| Project/Area Number |
05670795
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
Radiation science
|
|---|
| Research Institution | Tokyo Metropolitan Institute of Gerontology. |
|---|
Principal Investigator |
TOYAMA Hinako TMIG PET Medical center, Senior Research Scientist, ポジトロン医学研究部門, 主任研究員 (50180188)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
ODA Keiichi TMIG PET Medical center, assistant researcher, ポジトロン医学研究部門, 研究員 (70224235)
SENDA Michio TMIG PET Medical center, head positron medical center, ポジトロン医学研究部門, 副参事研究員 (00216558)
|
|---|
| Project Period (FY) |
1993 – 1994
|
| Project Status |
Completed (Fiscal Year 1994)
|
| Budget Amount *help |
¥2,000,000 (Direct Cost: ¥2,000,000)
Fiscal Year 1994: ¥800,000 (Direct Cost: ¥800,000)
Fiscal Year 1993: ¥1,200,000 (Direct Cost: ¥1,200,000)
|
|---|
| Keywords | Data Base / user-friendly-interface / Modeling / nonlinear least-square-fitting / Marquardt method / Fuzzy theory / Fractal dimension / 3D-surface rendering / parametric image / ファジー理論 / 動態解析 / インターフェイス |
|---|
| Research Abstract |
The purposes of this study are to reconstruct a new database system and to develope the new technique of the data analysis for medical images by using a fuzzy theory and a arterial intelligence technic. There are five characters of this database system. A user friendly interface was developed to retrieve and update the database quickly and easily by using worksheets explicitly or implicitly. This database works on our network system and we can access the database from any work stations. A directory address, in which a data file (image or data) is stored, is managed by this database. The new records such as PATIENT,STUDY,SCAN are created automatically when a new datafile is created. The informations for a patient, a study and a scan are recorded in the header of a datafile. We connected this database to the data processing system and reporting system. For example, a history of data processing was saved in the header of the data file and registerd into the data comment field of the data
… More
record in the database when the datafile was registerd.
We developed a method to generate a fractal dimension image for the T1 weighted MRI image and PET functional images. The fractal dimension at the target point is established as Hausdorff dimension (D) which is defined as N (r) =l/rD,where N (r) is the number of unit voxel with size r covering the three dimensional region enclosed by the area of 7*7 pixel around the target pixel. This method is called "boxcounting method". We modified the method of counting the number of unit voxels in "boxcounting method" to generate a new filter for edge enhancement. In stead of N*N pixel area (N=7), we counted the number of unit voxels in the r*((N-i)/r+r) region, where i and r varied from 1 to N-1. Edge enhanced image was obtained by our new method in a normal case and compared to the conventional fractal dimension image. Using this image we could remove the structure surrounding the brain such as dura and scalp by masking the pixels with the value above the threshold (about 1900) in the filtered image. The contour of the brain was obtained automatically from this image for all slices except a few top slices. Calculation time for filtering and edge detection was about 15 seconds per a slice.
In the model analysis of the dynamic PET images, a fuzzy theory was applied to save the conputation time in the non-linear least square curve fitting. A new searching point in the Flecher algorithm was determind by using a fuzzy set theory. This method was useful to save the calculation time in the noisy data Less
|
