目的 高分辨率多层螺旋CT是临床医生研究肺部解剖结构功能、评估生理状态、检测和诊断病变的主要影像学工具。鉴于肺部各解剖结构间特殊的关联关系和图像成像缺陷、组织病变等干扰因素对分割效果的影响，学术界已在经典图像处理方法基础上针对CT图像中的肺部解剖结构分割进行了大量研究。方法 本文通过对相关领域有代表性或前沿性文献的归纳总结，系统性地梳理了现有肺组织、肺气管、肺血管、肺裂纹、肺叶或肺段等解剖结构CT图像分割方法的主要流程、方法理论、关键技术和优缺点，讨论了各解剖结构分割的参考数据获取、实验设计方法和结果评价指标。结果 分析了现有研究在结果精度和鲁棒性方面所面临的挑战性问题，以及基于分割结果在定位病变、定量测量、提取其它结构等方面展开的热点应用，特别详述了当前被重点关注的深度学习方法在本领域的工作进展，同时展望了本领域在分割理论方法和后续处理等步骤的发展趋势，并探索了如何在实践中根据分割结果发现新的临床生物标志。结论 快速精确地从CT图像中分割肺部各解剖结构可以获取清晰直观的3维可视化结构影像，展开解剖结构内部的定量参数测量或结构之间的关联关系分析能提供客观、有效的肺部组织疾病辅助诊断依据信息，可以大大减轻临床医生的阅片负担、提高工作效率，具有重要的理论研究意义和临床应用价值。
Objective Pulmonary disorder is with high morbidity and mortality world-wide according to the reports by World Health Organization. Some common pulmonary disorders include lung nodule and cancer, interstitial lung disease, chronic obstructive pulmonary disease, bronchiectasis and pulmonary embolism, etc. The disorders are typically characterized by long-term poor breath quality, irregular blood supply, and obstructive airflow and lesser circulation. Since pulmonary disorders bring not only enormous societal financial burden but also physical and mental suffering to patients, the recognition and comprehension of the disorders are widely considered to be the most basic and crucial medical tasks. High-resolution multi-slice computed tomography (CT) receives dominant prevalence among pulmologists and radiologists due to its allowance of investigating pulmonary anatomic function, assessing physiological conditions, and detecting and diagnosing pulmonary disorders. Hundreds of isotropic thin slices reconstructed real-timely from single spiral CT scanning could realize objective, repeatable and non-invasive clinical inspections. This over-performs against traditional tools, especially in early diseased stage. However, manual delineation, measurement and evaluation of volumetric scans turn out to be extremely time-consuming and put intensively laborious work load for clinicians. The biomedical engineering community then aims at developing semi- and fully-automated segmentations in CT images by means of voxel-by-voxel labeling using computer software to separate sub-divided pulmonary anatomic structures from each other. In the presence of the unique inter- and intra-anatomy relationships, and the impact of imaging defects, abnormalities or other interference factors, classical image processing methods suffer from performance limitations. The anatomic CT visibility attenuates and morphology deforms both spatially and pathologically, which also play a negative role on the segmentation results. A number of researches, usually incorporating traditional work and carefully defined processing rules, have been employed focusing on thoracic CT images. Method In this paper a systematic review of anatomic segmentation methods of pulmonary tissues, airways, vasculatures, fissures and lobes is presented by tracking and summarizing the representative or up-to-date published literatures. In addition, a sequence of attractive practices and extractions of sub-divided or related structures, derived from segmented anatomic results, are also attached to corresponding anatomic subsection. They include the segmentation of adhesive, pleura nodular and interstitial diseased lungs, centerline extraction of airways and vasculatures, airway wall quantification and segmentation, pulmonary artery and vein separation, and pulmonary segment approximation. Moreover, for all the referred segmentation methods, the full implementation pipeline, background image processing methodology, and key techniques to ensure result performance are elaborated. Analogous methods are further classified on the basis of their designed frameworks or mathematical theories, and the merits and demerits of each method type are analyzed at the end of the classification content. In general, it is troublesome for researchers to evaluate their segmentations and compare with fellow works. This is mainly because the difficulty in the acquisition of ground truth. LOLA11, EXACT09 and VESSEL12, three of the public and authoritative MICCAI grand challenges in chest image analysis, are then introduced for result comparisons in the directions of pulmonary tissue and lobe, airway, and vasculature. The complete procedures for the challenge owners to construct their reference repository and quantify submission performance are emphasized. The reported standard generation approaches for other anatomic-based applications are illuminated in parallel. Evaluated indices contain the anatomic boundary alignment and volume overlap, and the trade-off between true positive and false positive detection. Next, experimental validation approaches are explained based on the reference standard and indices. The qualitative and quantitative results of different methods are shown specifically with the description of the test datasets. Result Per individual anatomic topic, the existing challenges of state-of-the-art studies are put forward in detail, targeting on the accuracy performance in both true positive detection and false positive removal, and the robustness performance against the diversity of CT scanners, imaging protocols and the appearance of various abnormalities. A set of practical problems, such as lesion locating, qualitative and quantitative anatomic measuring, and sequential component segmenting, is also discussed in the paper. Besides, deep learning algorithms, in particular convolutional neural network-based algorithms, have rapidly become a preference of choice for medical imaging institutions. At present, two mature applied utilizations in chest CT imaging field are nodule detection and malignancy prediction in lung cancer screening, and interstitial lung disease type classification. The major deep learning-based efforts are surveyed in regard to the contribution to pulmonary tissue bounding box localization as well as pulmonary airway segmentation and leakage removal, most of which were proposed in the recent two years. The improvements of the efforts are compared with previous methods. Concerning the frontier requirements from scientific groups, industrial units and pulmological domains, future work trends and open issues are listed pertinent to methodology and post-processing steps together with the applications such as the identification of pulmonary lesion sites and the following anatomic segmentation. Since it is commonly recognized that the parameters relied on anatomic measuring is of vital importance to characterize the progress and severity of pulmonary disorders, a few possible innovative points for achieving these biomarkers are also recommended. Conclusion It is beneficial for not only theoretical researches but also clinical practices to implement accurate, fast and robust pulmonary anatomic segmentation from large amounts of CT images. The transfer or modification of succeeded pulmonary segmentation methodology would facilitate the segmentation of other tissues, organs and multi-modality images. Inter- and intra-structure measurements and the relationship mapping information obtained ranging from global to local analyses can provide objective and effective evidence in computer aided pulmonary disease detection and diagnosis. The 2D transversal or 3D visualization of those points is able to present intuitive, legible and proportional views of the anatomic structures with the help of volume rendering techniques and grayscale Dicom slices overlaid by chromatic tissue marks. The contribution of these aspects reduces pulmologists’ and radiologists’ labor involvement and increase their efficiency significantly. Although deep learning algorithms are currently just started to participate in the topic, and not perfectible in segmentation time cost and refinement steps, they still have considerable potency and space for the studies in pulmonary segmentation areas, which is believed to dominate in this field in the coming years.