Recent advances in CT technologies have enabled clinicians to obtain threedimensional (3D) volumetric images with high resolution. In this research, fullyautomated methods to analyze anatomical structures from chest CT images were developed and evaluated. The main focus of this research was on analyzing low-dose CT images to obtain diagnostic information. All automated analysis methods presented have been designed for and evaluated on CT images taken with low radiation exposure to the patients. A method was developed for analyzing intrathoracic airways, and the precision of the automated measurement was quantified. A novel contribution of this work was the development of the method for comparative measurement of airways using repeat scans of the same patient, which is clinically relevant for monitoring a patient's condition over time but has not yet been explored by others. A technique for precisely measuring airway's wall thickness was developed, which showed a significant improvement in measurement precision over the conventional full-width half-maximum (FWHM) based measurements. The segmentation is often a first step in the automated analysis as the segmented organs or structures may be used to retrieve useful diagnostic measures. The algorithms to segment various anatomical structures were developed and validated using large datasets. Top-down approach was used by first performing segmentations of the structures that were robustly identifiable and using those as a basis for segmenting other structures. The segmentations were performed for airway tree, spinal canal, ribs, and vertebrae, and the experimental results showed that these structures can be segmented robustly from low-dose CT images. Another aspect of this dissertation was on establishment of a chest frame of reference (CFOR) that serves as a common reference grid for the chest region. Such a reference frame is useful for normalizing chest regions for different-sized individuals, for studying spatial distribution of a certain anatomical point of interest, or for matching anatomical point across different intra-subject CT scans. Experimental results showed that the anatomical points are well-localized when the proposed CFOR was used.