1994 Fiscal Year Final Research Report Summary
Studies on Computerized lmage Recognition of Digital Chest Radiographs
Project/Area Number |
05680303
|
Research Category |
Grant-in-Aid for General Scientific Research (C)
|
Allocation Type | Single-year Grants |
Research Field |
Intelligent informatics
|
Research Institution | Sapporo Medical University |
Principal Investigator |
KOBA Hiroyuki Sapporo Medical University School of Medicine, Assistant Professor, 医学部, 助教授 (40186721)
|
Co-Investigator(Kenkyū-buntansha) |
TORIWAKI Junichiro Nagoya University Faculty of Engineering, Professor, 工学部, 教授 (30023138)
YAMAGISHI Masahiko Sapporo Medical University Scholl of Medicine Instructor, 医学部, 助手 (10211624)
|
Project Period (FY) |
1993 – 1994
|
Keywords | Chest radiograph / lmage processing / Pulmonary nodule / Pulmonary emphysema / Pulmonary vasculature |
Research Abstract |
(1) Differentiation of benign from malignant pulmonary nodules-radiographic features and dihital parameters- We evaluated radiographic features and digital parameters of 80 solitary pulmonary nodules (benign 40, malignant 40). Benign nodules tend to show regular margin, high density, clear contour, less spicular formation, and less vascular convergence. Digital parameters were obtained by tow kinds of concentric circular windows which radius was adjusted to the area of each nodule. DCF-N (directional contrast filter for no dule) output of central nodular region, dendity gradient of marginal nodular region, and density and density entropy value of surrounding region were measured. Discrimination ratio of benign from malignant nodules with these parameters was 78%, while that with radiographic features was 74%. These digital parameters seemed to be useful to differentiate the nodules. (2) Quantitative evaluation of the severity of pulmonary emphysematous change using conventional chest radiographs Comouterized method to quantitatively evaluate the degree of advance of pulmonary emphysema on chest radiographs was investigated by the following steps : 1.setting of ROIs (regions of interests) in intercostal spaces on digital chest radiographs 2.filtering by the Max-DD filter to extract vascular structures 3.estimation of thickness of vessels 4.quantitative evaluation of severity of emohysematous change The measured thickness of pulmonary vasculature in pulmonary emphysema was significantly smaller than that in normal control. Clinical usefulness of this computerized method must be studied in the future.
|
Research Products
(19 results)