Computed Tomography is a technique that allows the non-destructive evaluation of internal structures of objects. It can be analyzed as a two-step process: exploration of the samples and reconstruction of the images. The software that performs image reconstruction must reproduce, in the reconstructed volume, the geometric characteristics of the internal objects and maintain the density relationships between their materials. A methodology for evaluating the quality of reconstruction software is presented. To show the use of the methodology, a simulate phantom is build and the tomography projections are simulated with GATE and XRMC codes. The projections were processed by reconstruction software and results volumes were verified. The application of the methodology proved for the two reconstructed volumes the correct positions of objects and the relations between their densities. In the reconstructed volumes, it was possible to separate objects with different densities with a probability error less than 5%.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Methodology for Evaluating the Quality of Computed Tomography Reconstruction Software

  • A. M. Martins,
  • J. S. Domínguez,
  • J. P. F. Guimarães,
  • A. A. L. Bezerra,
  • C. B. de Jesus,
  • E. T. V. Orellana,
  • F. M. Millian

摘要

Computed Tomography is a technique that allows the non-destructive evaluation of internal structures of objects. It can be analyzed as a two-step process: exploration of the samples and reconstruction of the images. The software that performs image reconstruction must reproduce, in the reconstructed volume, the geometric characteristics of the internal objects and maintain the density relationships between their materials. A methodology for evaluating the quality of reconstruction software is presented. To show the use of the methodology, a simulate phantom is build and the tomography projections are simulated with GATE and XRMC codes. The projections were processed by reconstruction software and results volumes were verified. The application of the methodology proved for the two reconstructed volumes the correct positions of objects and the relations between their densities. In the reconstructed volumes, it was possible to separate objects with different densities with a probability error less than 5%.