1994 Fiscal Year Final Research Report Summary
Improvement in Resolution by Using Multiple Projections for Coded Aperture Emission CT
Project/Area Number |
05452405
|
Research Category |
Grant-in-Aid for General Scientific Research (B)
|
Allocation Type | Single-year Grants |
Research Field |
Biomedical engineering/Biological material science
|
Research Institution | University of Tokyo |
Principal Investigator |
FUJIMURA Sadao University of Tokyo, Faculty of Engin., Professor, 工学部, 教授 (30010961)
|
Co-Investigator(Kenkyū-buntansha) |
HANAIZUMI Hiroshi Hosei University, Fac.of Eng.Ass.Prof., 工学部, 助教授 (60143385)
KIYASU Senya Univ.of Tokyo, Fac.of Eng.Rsearch Associate, 工学部, 助手 (20234388)
ITO Tadashi Univ.of Tokyo, Fac.of Eng.Rsearch Associate, 工学部, 助手 (20223159)
ISHIKAWA Masatoshi Univ.of Tokyo, Fac.of Eng.Ass.Prof., 工学部, 助教授 (40212857)
|
Project Period (FY) |
1993 – 1994
|
Keywords | Emission CT / Coded Aperture / Resolution / Multiple Projections / Projection Method / Differential Imaging / Optimal Filtering / Quadratic programming |
Research Abstract |
Results are summarized as follows : 1.An optimal filter method we have developed in this project gives highest resolution in the direction of depth and the lowest root-mean-square error among the methods we have developed these years in the process of improvement : the depth resolution was two or three times better than, and the rms error was one fourth of, that given by a matched filter method. This was confirmed by computer simulation and experiments using radio isotope in a 3-D phantom. The evaluation was made by a point spread function for a point source, and also by reconstructed patterns for distributed objects, which were yielded by each method. 2.It is shown by computer simulation that an idealized observation system would yield 5 mm depth resolution. How to realize the system, however, is the subject for a future study. 3.Quadratic programming algorithm using a priori knowledge was also applied to reconstruction. An ideal result was obtained for a point source. As this algrithm is not linear, it is necesary to investigate the behavior of this algorithm when applied to reconstruction of patterns for distributed objects.
|