitative analysis of the segmented structures. Methods for matching of structures over time that allow longitudinal comparisons of the measured quantitative indices are discussed. In addition, texture analysis and pattern recognition techniques used for tissue characterization of pulmonary parenchyma, and validation results, are reported. The future directions of quantitative and highly automated approaches to pulmonary image analysis are addressed.