核磁気共鳴画像を利用したPET画像再構成法

• 村瀬 研也

  愛媛大学医学部放射線医学教室

• ZHANG Youpu

  McConnell Brain Imaging Center, Montreal Neurological Institute, McGill University

• MA Yilong

  McConnell Brain Imaging Center, Montreal Neurological Institute, McGill University

• 山本 皓二

  宮崎大学工学部情報工学科情報処理システム講座

• EVANS Alan C.

  McConnell Brain Imaging Center, Montreal Neurological Institute, McGill University

書誌事項

タイトル別名

• Positron Emission Tomography Image Reconstruction Method Utilizing Magnetic Resonance Image.
• Positron Emission Tomography Image Reco

抄録

Magnetic resonance imaging (MRI) provides high-resolution anatomical information. In this study, we investigated the usefulness of an MRI-constrained method for image reconstruction of positron emission tomography (PET) . We started the reconstruction process with a conventional Bayesian reconstruction (BAY) method. Typically after 7 iterations, we calculated the apparent radioactivity ratio between gray matter and other tissues such as white matter and cerebrospinal fluid from both the PET image and tissue classification probability images generated from segmented MR images, and then increased this ratio to enhance the functional boundary gradient. We multiplied the apparent ratio by 2.0. Subsequently, the cross-correlation between anatomical and enhanced functional boundary maps was incorporated into BAY using the weighted line site method. This algorithm was tested with simulated PET projection data generated from an MRI based three-dimensional brain phantom, including the effects of resolution, attenuation, scatter and Poisson noise. The reconstructed images were evaluated in terms of root mean square distance (RMSD) from the true image. This method improved RMSD by 39.6%, 33.5%, and 33.0% on average, compared to (i) the filtered backprojection method, (ii) the maximum-likelihood expectation-maximization method and (iii) BA Y method, respectively. Our preliminary results suggest that incorporation of both anatomical and tissue classification information into the image reconstruction process is useful for improving the qualitative and quantitative accuracy of PET images.

収録刊行物

• ＲＡＤＩＯＩＳＯＴＯＰＥＳ

  ＲＡＤＩＯＩＳＯＴＯＰＥＳ 46 (4), 219-225, 1997

  公益社団法人 日本アイソトープ協会

キーワード

• MRI
• PET
• Bayesian image reconstruction
• anatomical information
• tissue classification

詳細情報

• CRID: 1390001204154050176
• NII論文ID: 130004383000
• NII論文ID: 10003721247
• NII書誌ID: AN00351589
• DOI: 10.3769/radioisotopes.46.219
• ISSN: 18844111
• ISSN: 00338303
• NDL書誌ID: 4183241
• Web Site: https://ndlsearch.ndl.go.jp/books/R000000004-I4183241
• Web Site: http://www.jstage.jst.go.jp/article/radioisotopes1952/46/4/46_4_219/_pdf
• Web Site: https://search.jamas.or.jp/link/ui/1997202704
• 本文言語コード: en
• データソース種別:
  • JaLC
  • NDL
  • Crossref
  • CiNii Articles
• 抄録ライセンスフラグ: 使用不可