CT perfusion imaging of lung cancer: benefit of motion correction for blood flow estimates

• Published in: European radiology 2018-12, Vol.28 (12), p.5069-5075
• Main Authors: Chu, Lisa L., Knebel, Robert J., Shay, Aryan D., Santos, Jonathan, Badawi, Ramsey D., Gandara, David R., Knollmann, Friedrich D.
• Format: Article
• Language: English
• Subjects: Adult; Aged; Algorithms; Blood flow; Cancer; Carcinoma, Non-Small-Cell Lung - blood supply; Carcinoma, Non-Small-Cell Lung - diagnostic imaging; Carcinoma, Non-Small-Cell Lung - physiopathology; Computed Tomography; Diagnostic Radiology; Estimates; Female; Humans; Imaging; Informed consent; Internal Medicine; Interventional Radiology; Lung cancer; Lung Neoplasms - blood supply; Lung Neoplasms - diagnostic imaging; Lung Neoplasms - physiopathology; Male; Measurement methods; Medical diagnosis; Medical imaging; Medicine; Medicine & Public Health; Middle Aged; Motion; Neovascularization, Pathologic - diagnostic imaging; Neovascularization, Pathologic - physiopathology; Neuroradiology; Observer Variation; Perfusion; Perfusion Imaging - methods; Prospective Studies; Radiographic Image Interpretation, Computer-Assisted - methods; Radiology; Reproducibility; Reproducibility of Results; Respiration; Software; Statistical analysis; Tomography, X-Ray Computed - methods; Tumors; Ultrasound; Variability
• ISSN: 0938-7994; 1432-1084
• DOI: 10.1007/s00330-018-5492-1

Purpose CT perfusion (CTP) imaging assessment of treatment response in advanced lung cancer can be compromised by respiratory motion. Our purpose was to determine whether an original motion correction method could improve the reproducibility of such measurements. Materials and methods The institutional review board approved this prospective study. Twenty-one adult patients with non-resectable non-small-cell lung cancer provided written informed consent to undergo CTP imaging. A motion correction method that consisted of manually outlining the tumor margins and then applying a rigid manual landmark registration algorithm followed by the non-rigid diffeomorphic demons algorithm was applied. The non-motion-corrected and motion-corrected images were analyzed with dual blood supply perfusion analysis software. Two observers performed the analysis twice, and the intra- and inter-observer variability of each method was assessed with Bland-Altman statistics. Results The 95% limits of agreement of intra-observer reproducibility for observer 1 improved from −84.4%, 65.3% before motion correction to −33.8%, 30.3% after motion correction (r = 0.86 and 0.97, before and after motion correction, p < 0.0001 for both) and for observer 2 from −151%, 96% to −49 %, 36 % (r = 0.87 and 0.95, p < 0.0001 for both). The 95% limits of agreement of inter-observer reproducibility improved from −168%, 154% to −17%, 25%. Conclusion The use of a motion correction method significantly improves the reproducibility of CTP estimates of tumor blood flow in lung cancer. Key Points • Tumor blood flow estimates in advanced lung cancer show significant variability. • Motion correction improves the reproducibility of CT blood flow estimates in advanced lung cancer. • Reproducibility of blood flow measurements is critical to characterize lung tumor biology and the success of treatment in lung cancer.