1987 Fiscal Year Final Research Report Summary
Quantitative categorization of pneumoconiosis based on digital radiography and its application to automatic diagnosis
| Project/Area Number |
60302087
|
|---|
| Research Category |
Grant-in-Aid for Co-operative Research (A)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
Radiation science
|
|---|
| Research Institution | Tokyo University of Agriculture and Technology |
|---|
Principal Investigator |
KOBATAKE HIDEFUMI Professor of the Fsculty of Technology,Tokyo Univ. of Agir. & Tech., 工学部, 教授 (80013720)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
NOBECHI Tokuro The fromer director of St. Lukas International Hospital, Tokyo, 放射線科, 顧問
HOSODA Yutaka Chief of Clinics, Radiation Effects Research Foundation, Hiroshima, 臨床部, 部長
SHIDA Hisao Chief of Radiology, Rosai Hospital for Silicosis, Tochigi, 放射線科, 部長
TATENO Yukio Director of the Division of Clinical Research, National Institute of Rediologica, 臨床研究部, 部長
MIYAMICHI Juichi Associate Progessor of the Faculty of Technology, Utunomiya University, 工学部, 助教授 (50107166)
|
|---|
| Project Period (FY) |
1985 – 1987
|
| Keywords | Digital Radiography / Pneumoconiosis / Automatic Diagnosis / Computed Radiography / Digital Radiography / 画質評価 / Fuji Computed Radiography / CR |
|---|
| Research Abstract |
The objectives of this research are to confirm the usefulness of the application of computed radiography technology to the reliable deagnosis of pneumoconiosis and to develop a computerized system for its automatic diagnosis. The main achievements are as follows.
1. X-ray image data base of pneumoconiosis has been made. It consists of 17 ILO 1980 Standard films, other 26 semi-standard films and 44 FCR (Fuji Computed Radiography) films. Not only profusion categorization has been made for these FCR films, but also partial lung area has been identified whose opacity density reflects the film categorizstion typically.
2. Effects of digital image processing on radiographic interpretation of pneumoconiosis have been investigated. Digitized images of pneumoconiosis were processed with the seversl contrast and spatial frequency characteristics, and the experiments on interpretation for the processed rsdiograph of pneumoconiosis wereperformed. It has been concluded that diagnosis of pneumoconiosis can be made by the processed film of reduced sizu(1/2 in length) with the same accuracy as the conventional films.
3. Six kinds of automatic diagnostic system have been developed. Three of them belong to the methods based on texture analysis and the others are based on the identification of small poacities of pneumoconiosis. Their correct classification rate for normal/abnormal categorization lie between 80 and 100 percent. Experiment for 4 major categorization have been performed and the correct classigication rates range between 60 and 80 percent. Each method has its own characteristics, and effective consolidation of these methods may have a potential capability of reliable automatic diagnosis.
Such total system is now under developing.
|
