1993 Fiscal Year Final Research Report Summary
Development of an Equipment for Automatic Evaluation of Cardiovascular Dynamics : Using Echocardiography
| Project/Area Number |
04557115
|
|---|
| Research Category |
Grant-in-Aid for Developmental Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
医学一般
|
|---|
| Research Institution | Osaka University |
|---|
Principal Investigator |
TAKEDA Hiroshi OSAKA UNIVERSITY,OSAKA UNIVERSITY HOSPITAL,ASSOICIATE PROFESSOR, 医学部・付属病院, 助教授 (20127252)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
MASUYAMA Tohru OSAKA UNIVERSITY,OSAKA UNIVERSITY HOSPITAL,FELLOW, 医学部・付属病院, 医員
TANOUCHI Jun OSAKA UNIVERSITY,MEDICAL SCHOOL,ASSISTANT, 医学部, 助手 (20197544)
INOUE Michitoshi OSAKA UNIVERSITY,OSAKA UNIVERSITY HOSPITAL,PROFESSOR, 医学部・付属病院, 教授 (30028401)
|
|---|
| Project Period (FY) |
1992 – 1993
|
| Keywords | ECHOCARDIOGRAPHY / AUTOMATIC MEASUREMENT / IMAGE PROCESSING |
|---|
| Research Abstract |
Automatic measurement has been expected in clinical fields for more objective clinical evaluation. We attempt to apply an image processing technique to automatic measurement of left ventricular wall motion on echocardiotomography. This approach may allow us objective evaluation of regional left ventricular function without any cardiovascular experts. In this study, we examined ideal accuracy of automated measuring method on echocardiotomograms and evaluated accuracy in clinical use.
The basic approach of automated evaluation in left ventricular function was measurement of shortening in systole between two points on left ventricular wall. We set multiple ROIs on an initial image of left ventricular wall in echocardiography. A texture tracking procedure based on cross correlation of image texture automatically traced ROIs on following tomograms. Thus, we could calculate fractionals hortening from changes of distance between ROIs. In an ideal condition, the mean accuracy of ROI tracing was 0.4mm that was same as special resolution of tomogram. Although, 23% error was observed for clinical images. We introduced pre-processing procedure to improve error fraction. This procedure reduced the error to 5%.
These results suggested that automated measurement of left ventricular function would be applied in clinical field.
|
