| Project/Area Number |
11670872
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Radiation science
|
|---|
| Research Institution | The university of Tokyo |
|---|
Principal Investigator |
ONEGE Yuze University of Tokyo, Facuulty of Medicin. Lecturer, 医学部・附属病院, 講師 (90233593)
|
|---|
| Project Period (FY) |
1999 – 2000
|
| Project Status |
Completed (Fiscal Year 2000)
|
| Budget Amount *help |
¥2,300,000 (Direct Cost: ¥2,300,000)
Fiscal Year 2000: ¥800,000 (Direct Cost: ¥800,000)
Fiscal Year 1999: ¥1,500,000 (Direct Cost: ¥1,500,000)
|
|---|
| Keywords | Speeds recognition / radiology report / dictation / recognition rate / 放射線診断報告書 / 放射線検査 / 読影報告書 |
|---|
| Research Abstract |
We applied a continuous speech recognition technique to our radiological reporting system and compared its registration time with the keyboard input system. The recognition rate for CAT scan report was about 65% when we used a continuous speech recognition system for daily conversation and politics-economics domain, which did not contain radiological terms. This rate was much improved with our new recognition system for radiology domain and the result was about 90% recognition rate. But we found that it took much time to correct the miss-recognized terms with keyboard and even if the recognition rate was 90%, total report input time was almost same between keyboard and continuous speech recognition. The problems to be solved in future are as follows, attaining better recognition rate, and finding some good method to correct miss-recognized terms. And also we find that it requires some special training for users to get mood performance on recognition. To say nothing of our being not used to dictate, it requires another way of thinking for uttering findings in clear and smooth voice. This technique leaves much room for improvement, however we think it is getting a position to a fine practical technique.
|
