首页 | 本学科首页   官方微博 | 高级检索  
     


Use of an optical flow method for the analysis of serial CT lung images
Authors:Dougherty Lawrence  Torigian Drew A  Affusso John D  Asmuth Jane C  Gefter Warren B
Affiliation:Hospital of the University of Pennsylvania, Department of Radiology/1 Silverstein, 3400 Spruce Street, Philadelphia, PA 19104, USA. dougherty@rad.upenn.edu
Abstract:RATIONALE AND OBJECTIVES: Serial CT lung studies are difficult to compare due to misregistration between image sets. An optical flow method (OFM) was adapted for use on CT lung images to register images and visualize changes between studies. Three applications were investigated: lung nodule assessment; evaluation of pulmonary enhancement; and functional changes due to air trapping. MATERIALS AND METHODS: From an initial clinical study, a follow-up study was created by digitally manipulating the images to simulate patient positioning errors and nodule growth. Nodule growth was measured from the temporal subtraction of registered images. In application to the assessment of pulmonary enhancement, pre and postcontrast images from a patient with acute pulmonary embolism (PE) were registered. A map of the perfused blood volume was computed from the ratio of aligned lung volumes. Functional changes in the lung were demonstrated using images from a patient with air trapping. End-inspiratory and end-expiratory volumes were aligned and displacement fields estimated using the OFM. Principal strains were computed from the displacement fields. RESULTS: All image volumes were aligned with at least 0.95 correlation. OFM estimates of displacement showed excellent agreement with the prescribed displacements with 0.33 pixel RMS error. Nodule growth was evident in the presence of significant positioning errors. In the PE case, enhancement ratios indicated a hypoperfused area consistent with an occlusive hypodense filling defect. For the air trapping case, a strain map showed functional changes along the interface of the air trap. CONCLUSIONS: The OFM can facilitate the detection and quantification of changes between serial CT lung studies.
Keywords:Computed tomography (CT)   lung   nodule   perfusion   registration
本文献已被 ScienceDirect PubMed 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号