Japanese

Effect of deep learning reconstruction on the image quality of ultra-high-resolution computed tomography for diffuse lung diseases Kohei Mitsuhashi 1 1Department of Radiology Kanagawa Cardiovascular and Respiratory Center Keyword: びまん性肺疾患 , ultra-high resolution CT , deep learning reconstruction pp.17-24
Published Date 2020/1/10
DOI https://doi.org/10.18888/rp.0000001105
  • Abstract
  • Look Inside
  • Reference

We compared ultra-high-resolution CT images reconstructed with Hybrid Iterative Reconstruction(HIR)(AIDR3D-FC52)and Deep Learning Reconstruction(DLR)(AiCE-body)quantitatively and qualitatively. Noise for DLR images was significantly smaller than those for HIR. The extent of each lesion was almost the same between two kinds of CT images. The peripheral bronchiectasis and intralobular abnormalities were easily recognized in DLR. The score of image quality in DLR was higher than those in HIR. DLR can reduce image noise keeping with sharpness of the peripheral lung structures. Ultra-high-resolution CT using DLR is useful to evaluate diffuse lung diseases.


Copyright © 2020, KANEHARA SHUPPAN Co.LTD. All rights reserved.

基本情報

電子版ISSN 印刷版ISSN 0009-9252 金原出版

関連文献

もっと見る

文献を共有